From CMOSCCD at aol.com Mon Jan 3 17:28:59 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Mon, 3 Jan 2005 15:28:59 EST Subject: CCD-world: PHOTON TRANSFER Message-ID: <7b.3bc0063e.2f0b050b@aol.com> Folks, Here we go again. . . happy new year !! The photon transfer technique has acquired a new dimension in the way it handles nonlinear detectors (e.g., CMOS).? For example. . check out the CMOS data below. ..? where sense node capacitance is plotted as a function of shot noise (rms volts) at the sense node and source follower output.? Note, we can now separate source follower gain nonlinearity (V/V) from sense node conversion gain nonlinearity (e-/V).? The ratio of gains is shown in the second plot.? The third plot, the standard photon transfer format, also reflects the gain differences when sense node and source follower outputs are plotted independently.? It's amazing that all information that photon transfer collects is reflected in the first plot. Let me know (off line) if you want to know more details in how absolute nonlinear data is collected (hardly more difficult than linear photon transfer collection schemes). Jim PS. . . note that three transistor (3T) CMOS pixels are very nonlinear because of sense node capacitance and source follower nonlinearities. On the otherhand, charge coupled CMOS pixels exhibit significantly better linearity because the sense capacitance for the most part is fixed (i.e., the gate capacitance of the source follower). Also, the voltage swing on the sense node and source follower is limited in comparison to 3T (with a corresponding full well reduction). Scientific CMOS is focusing on charge coupling for this important reason (but not the only reason). -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050103/7cb0352b/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 60397 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050103/7cb0352b/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 53200 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050103/7cb0352b/attachment-0001.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 74049 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050103/7cb0352b/attachment-0002.jpe From tg.burke at ngc.com Tue Jan 4 10:13:11 2005 From: tg.burke at ngc.com (Burke, Thomas G.) Date: Tue, 4 Jan 2005 08:13:11 -0500 Subject: CCD-world: IR remote controls Message-ID: Howdy all, I know this may be a little off-topic, but I'd like to ask - does anyone here know the output power of an IR Remote control? Their outputs are generally centered at one of a few discrete frequencies in the 800-950nm range, and I know it's a fairly gaussian output, but I'm curious about output power, both at the peak f, and integrated over f (eg total power)... Also, they generally encode their outputs on a 30-60kHz square-wave carrier.. I'm putting together a detector, & am thinking of using a remote as a quick "sanity check" type of thing to ensure it's working, but I don't want to spend a heap of dineros on it. A remote seems right up the alley, but I need to know output levels to ensure I don't saturate & so forth. Thanks, Tom From CMOSCCD at aol.com Thu Jan 6 05:14:19 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 6 Jan 2005 03:14:19 EST Subject: CCD-world: NONLINEAR PHOTON TRANSFER Message-ID: <8a.1da11f8b.2f0e4d5b@aol.com> There are several folks here working on APS devices who would be interested in your techniques for handling non-linear outputs.? If you could send the details to me I will forward them to those interested. *************jj Barry. . . thanks for your interest. Lets start with the discussion below.. . .. . . <><> First. . . it is important to note that sense node capacitance Capplies to nonlinear detectors. . . which is a lot different than C exhibited by linear CCDs. Therefore, we must keep track of volts AND electrons as the light is varied to the pixels. Recall that linear photon transfer only needs to measure output DN (or volts). Electrons are readily calculated through the conversion constants .. . . e-/DN or e-/V .. . . which are NOT true relationships for nonlinear detectors (and I mean significantly !). We keep track of electrons by first taking a (very) low light level reading and applying the photon transfer equations above to find electrons. For higher light levels, we (must) find electrons through exposure time from this reference point. . . or reading a standard diode next to the detector who's output is linear with light level. In either case signal electrons are proportional to exposure time or standard diode photo current. At this point we now know how DN (or volts at the source follower output) varies with electrons (or Q). We then convert source follower volts to sense node volts (for a good approximation we can just multiply by 1.2 and assume the source follower is linear). This produces the plot Q vs sense node voltage. Next the curve is differentiated (i.e., dQ/dV). I've been doing this by first curve fitting Q vs V (typically this results in a log curve. . . provided by Excel trendline). . and then differentiating (typically resulting in a k/V curve). The result gives the sense node capacitance curve as a function of sense node voltage.. . . . our desired result. >From there you can find everything else. . . e.g., true sense sensitivity (V/e-) and e-/DN. I've been working on the same problem above when the source follower has some V/V nonlinearity. .. . but that is another discussion. What do you think? :-) Let me know if you have questions. . or suggestions in making this photon transfer measurement technique more user friendly. We can go from there. Take care.. . Jim PS. . I thought I send this one to CCD World. . because I've received a half dozen inquiries on the subject. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/03a1ae22/attachment.html From pmooney at gatan.com Thu Jan 6 14:50:01 2005 From: pmooney at gatan.com (Paul Mooney) Date: Thu, 6 Jan 2005 09:50:01 -0800 Subject: FW: CCD-world: NONLINEAR PHOTON TRANSFER - why not? Message-ID: Jim, The basic photon-transfer method is differential for all intents and purposes since one obtains gain from measuring the slope of the variance vs. mean curve. That means that the bias signal doesn't create an error - as it would if one simply divided variance by mean at some particular dose level. For this reason, I don't understand why it wouldn't also work to simply generalize this concept for non-linear devices on the assumption that the rms variation of electron shot noise keeps the signal within a range for which the gain is approximately constant - and therefore doesn't mess up the basis assumptions needed for photon transfer. From this, then, the photon transfer method should in principal be able to measure the net response of non-linear capacitance and transistor gain by integration of the marginal gain. Am I missing something? Paul Mooney _____ From: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] On Behalf Of CMOSCCD at aol.com Sent: Thursday, January 06, 2005 12:14 AM To: bburke at ll.mit.edu Cc: ccd-world at ctio.noao.edu; ccd-world at ctios2.ctio.noao.edu Subject: CCD-world: NONLINEAR PHOTON TRANSFER There are several folks here working on APS devices who would be interested in your techniques for handling non-linear outputs. If you could send the details to me I will forward them to those interested. *************jj Barry. . . thanks for your interest. Lets start with the discussion below.. . .. . . <><> First. . . it is important to note that sense node capacitance C= dQ/dV applies to nonlinear detectors. . . which is a lot different than C = fixed as exhibited by linear CCDs. Therefore, we must keep track of volts AND electrons as the light is varied to the pixels. Recall that linear photon transfer only needs to measure output DN (or volts). Electrons are readily calculated through the conversion constants .. . . e-/DN = S(DN) / N(DN)^2 or e-/V = S(V) / N(V)^2 .. . . which are NOT true relationships for nonlinear detectors (and I mean significantly !). We keep track of electrons by first taking a (very) low light level reading and applying the photon transfer equations above to find electrons. For higher light levels, we (must) find electrons through exposure time from this reference point. . . or reading a standard diode next to the detector who's output is linear with light level. In either case signal electrons are proportional to exposure time or standard diode photo current. At this point we now know how DN (or volts at the source follower output) varies with electrons (or Q). We then convert source follower volts to sense node volts (for a good approximation we can just multiply by 1.2 and assume the source follower is linear). This produces the plot Q vs sense node voltage. Next the curve is differentiated (i.e., dQ/dV). I've been doing this by first curve fitting Q vs V (typically this results in a log curve. . . provided by Excel trendline). . and then differentiating (typically resulting in a k/V curve). The result gives the sense node capacitance curve as a function of sense node voltage.. . . . our desired result. >From there you can find everything else. . . e.g., true sense sensitivity (V/e-) and e-/DN. I've been working on the same problem above when the source follower has some V/V nonlinearity. .. . but that is another discussion. What do you think? :-) Let me know if you have questions. . or suggestions in making this photon transfer measurement technique more user friendly. We can go from there. Take care.. . Jim PS. . I thought I send this one to CCD World. . because I've received a half dozen inquiries on the subject. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/9e1fbab1/attachment.html -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: ATT00117.txt Url: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/9e1fbab1/attachment.txt From CMOSCCD at aol.com Thu Jan 6 21:00:05 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 6 Jan 2005 19:00:05 EST Subject: FW: CCD-world: NONLINEAR PHOTON TRANSFER - why not? Message-ID: <7a.6a371506.2f0f2b05@aol.com> Jim, ? The basic photon-transfer method is differential for all intents and purposes since one obtains gain from measuring the slope of the variance vs. mean curve.? That means that the bias signal doesn't create an error - as it would if one simply divided variance by mean at some particular dose level.? For this reason, I don't understand why it wouldn't also work to simply generalize this concept for non-linear devices on the assumption that the rms variation of electron shot noise keeps the signal within a range for which the gain is approximately constant - and therefore doesn't mess up the basis assumptions needed for photon transfer.? From this, then,?the photon transfer method should in principal be able to measure the net response of non-linear capacitance and transistor gain by integration of the marginal gain. ? Am I missing something? ? Paul Mooney *************jj Paul, I (really) hope you are on to something. . . but .. . :-) Let me show you the raw photon transfer data below taken from a nonlinear detector for your comment. Here we plot Noise (DN) vs Signal (DN).. . . standard photon transfer format. Note, if we use the linearily assumed photon transfer equation. . . e-/DN that e-/DN varies from 1.2 e-/DN to 19 e-/DN. We plot the result in the second plot in "red." The true result, using the nonlinear measuring technique described in a previous e-mail (below), is shown in "blue". As you can see. . there is a huge discrepancy in e-/DN and signal charge measured. Linear photon transfer fails. The question for you is. . . . how would you take the raw DN data below and get to the correct blue curve? . . . without having further information? Your turn. . . :-) Jim ****************jj There are several folks here working on APS devices who would be interested in your techniques for handling non-linear outputs.? If you could send the details to me I will forward them to those interested. *************jj Barry. . . thanks for your interest. Lets start with the discussion below.. . .. .? . <><> First. . .? it is important to note that sense node capacitance Capplies to nonlinear detectors. . . which is a lot different than C exhibited by linear CCDs.? Therefore, we must keep track of volts AND electrons as the light is varied to the pixels.? Recall that linear photon transfer only needs to measure output DN (or volts).? Electrons are readily calculated through the conversion constants .. .? . e-/DN or e-/V ? .. . . which are NOT? true relationships for nonlinear detectors (and I mean significantly !). We keep track of electrons by first taking a (very) low light level reading and applying the photon transfer equations above to find electrons.? For higher light levels, we (must) find electrons through exposure time from this reference point. . .? or reading a standard diode next to the detector who's output is linear with light level. In either case signal electrons are proportional to exposure time or standard diode photo current. At this point we now know how DN (or volts at the source follower output) varies with electrons (or Q).? We then convert source follower volts to sense node volts (for a good approximation we can just multiply by 1.2 and assume the source follower is linear).? This produces the plot Q vs sense node voltage.? Next the curve is differentiated (i.e., dQ/dV).? I've been doing this by first curve fitting Q vs V (typically this results in a log curve. . . provided by Excel trendline). .? and then differentiating (typically resulting in a k/V curve).? The result gives the sense node capacitance curve as a function of sense node voltage.. . . . our desired result. >From there you can find everything else. . .? e.g.,? true? sense sensitivity (V/e-) and e-/DN. I've been working on the same problem above when the source follower has some V/V nonlinearity. ..? . but that is another discussion. What do you think? :-)? Let me know if you have questions. .? or suggestions in making this photon transfer measurement technique more user friendly.? We can go from there. Take care.. . Jim PS. .? I thought I send this one to CCD World. . because I've received a half dozen inquiries on the subject. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/827b99bd/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 65519 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/827b99bd/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 77737 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/827b99bd/attachment-0001.jpe From CMOSCCD at aol.com Thu Jan 6 21:26:28 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 6 Jan 2005 19:26:28 EST Subject: CCD-world: e-/V vs V/V NONLINEARITY Message-ID: Folks, An addition to the last e-mail. . . . There are "good" and "bad" nonlinearities .. . as far as photon transfer is concerned. :-) Photon transfer can handle nonlinearity in the form of V/V (e.g., a nonlinear amplifier). It's the conversion gain (V/e-) that linear photon transfer is upset with. jj -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/23ebe6be/attachment.html From pmooney at gatan.com Thu Jan 6 22:49:40 2005 From: pmooney at gatan.com (Paul Mooney) Date: Thu, 6 Jan 2005 17:49:40 -0800 Subject: FW: CCD-world: NONLINEAR PHOTON TRANSFER - why not? In-Reply-To: <7a.6a371506.2f0f2b05@aol.com> Message-ID: Jim, I have been playing with this as a way to measure small deviations from linearity. When I do the math for a general case I get an ugly diff EQ: for c [counts] = f ( n [electrons] ) and the inverse function n = finv(c): var(c) = [f'(finv(c))]**2 * finv(c) taking the derivative of var(c) with respect to c: var'(c) = 2 * finv(c) * f''(finv(c)) + f'(finv(c)). The first term drops out for linear functions (f''=0) and f'(finv(c)) is just the gain, as expected. For general functions, I have to think some more. Maybe there is someone out there who can suggest an inversion method (either using var(c)-your curve- or var'(c)). I have to admit that I underestimated the problem. Paul _____ From: CMOSCCD at aol.com [mailto:CMOSCCD at aol.com] Sent: Thursday, January 06, 2005 4:00 PM To: pmooney at gatan.com; ccd-world at ctio.noao.edu; ccd-world at ctios2.ctio.noao.edu Subject: Re: FW: CCD-world: NONLINEAR PHOTON TRANSFER - why not? Jim, The basic photon-transfer method is differential for all intents and purposes since one obtains gain from measuring the slope of the variance vs. mean curve. That means that the bias signal doesn't create an error - as it would if one simply divided variance by mean at some particular dose level. For this reason, I don't understand why it wouldn't also work to simply generalize this concept for non-linear devices on the assumption that the rms variation of electron shot noise keeps the signal within a range for which the gain is approximately constant - and therefore doesn't mess up the basis assumptions needed for photon transfer. From this, then, the photon transfer method should in principal be able to measure the net response of non-linear capacitance and transistor gain by integration of the marginal gain. Am I missing something? Paul Mooney *************jj Paul, I (really) hope you are on to something. . . but .. . :-) Let me show you the raw photon transfer data below taken from a nonlinear detector for your comment. Here we plot Noise (DN) vs Signal (DN).. . . standard photon transfer format. Note, if we use the linearily assumed photon transfer equation. . . e-/DN = SIGNAL(DN) / NOISE(DN)^2 that e-/DN varies from 1.2 e-/DN to 19 e-/DN. We plot the result in the second plot in "red." The true result, using the nonlinear measuring technique described in a previous e-mail (below), is shown in "blue". As you can see. . there is a huge discrepancy in e-/DN and signal charge measured. Linear photon transfer fails. The question for you is. . . . how would you take the raw DN data below and get to the correct blue curve? . . . without having further information? Your turn. . . :-) Jim ****************jj There are several folks here working on APS devices who would be interested in your techniques for handling non-linear outputs. If you could send the details to me I will forward them to those interested. *************jj Barry. . . thanks for your interest. Lets start with the discussion below.. . .. . . <><> First. . . it is important to note that sense node capacitance C= dQ/dV applies to nonlinear detectors. . . which is a lot different than C = fixed as exhibited by linear CCDs. Therefore, we must keep track of volts AND electrons as the light is varied to the pixels. Recall that linear photon transfer only needs to measure output DN (or volts). Electrons are readily calculated through the conversion constants .. . . e-/DN = S(DN) / N(DN)^2 or e-/V = S(V) / N(V)^2 .. . . which are NOT true relationships for nonlinear detectors (and I mean significantly !). We keep track of electrons by first taking a (very) low light level reading and applying the photon transfer equations above to find electrons. For higher light levels, we (must) find electrons through exposure time from this reference point. . . or reading a standard diode next to the detector who's output is linear with light level. In either case signal electrons are proportional to exposure time or standard diode photo current. At this point we now know how DN (or volts at the source follower output) varies with electrons (or Q). We then convert source follower volts to sense node volts (for a good approximation we can just multiply by 1.2 and assume the source follower is linear). This produces the plot Q vs sense node voltage. Next the curve is differentiated (i.e., dQ/dV). I've been doing this by first curve fitting Q vs V (typically this results in a log curve. . . provided by Excel trendline). . and then differentiating (typically resulting in a k/V curve). The result gives the sense node capacitance curve as a function of sense node voltage.. . . . our desired result. >From there you can find everything else. . . e.g., true sense sensitivity (V/e-) and e-/DN. I've been working on the same problem above when the source follower has some V/V nonlinearity. .. . but that is another discussion. What do you think? :-) Let me know if you have questions. . or suggestions in making this photon transfer measurement technique more user friendly. We can go from there. Take care.. . Jim PS. . I thought I send this one to CCD World. . because I've received a half dozen inquiries on the subject. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/831ebe7e/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 65519 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/831ebe7e/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 77737 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050106/831ebe7e/attachment-0001.jpe From djs at aaocbn.aao.gov.au Fri Jan 7 08:14:02 2005 From: djs at aaocbn.aao.gov.au (Darren Stafford) Date: Fri, 07 Jan 2005 22:14:02 +1100 Subject: CCD-world: Dewar decontamination In-Reply-To: <7b.3bc0063e.2f0b050b@aol.com> References: <7b.3bc0063e.2f0b050b@aol.com> Message-ID: <41DE6EFA.3030402@aaocbn.aao.gov.au> Hi CCD World, The Anglo-Australian Observatory has 3 cryogenic schmidt cameras used in spectrographs. All suffer vacuum oil contamination visible as condensation on the field flattener near the ccd.. There has been a program to clean out these cameras and some interesting findings have arisen that are worth sharing. If you're interested please read the attached. Darren Stafford djs at aaocbn.aao.gov.au Manager Vacuum and CCD systems Anglo-Australian Observatory Coonabarabran Australia -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050107/094a08a5/attachment.html From rsmith at astro.caltech.edu Fri Jan 7 15:04:25 2005 From: rsmith at astro.caltech.edu (Roger Smith) Date: Fri, 07 Jan 2005 10:04:25 -0800 Subject: CCD-world: Dewar decontamination References: <7b.3bc0063e.2f0b050b@aol.com> <41DE6EFA.3030402@aaocbn.aao.gov.au> Message-ID: <41DECF29.81E8142E@astro.caltech.edu> Darren, Can you tell us how much time is saved by Nitrogen flushing versus simply pumping on a dewar which is heated to the same temperature used in your flushing method (50C)? The virtue of pumping a (hot) dewar without Nitrogen flushing is that a cold trap can be placed in the pumping line to prevent contamination of the pump. The stainless steel vacuum line between the dewar (in an oven) and the trap (outside the oven) still gets contaminated but this can be washed and baked afterwards. A distinct advantage of this method is that the cold trap can be removed periodically to check for buildup of contaminants so you can measure progress. I've never had a Residual Gas Analyzer, but I find that the smell of the condensates on the trap (when warmed) goes a long way to identifying the contaminant. You can tell how much contaminent there was, and often what it was. e.g. pump oil or vacuum grease. Of course this wont work for very low levels of contamination, but I don't think we are talking about this regime. -- Roger -------------- next part -------------- A non-text attachment was scrubbed... Name: rsmith.vcf Type: text/x-vcard Size: 456 bytes Desc: Card for Roger Smith Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050107/7df1df6f/attachment.vcf From CMOSCCD at aol.com Tue Jan 11 03:36:38 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Tue, 11 Jan 2005 01:36:38 EST Subject: CCD-world: NONLINEAR PHOTON TRANSFER II Message-ID: <1a0.2d853c9d.2f14cdf6@aol.com> Paul, A hard nut to crack. I don't think it possible to find e-/volts, signal(e-), noise(e-), e-/DN . . . from signal(V) and noise(V) data alone (as can be accomplished for a linear detector, through photon transfer, like a CCD). However, if we generally know how the sense node capacitance varies with sense node voltage, then electron information can be found through equation. For example, sense capacitance for photon transfer data shown in Figure 1 below is described by, Cs (V) where Cs(V) is the sense node capacitance, K is a constant and V is the sense node voltage. Figure 2 plots measured sense capacitance as a function of sense voltage. A perfect fit is made to Eq. (1) once K is determined. As it turns out, the constant K can be solved with voltage data, K where Vs is sense node voltage, Ns is the sense node noise voltage and Vi is the sense node voltage at zero charge (i.e., the reference voltage,VREF less reset feedthrough assuming hard reset is applied to the gate of the reset). Once K is found, electron signal and noise can be determined on the sense node. S(e-) where S(e-) is the sense node signal (electrons) as a function of voltage data. Figure 3 plots Eq. 3 along with measured data. . . showing good agreement. Eq. 1 could have been in a more complex form like Cs(V) where n is other than 1. K can't be found until n is known.. . which appears can only come from measurement. As explained in previous e-mails. . . sense capacitance, Cs(V), can be measured by photon transfer and keeping track of exposure time to find signal electrons as a funtion of sense node voltage (i.e., Figure 3). As can be seen, nonlinear behavior represents a horrible (scientific) problem for 3T photo diode CMOS pixels. This is why our focus is on charge coupled CMOS pixels (PPD, PG) where the sense node is decoupled from photo region. Good linearity is observed. Jim FIGURE 1 FIGURE 2 FIGURE 3 Jim, ? I have been playing with this as a way to measure small deviations from linearity.? When I do the math for a general case I get an ugly diff EQ: ? ??? for c [counts] finv(c): ? ??????? var(c) ? ??? taking the derivative of var(c) with respect to c: ? ??????? var'(c) ? The first term drops out for linear functions (f''=0) and f'(finv(c)) is just the gain, as expected. ? For general functions, I have to think some more.? Maybe there is someone out there who can suggest an inversion method (either using var(c)-your curve-?or var'(c)). ? I?have to admit that I underestimated the problem. ? Paul ? -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050111/a8061ee8/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 56455 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050111/a8061ee8/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 62559 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050111/a8061ee8/attachment-0001.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 54092 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050111/a8061ee8/attachment-0002.jpe From djs at aaocbn.aao.gov.au Tue Jan 11 02:52:39 2005 From: djs at aaocbn.aao.gov.au (Darren Stafford) Date: Tue, 11 Jan 2005 16:52:39 +1100 Subject: CCD-world: Dewar decontamination In-Reply-To: <41DECF29.81E8142E@astro.caltech.edu> References: <7b.3bc0063e.2f0b050b@aol.com> <41DE6EFA.3030402@aaocbn.aao.gov.au> <41DECF29.81E8142E@astro.caltech.edu> Message-ID: <41E369A7.70305@aaocbn.aao.gov.au> Roger, Pumping the heated dewar was tried but found to be ineffective for removing vacuum pump oil contamination, however, due to operational requirements we had access to the dewar for a maximum of 5 days at a time per month or two. The dewar concerned contains anodized surfaces that act like getters so even at elevated temperatures the oil is reluctant to leave the dewar wall. In this situation decontamination efficiency needs to be maximized. With low pressure nitrogen flushing each molecule is nabbed as soon as it leaves the dewar wall rather than allowed to bounce around the dewar until it falls down the vacuum port. For traditional pumping the probability of a particular molecule travelling into the vacuum port is a function of the area of the vacuum port / the surface area the dewar. This becomes extermely small when the surface area of the getter is considered(800m square per gram). The efficiency also trails off as the contamination is reduced. Also each molecule has a 50% chance of bouncing back in to the dewar from the pumping line. Your point about using a cold trap is a excellent one . I had one of those "why didn't I think of that?" experiences It may be possible to use a cold trap with low pressure nitrogen flushing. I'll do some tests and get back to you. Many Thanks Darren Roger Smith wrote: >Darren, > >Can you tell us how much time is saved by Nitrogen flushing versus >simply pumping on a dewar which is heated to the same temperature used >in your flushing method (50C)? > >The virtue of pumping a (hot) dewar without Nitrogen flushing is that a >cold trap can be placed in the pumping line to prevent contamination of >the pump. The stainless steel vacuum line between the dewar (in an >oven) and the trap (outside the oven) still gets contaminated but this >can be washed and baked afterwards. > >A distinct advantage of this method is that the cold trap can be removed >periodically to check for buildup of contaminants so you can measure >progress. I've never had a Residual Gas Analyzer, but I find that the >smell of the condensates on the trap (when warmed) goes a long way to >identifying the contaminant. You can tell how much contaminent there >was, and often what it was. e.g. pump oil or vacuum grease. Of >course this wont work for very low levels of contamination, but I don't >think we are talking about this regime. > > -- Roger > > > >-- CCD-world -- >CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu >Standard replies will go to the list; address personal replies manually. >For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world > > From waterson at ifa.hawaii.edu Tue Jan 11 15:22:16 2005 From: waterson at ifa.hawaii.edu (Mark Waterson) Date: Tue, 11 Jan 2005 08:22:16 -1000 Subject: CCD-world: IR remote controls In-Reply-To: References: Message-ID: <41E41958.8020203@ifa.hawaii.edu> I think it really depends on how expensive the remote is... The spec for home remote controls is >.4mW/sterRadian recieved at 850nm, something like 10m range, but I have seen remotes that won't work at half that distance. The IRDA standard for data links (use your pda for a cal source?) requires higher power (~10x, due to a much shorter on-time for the "0" code of 1.6 uS), at 960nm and 1m range. The emitters are available for <<$1 each from Digikey, typical output powers for some 950nm parts are 50-250mW/SR at 100ma (OSRAM), beam angles range from 8-40 degrees depending on package. Some further interesting/relevant info at mfw Mark Waterson University of Hawaii - Institute for Astronomy Haleakala Observatories, Maui 808-876-7600 x108 Kula Office 808-243-5892 Observatory waterson at ifa.hawaii.edu Burke, Thomas G. wrote: > Howdy all, > > I know this may be a little off-topic, but I'd like to ask - does anyone here > know the output power of an IR Remote control? Their outputs are generally > centered at one of a few discrete frequencies in the 800-950nm range, and I > know it's a fairly gaussian output, but I'm curious about output power, both > at the peak f, and integrated over f (eg total power)... Also, they > generally encode their outputs on a 30-60kHz square-wave carrier.. > > I'm putting together a detector, & am thinking of using a remote as a quick > "sanity check" type of thing to ensure it's working, but I don't want to > spend a heap of dineros on it. A remote seems right up the alley, but I need > to know output levels to ensure I don't saturate & so forth. > > Thanks, Tom > > > -- CCD-world -- CCD-world is fully moderated. Send posts to > CCD-world at ctio.noao.edu Standard replies will go to the list; address > personal replies manually. For more information, please go to: > http://www.ctio.noao.edu/mailman/listinfo/ccd-world > > From alain at spaceobs.com Tue Jan 11 20:57:06 2005 From: alain at spaceobs.com (Alain) Date: Tue, 11 Jan 2005 20:57:06 -0300 Subject: CCD-world: CCD window In-Reply-To: <41DECF29.81E8142E@astro.caltech.edu> Message-ID: <000c01c4f839$41989700$ae5f1cc8@SPACE> Hello all, I would like to know if there is a special technique in order to unglue the window glued on some of the commercial CCDs... A safe technique I mean... :) Thanks in advance, Alain From CMOSCCD at aol.com Wed Jan 12 13:16:08 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Wed, 12 Jan 2005 11:16:08 EST Subject: CCD-world: CCD window Message-ID: In a message dated 1/12/2005 5:17:41 AM Pacific Standard Time, alain at spaceobs.com writes: > Hello all, > I would like to know if there is a special technique in order to unglue > the window glued on some of the commercial CCDs... A safe technique I > mean... :) > Thanks in advance, > Alain > *********jj Alain, At JPL we used a hot plate. . . sometimes windows just fall off. Best to call manufacturer . . . and also ask for some mechanical samples to experiment. Who makes the CCD? Good luck :-) Jim **************jj -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050112/216322c0/attachment.html From Venkatraman.Iyer at kla-tencor.com Wed Jan 12 15:18:05 2005 From: Venkatraman.Iyer at kla-tencor.com (Iyer, Venkatraman) Date: Wed, 12 Jan 2005 10:18:05 -0800 Subject: CCD-world: CCD window Message-ID: <321D44776C348447B1440547A77C2F68504C07@CA1EXCLV03.adcorp.kla-tencor.com> You will need to give some more information on the type of glue and window. Is there a metal frame to the window? How far away from the surface of the active CCD? Preferably thickness of the window... -----Original Message----- From: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] On Behalf Of Alain Sent: Tuesday, January 11, 2005 3:57 PM To: 'Optical & SWIR imager development for professional astronomy' Subject: CCD-world: CCD window Hello all, I would like to know if there is a special technique in order to unglue the window glued on some of the commercial CCDs... A safe technique I mean... :) Thanks in advance, Alain -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world From vincekasprzak at frontiernet.net Wed Jan 12 20:06:13 2005 From: vincekasprzak at frontiernet.net (Vince Kasperzak) Date: Wed, 12 Jan 2005 15:06:13 -0800 Subject: CCD-world: CCD window References: Message-ID: <000501c4f8fb$80b17610$01fea8c0@vince> Hi Alain, Try begging this capability from a company that already routinely does this like Mike Buchin at Stanford Photonics, mbuchin at stanfordphotonics.com The process usually relies on heating the window and then perhaps thermally stressing by cooling. I've heard that this has also been done using a machining process. Not all window bonding is equal. Some windows will come off easier than others, depending on which CCD manufacturer you use. Some CCD manufacturers will supply CCDs with temporary (removable) windows. Vince Kasprzak ----- Original Message ----- From: CMOSCCD at aol.com To: alain at spaceobs.com ; ccd-world at ctio.noao.edu Sent: Wednesday, January 12, 2005 8:16 AM Subject: Re: CCD-world: CCD window In a message dated 1/12/2005 5:17:41 AM Pacific Standard Time, alain at spaceobs.com writes: Hello all, I would like to know if there is a special technique in order to unglue the window glued on some of the commercial CCDs... A safe technique I mean... :) Thanks in advance, Alain *********jj Alain, At JPL we used a hot plate. . . sometimes windows just fall off. Best to call manufacturer . . . and also ask for some mechanical samples to experiment. Who makes the CCD? Good luck :-) Jim **************jj ------------------------------------------------------------------------------ -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050112/3d9a42f9/attachment.html From CMOSCCD at aol.com Thu Jan 13 18:31:59 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 13 Jan 2005 16:31:59 EST Subject: CCD-world: ? SENSE CAPACITANCE ? Message-ID: <12c.55292feb.2f1842cf@aol.com> Solid Staters. . . We are picking up an interesting solid state characteristic exhibited by a 3T photo diode CMOS pixel. As can be seen from a nonlinear PT measurement below. . sense capacitance is inversely proportional to sense voltage (exactly!). >From SZE and past understandings. . . we thought that capacitance for a diode should go by the square root of voltage. As can also be seen in the plot. . PISCES modeling agrees very well with measurement. This is a pinned diode pixel (i.e., highly doped boron pinning implant on top of a lightly doped n region. . both which are over high epi resistivity). Any ideas ? Thanks, Jim 1/V PHOTO DIODE CHARACTERISTIC -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050113/92136513/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 58654 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050113/92136513/attachment.jpe From CMOSCCD at aol.com Thu Jan 13 18:43:11 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 13 Jan 2005 16:43:11 EST Subject: CCD-world: NONLINEAR PT III Message-ID: Folks, We also show another CMOS pixel plot below that compares signal(e-) measured by linear PT (i.e., e-/DN = S(DN) / N(DN)^2 is employed) to reality (i.e., nonlinear PT technique). It's amazing how far off linear PT takes you. . . . and the error comes up very quickly. I wonder if there are high sensitivity (V/e-) CCDs out there that have calibration problems (Kodak?)? As noted before, linear PT does in fact handle certain kinds of nonlinearity (e.g., the V/V kind). It highly disagrees with conversion gain nonlinearity (i.e., V/e- because of dQ/dV). Therefore, nonlinearity associated with the source follower is ok with linear PT (i.e., e-/DN = S(DN) / N(DN)^2 applies). Shown below is the nonlinearity related to a 0.3 x 0.6 um source follower. Bottom line. . .don't push it too hard. Jim -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050113/0345caa8/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 45743 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050113/0345caa8/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 63519 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050113/0345caa8/attachment-0001.jpe From kossov at silar.spb.ru Tue Jan 18 10:40:14 2005 From: kossov at silar.spb.ru (Kossov) Date: Tue, 18 Jan 2005 16:40:14 +0300 Subject: CCD-world: Epi wafers for CCDs Message-ID: Hello CCD World! Does anybody know what companies produce 100 mm epitaxial wafers for manufacturing of CCDs? Vladimir Kossov Electron Optronic mailto:v.kossov at mail.ru From jbeletic at rwsc.com Wed Jan 19 03:00:13 2005 From: jbeletic at rwsc.com (James W Beletic) Date: Tue, 18 Jan 2005 22:00:13 -0800 Subject: CCD-world: SDW2005 registration opens on Monday Message-ID: Dear CCD-World (and CMOS and IR Worlds!), We write to inform you that registration for the Scientific Detector Workshop 2005 to be held in Taormina, Sicily (June 19-25, 2005) will open this Monday, January 24, 2005. We delayed release of the Web pages and opening registration while we finalized all details of the hotel and other arrangements. Everything is now set and with a few more modifications to the Web site, we will open registration on Monday. Thank you for your patience. We can assure you that this workshop will be the best ever, and hope you can attend. Ciao, Jim e Paola -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050118/0d9bc4dd/attachment.html From CMOSCCD at aol.com Wed Jan 19 21:35:24 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Wed, 19 Jan 2005 19:35:24 EST Subject: CCD-world: Epi wafers for CCDs Message-ID: <1dc.34cd6219.2f2056cc@aol.com> In a message dated 1/18/2005 6:16:23 AM Pacific Standard Time, kossov at silar.spb.ru writes: Does anybody know what companies produce 100 mm epitaxial wafers for manufacturing of CCDs? ********jj Try. . . Wacker's Burghausen site has 130 crystal pullers, half of ... Siltronic's Burghausen plant produces the widest range of ... wafers for everything from 4-inch (100-mm ... _http://www.micromagazine.com/archive/02/07/facilityrprt.html_ (http://www.micromagazine.com/archive/02/07/facilityrprt.html) *********jj -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050119/d9262de4/attachment.html From ltqf341 at ifa.cetef.csic.es Mon Jan 24 06:41:34 2005 From: ltqf341 at ifa.cetef.csic.es (Alejandro Ferrero) Date: Mon, 24 Jan 2005 10:41:34 +0100 Subject: CCD-world: NONLINEAR PHOTON TRANSFER II In-Reply-To: <1a0.2d853c9d.2f14cdf6@aol.com> Message-ID: Jim, I dont know if I have understood it well. Firsty, we have been studied the linearity in a Sony CCD as it follows: With an uniform an constant irradiance, we have registered Ni-Nio (illumination minus dark in DN) versus integration time (1000-10000 us). Integration time is proportional to charge and Ni-Nio is proportional to V (?). Then (dont taking in account MOSFET linearity), the amount tint/(Ni-Nio) should be proportional to sense capacitance (constant in linear CCDs and no not constant in no linear CCDs). I am right? Well, we have obtained similar data to your figure 3 (tint/((Ni-Nio) versus Ni)), but we observed a slightly increasing in capacitance at high Ni. Is our method correct? Why this slighty increase in capacitance? I enclose my figure. One thing more. In you textbook "Scientific Charge Coupled devices", you write (6.2.9.1): "The depeletion width decreases when charge is transfered onto the sense node and, in turn, increases the node capacitance and lowers the sensitivity." But we have seen that the node capacitance decreases with V. I dont understand. What am I misunderstanding? Excuse my awful english and thank you. Alejandro Ferrero Turri?n Instituto de F?sica Aplicada (CSIC) c/Serrano 144, 28006 Madrid Tel: (34) 91 5618806 Fax: (34) 91 4117651 email: aferrero at ifa.cetef.csic.es -----Mensaje original----- De: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu]En nombre de CMOSCCD at aol.com Enviado el: Martes 11 de Enero de 2005 07:37 AM Para: pmooney at gatan.com; ccd-world at ctio.noao.edu; ccd-world at ctios2.ctio.noao.edu Asunto: CCD-world: NONLINEAR PHOTON TRANSFER II Paul, A hard nut to crack. I don't think it possible to find e-/volts, signal(e-), noise(e-), e-/DN . . . from signal(V) and noise(V) data alone (as can be accomplished for a linear detector, through photon transfer, like a CCD). However, if we generally know how the sense node capacitance varies with sense node voltage, then electron information can be found through equation. For example, sense capacitance for photon transfer data shown in Figure 1 below is described by, Cs (V) (1) where Cs(V) is the sense node capacitance, K is a constant and V is the sense node voltage. Figure 2 plots measured sense capacitance as a function of sense voltage. A perfect fit is made to Eq. (1) once K is determined. As it turns out, the constant K can be solved with voltage data, K where Vs is sense node voltage, Ns is the sense node noise voltage and Vi is the sense node voltage at zero charge (i.e., the reference voltage,VREF less reset feedthrough assuming hard reset is applied to the gate of the reset). Once K is found, electron signal and noise can be determined on the sense node. S(e-) where S(e-) is the sense node signal (electrons) as a function of voltage data. Figure 3 plots Eq. 3 along with measured data. . . showing good agreement. Eq. 1 could have been in a more complex form like Cs(V) where n is other than 1. K can't be found until n is known.. . which appears can only come from measurement. As explained in previous e-mails. . . sense capacitance, Cs(V), can be measured by photon transfer and keeping track of exposure time to find signal electrons as a funtion of sense node voltage (i.e., Figure 3). As can be seen, nonlinear behavior represents a horrible (scientific) problem for 3T photo diode CMOS pixels. This is why our focus is on charge coupled CMOS pixels (PPD, PG) where the sense node is decoupled from photo region. Good linearity is observed. Jim FIGURE 1 FIGURE 2 FIGURE 3 Jim, I have been playing with this as a way to measure small deviations from linearity. When I do the math for a general case I get an ugly diff EQ: for c [counts] finv(c): var(c) taking the derivative of var(c) with respect to c: var'(c) The first term drops out for linear functions (f''=0) and f'(finv(c)) is just the gain, as expected. For general functions, I have to think some more. Maybe there is someone out there who can suggest an inversion method (either using var(c)-your curve- or var'(c)). I have to admit that I underestimated the problem. Paul -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/d4e53189/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 56455 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/d4e53189/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 62559 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/d4e53189/attachment-0001.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 54092 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/d4e53189/attachment-0002.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/bmp Size: 1601670 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/d4e53189/attachment.bmp From pamico at swfactory.org Mon Jan 24 14:54:43 2005 From: pamico at swfactory.org (Paola Amico) Date: Mon, 24 Jan 2005 14:54:43 US/Mountain Subject: CCD-world: The SDW2005 web site and registrations are open! Message-ID: <20050124215443.18228.qmail@webmail.networkshosting.com> An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050124/4f222a08/attachment.html From cyril.cavadore at fr.thalesgroup.com Tue Feb 1 09:38:44 2005 From: cyril.cavadore at fr.thalesgroup.com (cyril cavadore) Date: Tue, 01 Feb 2005 13:38:44 +0100 Subject: CCD-world: CMOS sensors References: Message-ID: <41FF7854.A4DC2D6E@fr.thalesgroup.com> Dear Folks, For astronomy imaging I have noticed the oustanding performance of a Canon 20D digital camera. This camera has a 8 million pixel sensor (15x22mm), 6.5?m square pixels + bayer color filter. I decided to measure the readnoise and dark current of the embedded CMOS detector by using RAW files that the camera is able to output. It turns out the files to be proper for these measurments : raw 12 bits data, no Bayer to RGB reconstruction, consistent statistical behavior. I was amazed by the resulting figures, I could get, at ISO 800, 5e- of Readout noise (3.5e- at ISO 1600). Knowing that the whole 8 millions pixels sensor can be read out in less than 200 ms yes ! 0.2 sec ! (multiport readout ?), it looks to be very impressive figures. Also, regarding dark current measurements, I could measure 0.1 pA/cm2 at room temperature, which is 50 less than a Kodak CCD (known for they low dark current level). The sensor is able to withstand more than 300sec exposure without hexibiting large dark current (at room temperature, I must emphasize). Cosmetically, the device is perfect. Report can be seen here : http://astrosurf.com/cavadore/APN/carac/EOS20D/index.html I will not speak here about quantum efficiency of those devices, nor about the dynamic, nevertheless CANON seems to have achieved a very nice sensor. I could guess the readout conversion factor is large, helping to get low noise. I know this digital camera has also an embedded processor, I wonder how far this "guy" is changing the RAW data (if so). At least I have noticed an offset correction versus long exposures times. If someone has more informations about this detector... I would be interesting to know. If not, one can wonder if our CCDs used for Astronomy turns out to become dinosaurs those days ;-) Cyril Cavadore From djm at aaoepp.aao.gov.au Tue Feb 1 22:00:20 2005 From: djm at aaoepp.aao.gov.au (Don Mayfield) Date: Wed, 02 Feb 2005 12:00:20 +1100 Subject: CCD-world: E2V hot rows Message-ID: <42002624.2020707@aaoepp.aao.gov.au> Hi all, I have a problem with two E2V CCD44-82 detectors, a blue sensitive and a red frame transfer device. The problem is that a dark frame taken with an exposure time greater than a few tens of seconds shows bright pixels (200 seconds results in pixels up to 30e- above the bias level) in the first few rows adjacent to the serial readout register. There are several hundred bright pixels scattered randomly across the full width of the chip in these rows. These devices are being operated at settings very close to manufacturers specs. Has anyone noticed this effect and/or been able to explain or fix it. Best regards, -- Don Mayfield Anglo-Australian Observatory 167 Vimiera Rd Eastwood NSW 2122 Australia Ph. 61-2-9372-4836 Fax. 61-2-9372-4880 From djm at aaoepp.aao.gov.au Wed Feb 2 18:54:20 2005 From: djm at aaoepp.aao.gov.au (Don Mayfield) Date: Thu, 03 Feb 2005 08:54:20 +1100 Subject: CCD-world: E2V hot rows. Message-ID: <42014C0C.8090505@aaoepp.aao.gov.au> To Brian, Thanks for the reply, I have tried integrating without the serial clocks running but this didn't help, I also tried various clock and bias voltages with no success. Since no one seems to have seen this particular problem I must assume its the way I'm driving the chips. To Paul, Thanks for your interest and yes it is the same problem I discussed with you before. I'll put together some data for you when time permits. Cheers, -- Don Mayfield Anglo-Australian Observatory 167 Vimiera Rd Eastwood NSW 2122 Australia Ph. 61-2-9372-4836 Fax. 61-2-9372-4880 From apo at mso.anu.edu.au Wed Feb 2 19:06:21 2005 From: apo at mso.anu.edu.au (Dr. Paddy Oates) Date: Thu, 03 Feb 2005 09:06:21 +1100 Subject: CCD-world: E2V hot rows In-Reply-To: <42002624.2020707@aaoepp.aao.gov.au> References: <42002624.2020707@aaoepp.aao.gov.au> Message-ID: <42014EDD.60209@mso.anu.edu.au> Don Is this charge leaking back form the serial register into the image area - the Tek I have here and the e2v 4240 both show this effect - trick is to do serial scanning during the integration to clear out the register. I have some pretty remarkable frames showing this effect growing up the image area as the dark frame integration time increased. Serial idling stopped the problem straight away... Cheers Paddy -- ++Surfing in CyberSpace on the Wings of a Storm++++++\ Senior Detector Engineer. MSSSO, RSAA. | Ph: +61 2 6125 8909 Fax: +61 2 6125 5635 | W: http://www.mso.anu.edu.au/observing/detectors/ | e: apo at mso.anu.edu.au | -----------------------------------------------------/ From rsmith at astro.caltech.edu Wed Feb 2 21:48:59 2005 From: rsmith at astro.caltech.edu (Roger Smith) Date: Wed, 02 Feb 2005 16:48:59 -0800 Subject: CCD-world: E2V hot rows References: <42002624.2020707@aaoepp.aao.gov.au> <42014EDD.60209@mso.anu.edu.au> Message-ID: <420174FA.923A5FA5@astro.caltech.edu> "Dr. Paddy Oates" wrote: > Don > Is this charge leaking back form the serial register into the image > area - the Tek I have here and the e2v 4240 both show this effect - > trick is to do serial scanning during the integration to clear out the > register. I have some pretty remarkable frames showing this effect > growing up the image area as the dark frame integration time increased. > Serial idling stopped the problem straight away... > > Cheers > Paddy > Don, are the bright pixels only seen immediately adjacent to the serial register? I've also seen the effect Paddy refers to when a fast erase allows the serial register to saturate. I'm convinced that the effect it is due to surface traps. But hey, this chip has a dump drain so you can throw the charge overboard without any serial clocking. If your problem was surface trapping in a saturated serial register it could realse charge for several consecutive lines giving hot pixels at the trap sites. The traps would be delivering charge to all trailing pixels but slowly enough that the effect (smeared along the line) would not be noticeable except when serial clocking is paused for the vertical transfer. You can test this by putting a pause in after a few horizontal clocks to make a double image of the trap locations. ....Er, the above text was all fine until I learned a moment ago that the effect doesn't depend on prior illumination. It was fun while it lasted. (:-( -- Roger -------------- next part -------------- A non-text attachment was scrubbed... Name: rsmith.vcf Type: text/x-vcard Size: 456 bytes Desc: Card for Roger Smith Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050202/1cc5800f/attachment.vcf From CMOSCCD at aol.com Mon Feb 7 13:18:12 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Mon, 7 Feb 2005 11:18:12 EST Subject: CCD-world: CMOS READ NOISE Message-ID: <1ef.350c482a.2f38eec4@aol.com> Folks, CMOS read noise is very interesting. . . completely different behavior than CCD noise. . . even though both technologies read pixels with source follower amplifiers. CMOS amplifiers are physically very small compared to CCD amplifiers (0.32um x 0.55um are common) .. . so much so, that individual surface channel traps can significantly shake and influence the entire mobility of the drain bias current. Such noise generation is called Random Telegraph Noise (RTS). Figure 1 below shows an RTS signal taken directly from an output of a CMOS pixel source follower. The time response shows that two traps are at work. Every pixel in a CMOS array has a different RTS signature. Figure 2 shows a small collection of noise histograms taken from individual pixels of a CMOS array. Note that the trap (1/f) noise modulates the source follower's white noise floor (seen here as a Gaussian distribution). The double noise peaks indicate a trap turns the source follower drain current "off" and "on" equally. The first plot shown is multi peaked indicating multiple channel traps are at work. The net noise floor from a CMOS array is therefore complex (relative to CCD). Figure 3 plots read noise (electrons rms) as a function of pixel number. N ote that read noise varies between 4 - 20 e- rms along with a few very noisy pixels that are also present. The 4 e- pixels are essentially trap free. . and the read noise is determined by white (thermal) noise of the source follower (and therefore, influenced by the bandwidth of the signal CDS processor. . where RTS is not). Adding to the complexity above, RTS trap noise is influenced by the amount of drain current that flows. For example, Figure 4 plots read noise for a single very high noise CMOS pixel as a function of drain current. In general, trap noise increases with drain current and decreases as source follower geometry increases.. . . as 1/f noise should. The CMOS array characterized here has been designed with low pixel sensitivity (approximately 11 uV/e-) in order to achieve high full well (approximately 175,000 e-). In general, CMOS read noise decreases with increasing sensitivity (as do CCDs). High sensitivity CMOS pixels that are trap free exhibit read noise levels close to one electron. Amazing. . Jim . . . . the same RTS trap effect is also related to dark spikes (hot pixels) that randomly turn "off" and "on" in CCDs. Atomic effects in action. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/20dc7e49/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 45387 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/20dc7e49/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 67545 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/20dc7e49/attachment-0001.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 83681 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/20dc7e49/attachment-0002.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 66585 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/20dc7e49/attachment-0003.jpe From CMOSCCD at aol.com Mon Feb 7 16:22:21 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Mon, 7 Feb 2005 14:22:21 EST Subject: CCD-world: CMOS sensors Message-ID: <67.3e0f1ef8.2f3919ed@aol.com> ***********************jj Cyril, Commercial CCD and CMOS cameras do a frame dark current subtraction (i.e., pixel by pixel despiking). . eliminating average dark current and fixed pattern noise. So the comparison below is not fair to Kodak. . who quotes unprocessed dark current. The Canon dark current is specified at 60 pA at 60 C.. . . which is about 5 pA/cm^2 at room temp. . . see table below. Nevertheless. . . dark current is unusually very good for a CMOS array (but not as good at 0.1 pA). My $300 Sony 5.2 M CCD camera shows similar dark current characteristics (see below. . a 30 second room temperature image). Both the Canon CMOS and Sony CCD are pinned diode pixels. Silicon is silicon. CMOS and CCD will coexist. . . even as dinosaurs.. . . :-) Jim *******************jj Dear Folks, For astronomy imaging I have noticed the oustanding performance of a Canon 20D digital camera. This camera has a 8 million pixel sensor (15x22mm), 6.5?m square pixels + bayer color filter.? I decided to measure the readnoise and dark current of the embedded CMOS detector by using RAW files that the camera is able to output. It turns out the files to be proper for these measurments : raw 12 bits data, no Bayer to RGB reconstruction, consistent statistical behavior. I was amazed by the resulting figures, I could get, at ISO 800, 5e- of Readout noise (3.5e- at ISO 1600). Knowing that the whole 8 millions pixels sensor can be read out in less than 200 ms yes ! 0.2 sec ! (multiport readout ?), it looks to be very impressive figures. Also, regarding dark current measurements, I could measure 0.1 pA/cm2 at room temperature, which is 50 less than a Kodak CCD (known for they low dark current level). The sensor is able to withstand more than 300sec exposure without hexibiting large dark current (at room temperature, I must emphasize). Cosmetically, the device is perfect. Report can be seen here : http://astrosurf.com/cavadore/APN/carac/EOS20D/index.html I will not speak here about quantum efficiency of those devices, nor about the dynamic, nevertheless CANON seems to have achieved a very nice sensor. I could guess the readout conversion factor is large, helping to get low noise. I know this digital camera has also an embedded processor, I wonder how far this "guy" is changing the RAW data (if so). At least I have noticed an offset correction versus long exposures times. If someone has more informations about this detector... I would be interesting to know. If not, one can wonder if our CCDs used for Astronomy turns out to become dinosaurs those days ;-) Cyril Cavadore -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/69ad218f/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 16491 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/69ad218f/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 31653 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050207/69ad218f/attachment-0001.jpe From CMOSCCD at aol.com Wed Feb 9 04:26:37 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Wed, 9 Feb 2005 02:26:37 EST Subject: CCD-world: CMOS RTS NOISE - DOUBLE SAMPLING Message-ID: <15c.49bee88d.2f3b152d@aol.com> CMOS RTS noise addition. . . As the operating temperature is reduced. . . the frequency of RTS noise decreases (i.e., the emission time constant of the trap increases). In turn, RTS noise is reduced by CDS processing (see Figures 1 and 2 below). Amazing. . . -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050209/8d4eee6f/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 33972 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050209/8d4eee6f/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 36249 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050209/8d4eee6f/attachment-0001.jpe From JDefise at spd.aas.org Thu Feb 10 06:06:44 2005 From: JDefise at spd.aas.org (Jean-Marc Defise) Date: Thu, 10 Feb 2005 10:06:44 +0100 Subject: CCD-world: photon transfer curve with scintillator References: <005201c469bc$8a029650$efc7a58b@sedna> Message-ID: <021a01c50f4f$d84b3e50$ebb2a58b@csl.ulg.ac.be> Dear Jim & CCD World, Back to our CMOS device with scintillator coating for EUV imaging (13 to 60 nm) (see thread below). We have new results but still with a preliminary readout electronics. They indicate interesting outcomes: a.. e-/DN in the visible range is of the order of 5.5 e-/DN b.. e-/DN in the EUV range is of the order of 4.5 e-/DN This was obtained with statistical analysis of 30x30 pixel subframes of successive difference images (same illumination) using white light for visible PTC and synchroton beam for EUV. In the visible range, the scintillator is transparent. PTC gives the electronic gain. In the EUV, PTC results include the conversion of EUV photons in 54.5 nm photons. This conversion is not well documented in the litterature. If it is a one-to-one conversion (your experience of Lumigen with Lyman-alpha), we loose half of the generated photons, as the emission direction is random thus we loose all the "backwards" photons. It would reduce the EUV e-/DN by a factor 2. In our case, surprisingly, the PTC results gives a conversion of 1(incident to scintillator)-to-1.2 (in the CMOS), which is much better. This means 1 EUV photon is converted in 2.4 visible photons randomly oriented. Even with such a conversion rate, there are still a lot of losses in the scintillator, comparing the energy of one EUV photon and the visible photons produced. Does this make sense ? Thanks for your thoughts on this, Jean-Marc ================= Jean-Marc Defise Space Science Instruments Group CENTRE SPATIAL DE LIEGE Av Pre Aily - B 4031 ANGLEUR Tel/Fax: 32-43-676668 / 5613 www.ulg.ac.be/cslulg ================= ----- Original Message ----- From: CMOSCCD at aol.com To: CCD-world at astro.ku.dk Sent: Wednesday, July 07, 2004 8:30 PM Subject: Re: CCD-world: photon transfer curve with scintillator ****************************jj Jean-Marc, Yes. . it would be interesting to compare the gain coefficient . . electrons/digital number in the visible and EUV. Lumigen working at Lyman-alpha (1216A) did not exhibit a difference in shot noise characteristics indicating 1216 A to 5300 A conversion was complete. However, shorter wavelengths (< 400 A) showed a decrease in the conversion constant (presumably, EUV photons went through the thin lumigen layer and interacted with silicon directly). Please let us know what you detect. . . . results would be very interesting. By the way we are seeing the QE pinned condition for backside illuminated CMOS detectors. Would going direct offer more sensitivity? Jim Janesick ********************************jj Dear CCD-World, We are currently experimenting a CMOS detector with a scintillator (phosphor-like) coating, to provide sensitivity in the extreme UV range (15 to 30 nm) for a space telescope. I want to use the photon transfer curve (PTC) method to derive the effective sensitivity in the EUV and disantangle it from QE in the visible. We already have the gain (from electronic measurements, but should be confirmed with PTC with visible light). I suspect that conversion processes in the scintillator may add some statistical components to be accounted with PTC. Did any of you already elaborate further in that method with scintillator coatings ? Lumigen was already used in the past and was probably tested with PTC. I would also be interested to share EUV response measurements of Gd2O2S:Tb coating, which seems to be investigated only in the X-ray range for medical applications. Thanks for your inputs. Regards, Jean-Marc Defise -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050210/874cf32d/attachment.html From CMOSCCD at aol.com Fri Feb 11 18:29:09 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Fri, 11 Feb 2005 16:29:09 EST Subject: CCD-world: CMOS RTS NOISE vs TEMPERATURE Message-ID: <1a3.2e62ee91.2f3e7da5@aol.com> Gang, Theory also shows that CMOS RTS read noise will (always) decrease as temperature is lowered. The graph below plots read noise (source follower white + RTS) as a function of CDS clamp-to-sample time. We assume a single 0.35 eV trap is involved in generating RTS noise (purposely made very high to see the full effect with temperature relative to white noise).? The CCD noise equations used in 1/f noise analysis also work on RTS noise (the plot shown below was taken from my book that replaces 1/f noise with RTS for).? Read noise for CMOS source follower amps without RTS do not show a temperature dependence. These amps are probably trapless and only generate white noise (which is fairly temperature independent). Read noise for these amps (pixels) is 3.5 to 4 e-rms for the sensor measured here. It appears that all pixels will exhibit this same noise level if the device can be cooled enough to eliminate RTS. This characteristic is good news if cooling is an option. Jim PS. . check out Random telegraph signal noise simulation of decanano MOSFETs ... - ... Wel, EAM Klumperink, LKJ Vandamme, B. Nauta, Modeling random telegraph noise under switched ... the device scaling on the low-frequency noise in n-MOSFETs, Appl ... http://userweb.elec.gla.ac.uk/a/abrown/papers/SAM03_Lee.pdf which models the amplitude of RTS as function of trap position within the source follower channel. -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050211/aa608413/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 66976 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050211/aa608413/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 39737 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050211/aa608413/attachment-0001.jpe From alain at spaceobs.com Sun Feb 13 18:18:05 2005 From: alain at spaceobs.com (Alain Maury) Date: Sun, 13 Feb 2005 18:18:05 -0300 Subject: CCD-world: CMOS sensors In-Reply-To: <67.3e0f1ef8.2f3919ed@aol.com> Message-ID: <000001c51211$9c3dffa0$db5f1cc8@portablealain> Here comes a question then : In order to make an inexpensive survey instrument, something at the focus of a fast (short exposure time, photon limited) telescope it seems today better to use CMOS sensors rather than thinned CCDs. On one side, maybe only 40% of DQE, but fast readout, very inexpensive, almost no cooling, on the other a very high price tag, necessity to cool to -100?, slow readout. So if one has a large field of view, a smaller DQE can be compensated with more detector surface. It is easy to compare both, even with a 25% fill factor (I mean detectors separated by their own spacing in the focal plane, i.e. 24mm away between each 24mm detectors), the efficiency is very high for CMOS (using a constant $ amount to build the detector). So far I have found black and white CMOS sensors up to 35mm films for roughly speaking 1000 euros a piece, but the question to this distinguished list is : Are there larger CMOS sensors around ? buttable ones ? (I don?t believe, just asking). Alain -----Mensaje original----- De: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] En nombre de CMOSCCD at aol.com Enviado el: lundi 7 f?vrier 2005 16:22 Para: ccd-world at ctio.noao.edu Asunto: Re: CCD-world: CMOS sensors ***********************jj Cyril, Commercial CCD and CMOS cameras do a frame dark current subtraction (i.e., pixel by pixel despiking). . eliminating average dark current and fixed pattern noise. So the comparison below is not fair to Kodak. . who quotes unprocessed dark current. The Canon dark current is specified at 60 pA at 60 C.. . . which is about 5 pA/cm^2 at room temp. . . see table below. Nevertheless. . . dark current is unusually very good for a CMOS array (but not as good at 0.1 pA). My $300 Sony 5.2 M CCD camera shows similar dark current characteristics (see below. . a 30 second room temperature image). Both the Canon CMOS and Sony CCD are pinned diode pixels. Silicon is silicon. CMOS and CCD will coexist. . . even as dinosaurs.. . . :-) Jim *******************jj Dear Folks, For astronomy imaging I have noticed the oustanding performance of a Canon 20D digital camera. This camera has a 8 million pixel sensor (15x22mm), 6.5?m square pixels + bayer color filter. I decided to measure the readnoise and dark current of the embedded CMOS detector by using RAW files that the camera is able to output. It turns out the files to be proper for these measurments : raw 12 bits data, no Bayer to RGB reconstruction, consistent statistical behavior. I was amazed by the resulting figures, I could get, at ISO 800, 5e- of Readout noise (3.5e- at ISO 1600). Knowing that the whole 8 millions pixels sensor can be read out in less than 200 ms yes ! 0.2 sec ! (multiport readout ?), it looks to be very impressive figures. Also, regarding dark current measurements, I could measure 0.1 pA/cm2 at room temperature, which is 50 less than a Kodak CCD (known for they low dark current level). The sensor is able to withstand more than 300sec exposure without hexibiting large dark current (at room temperature, I must emphasize). Cosmetically, the device is perfect. Report can be seen here : http://astrosurf.com/cavadore/APN/carac/EOS20D/index.html I will not speak here about quantum efficiency of those devices, nor about the dynamic, nevertheless CANON seems to have achieved a very nice sensor. I could guess the readout conversion factor is large, helping to get low noise. I know this digital camera has also an embedded processor, I wonder how far this "guy" is changing the RAW data (if so). At least I have noticed an offset correction versus long exposures times. If someone has more informations about this detector... I would be interesting to know. If not, one can wonder if our CCDs used for Astronomy turns out to become dinosaurs those days ;-) Cyril Cavadore -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050213/72a53a77/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 16491 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050213/72a53a77/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 25780 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050213/72a53a77/attachment-0001.jpe From CMOSCCD at aol.com Mon Feb 14 04:09:08 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Mon, 14 Feb 2005 02:09:08 EST Subject: CCD-world: ? SIGNAL -TO -NOISE ? Message-ID: <190.397f51d7.2f41a894@aol.com> CCD (CMOS?) WORLD ;-) Ideally, signal-to-noise (S/N) should be proportional to the square-root of the signal (i.e., shot noise limit). For CMOS detectors it is (very) common to measure a S/N that is (much) greater than theory. This result is physically impossible. .. .. . so what gives ? S/N is often defined as. .. . S/N = SIGNAL(DN) / NOISE (DN) (1) where DN = output digital number or equivalently, S/N = SIGNAL(V) / NOISE (V) (2) where V = output volts These S/N equations are only valid if the conversion gain (V / electron) is constant. If the conversion gain decreases with signal then the equations above will produce a S/N result greater than the shot noise limit. For example, the figure below plots S/N from data generated by a (3T) CMOS imager for flat field illumination. Equation 1 is used to find S/N. Also plotted is the true S/N that is based on the absolute S/N equation, S/N = SIGNAL (e-) / NOISE (e-) (3) where e- = electrons The discrepancy between Eqs. (1) and (3) are striking and shows that familiar S/N equations as (1) and (2) are in error when applied on nonlinear conversion gain detectors like CMOS. However, the equations are still valid for linear and nonlinear gain detectors (e.g., CCD). Equation 3 is always problem free. Jim -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050214/3141d162/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 119255 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050214/3141d162/attachment.jpe From cyril.cavadore at fr.thalesgroup.com Mon Feb 14 06:43:13 2005 From: cyril.cavadore at fr.thalesgroup.com (cyril cavadore) Date: Mon, 14 Feb 2005 10:43:13 +0100 Subject: CCD-world: CMOS sensors In-Reply-To: <67.3e0f1ef8.2f3919ed@aol.com> References: <67.3e0f1ef8.2f3919ed@aol.com> Message-ID: <421072B1.3090803@fr.thalesgroup.com> Jim, To measure the dark current, I haven't used the dark signal offset increase (because it is corrected), instead I have considered the increase of pixel noise, knowing the bias floor noise and the conversion factor, I could compute the actual dark signal increase. Nevertheless, i would not be surprised, despite the raw mode, that a kind of trick is applied to the data. If you have more information about the sensor used by Canon (EOS20D cameras)... like a datasheet or a paper ;-) Cyril CMOSCCD at aol.com a ?crit : > > > > ***********************jj > Cyril, > > Commercial CCD and CMOS cameras do a frame dark current subtraction > (i.e., pixel by pixel despiking). . eliminating average dark current > and fixed pattern noise. So the comparison below is not fair to Kodak. > . who quotes unprocessed dark current. The Canon dark current is > specified at 60 pA at 60 C.. . . which is about 5 pA/cm^2 at room temp. > . . see table below. Nevertheless. . . dark current is unusually very > good for a CMOS array (but not as good at 0.1 pA). > > > > > My $300 Sony 5.2 M CCD camera shows similar dark current characteristics > (see below. . a 30 second room temperature image). Both the Canon CMOS > and Sony CCD are pinned diode pixels. Silicon is silicon. > > CMOS and CCD will coexist. . . even as dinosaurs.. . . :-) > > Jim > > > > *******************jj > > > > Dear Folks, > > For astronomy imaging I have noticed the oustanding performance of a > Canon 20D digital camera. This camera has a 8 million pixel sensor > (15x22mm), 6.5?m square pixels + bayer color filter. > I decided to measure the readnoise and dark current of the embedded CMOS > detector by using RAW files that the camera is able to output. It turns > out the files to be proper for these measurments : raw 12 bits data, no > Bayer to RGB reconstruction, consistent statistical behavior. > > I was amazed by the resulting figures, I could get, at ISO 800, 5e- of > Readout noise (3.5e- at ISO 1600). Knowing that the whole 8 millions > pixels sensor can be read out in less than 200 ms yes ! 0.2 sec ! > (multiport readout ?), it looks to be very impressive figures. > Also, regarding dark current measurements, I could measure 0.1 pA/cm2 at > room temperature, which is 50 less than a Kodak CCD (known for they low > dark current level). > The sensor is able to withstand more than 300sec exposure without > hexibiting large dark current (at room temperature, I must emphasize). > Cosmetically, the device is perfect. > > Report can be seen here : > http://astrosurf.com/cavadore/APN/carac/EOS20D/index.html > > I will not speak here about quantum efficiency of those devices, nor > about the dynamic, nevertheless CANON seems to have achieved a very nice > sensor. I could guess the readout conversion factor is large, helping to > get low noise. > I know this digital camera has also an embedded processor, I wonder how > far this "guy" is changing the RAW data (if so). At least I have noticed > an offset correction versus long exposures times. > If someone has more informations about this detector... I would be > interesting to know. If not, one can wonder if our CCDs used for > Astronomy turns out to become dinosaurs those days ;-) > > Cyril Cavadore > > > > ------------------------------------------------------------------------ > > > -- CCD-world -- > CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu > Standard replies will go to the list; address personal replies manually. > For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world From tg.burke at ngc.com Tue Feb 15 14:56:15 2005 From: tg.burke at ngc.com (Burke, Thomas G.) Date: Tue, 15 Feb 2005 12:56:15 -0500 Subject: CCD-world: CMOS sensors Message-ID: Rockwell (?) is making a 2-side buttable 4kx4k (IIRC)... We have been experimenting with development of a 4-side buttable array, but I don't know where this is right now, as I haven't worked the program in a couple of years. SiTE made some 3-side (CCD) buttables (but they're out of business, now?), and I believe Fairchild may be making some end-buttable (2 opposing sides) arrays (but CCD, I think) -----Original Message----- From: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu]On Behalf Of Alain Maury Sent: Sunday, February 13, 2005 4:18 PM To: 'Optical & SWIR imager development for professional astronomy' Subject: RE: CCD-world: CMOS sensors Here comes a question then : In order to make an inexpensive survey instrument, something at the focus of a fast (short exposure time, photon limited) telescope it seems today better to use CMOS sensors rather than thinned CCDs. On one side, maybe only 40% of DQE, but fast readout, very inexpensive, almost no cooling, on the other a very high price tag, necessity to cool to -100?, slow readout. So if one has a large field of view, a smaller DQE can be compensated with more detector surface. It is easy to compare both, even with a 25% fill factor (I mean detectors separated by their own spacing in the focal plane, i.e. 24mm away between each 24mm detectors), the efficiency is very high for CMOS (using a constant $ amount to build the detector). So far I have found black and white CMOS sensors up to 35mm films for roughly speaking 1000 euros a piece, but the question to this distinguished list is : Are there larger CMOS sensors around ? buttable ones ? (I don't believe, just asking). Alain -----Mensaje original----- De: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] En nombre de CMOSCCD at aol.com Enviado el: lundi 7 f?vrier 2005 16:22 Para: ccd-world at ctio.noao.edu Asunto: Re: CCD-world: CMOS sensors ***********************jj Cyril, Commercial CCD and CMOS cameras do a frame dark current subtraction (i.e., pixel by pixel despiking). . eliminating average dark current and fixed pattern noise. So the comparison below is not fair to Kodak. . who quotes unprocessed dark current. The Canon dark current is specified at 60 pA at 60 C.. . . which is about 5 pA/cm^2 at room temp. . . see table below. Nevertheless. . . dark current is unusually very good for a CMOS array (but not as good at 0.1 pA). My $300 Sony 5.2 M CCD camera shows similar dark current characteristics (see below. . a 30 second room temperature image). Both the Canon CMOS and Sony CCD are pinned diode pixels. Silicon is silicon. CMOS and CCD will coexist. . . even as dinosaurs.. . . :-) Jim *******************jj Dear Folks, For astronomy imaging I have noticed the oustanding performance of a Canon 20D digital camera. This camera has a 8 million pixel sensor (15x22mm), 6.5?m square pixels + bayer color filter. I decided to measure the readnoise and dark current of the embedded CMOS detector by using RAW files that the camera is able to output. It turns out the files to be proper for these measurments : raw 12 bits data, no Bayer to RGB reconstruction, consistent statistical behavior. I was amazed by the resulting figures, I could get, at ISO 800, 5e- of Readout noise (3.5e- at ISO 1600). Knowing that the whole 8 millions pixels sensor can be read out in less than 200 ms yes ! 0.2 sec ! (multiport readout ?), it looks to be very impressive figures. Also, regarding dark current measurements, I could measure 0.1 pA/cm2 at room temperature, which is 50 less than a Kodak CCD (known for they low dark current level). The sensor is able to withstand more than 300sec exposure without hexibiting large dark current (at room temperature, I must emphasize). Cosmetically, the device is perfect. Report can be seen here : http://astrosurf.com/cavadore/APN/carac/EOS20D/index.html I will not speak here about quantum efficiency of those devices, nor about the dynamic, nevertheless CANON seems to have achieved a very nice sensor. I could guess the readout conversion factor is large, helping to get low noise. I know this digital camera has also an embedded processor, I wonder how far this "guy" is changing the RAW data (if so). At least I have noticed an offset correction versus long exposures times. If someone has more informations about this detector... I would be interesting to know. If not, one can wonder if our CCDs used for Astronomy turns out to become dinosaurs those days ;-) Cyril Cavadore -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050215/6dc15147/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 16491 bytes Desc: image001.jpg Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050215/6dc15147/attachment.jpe -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: image/jpeg Size: 25780 bytes Desc: image002.jpg Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050215/6dc15147/attachment-0001.jpe From akana at cfht.hawaii.edu Wed Feb 16 16:26:20 2005 From: akana at cfht.hawaii.edu (akana at cfht.hawaii.edu) Date: Wed, 16 Feb 2005 09:26:20 -1000 Subject: CCD-world: Job Posting Message-ID: <42139E5C.7000102@cfht.hawaii.edu> Aloha ~ I would appreciate you posting this job in CCD-World please. I have copied the text within this email in an html version and also attached same in WORD format. If you have any questions please do not hesitate to contact me. Thank you for your assistance in this regard. _______________________ 1 Feb 2005 V A C A N C Y N O T I C E The Canada-France-Hawaii Telescope Corporation invites applications for the position of ELECTRONICS ENGINEER We operate a major optical telescope on Mauna Kea on the Big Island of Hawai'i and offer challenging work in an intellectually stimulating environment. We are at the forefront of wide-field imaging, developing a unique infrared mosaic camera and operating the world's largest CCD camera and state-of-the-art spectrographs. *DUTIES*: The successful candidate will be a member of a small, dynamic instrument team. He/she will be directly involved in the development, construction, integration, test and maintenance of state-of-the-art scientific instrument systems. *PROFESSIONAL QUALIFICATIONS*: B.S.E.E or equivalent degree in engineering/physical science with a minimum of five years experience in electronics engineering. Fundamental mathematical, analytical, and good communication skills and the ability to execute concurrent tasks are essential. Experience in the following areas is desirable - analog and digital circuits, feedback control, C and DSP programming, CCD and infrared detectors. *GENERAL QUALIFICATIONS*: The individual selected will undergo a physical examination to determine aptitude to carry out duties at the observatory's altitude of 4200 meters. The applicant must be able and willing to work irregular and night hours as occasionally required. Must be able to drive a 4-wheel-drive vehicle on an unpaved mountain road. Documentation to verify eligibility for employment in the United States will have to be furnished at the start of work. *COMPENSATION*: Salary will be dependent upon qualifications and experience. Comprehensive employer-paid benefit package and relocation assistance are available. Applications, which must include a detailed resume and three professional references, will be received until 4 Mar 2005 by Canada-France-Hawaii Telescope Corporation Attention: Electronics Engineer 65-1238 Mamalahoa Hwy. Kamuela, Hawaii 96743 -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050216/fda478f7/attachment.html -------------- next part -------------- A non-text attachment was scrubbed... Name: Vacancy_Notice.doc Type: application/msword Size: 31232 bytes Desc: not available Url : http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050216/fda478f7/attachment.doc From CMOSCCD at aol.com Thu Feb 17 22:16:46 2005 From: CMOSCCD at aol.com (CMOSCCD at aol.com) Date: Thu, 17 Feb 2005 20:16:46 EST Subject: CCD-world: CMOS SIZE Message-ID: <64.4f2dcaa1.2f469bfe@aol.com> **************jj Alain, CMOS arrays are normally limited in size to approximately 25 mm x 25 mm (although Tower Semi has produced a CMOS detector with an x dimension of 30 mm). Larger imagers required custom stitching. There are only a couple CMOS foundries that can (and willing) to do this. For example, we are involved with a 6k x 6k - 8um pixel backside illuminated CMOS imager which requires stitching (by Jazz Semiconductor in Newport Beach). I won't mention the cost involved for the 8-inch reticles based on 0.25/0.18um feature size :-). In theory, the design philosophy could produce a 8-inch wafer scale monolithic CMOS imager. Its scary to think about buttable CMOS. . .. because of the support circuitry on the sides. You're right. . readout speed for CMOS is amazing. The 6k sensor above is supposed to read at 30 frames/sec and deliver scientific (CCD?) performance. Jim ************************************jj > Are there larger CMOS sensors around ? buttable ones ? (I don?t believe, > just asking). Alain > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050217/eca4db91/attachment.html From rreiss at eso.org Tue Feb 22 08:02:09 2005 From: rreiss at eso.org (Roland Reiss) Date: Tue, 22 Feb 2005 12:02:09 +0100 Subject: CCD-world: Switching Power Supplies Message-ID: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> Hi CCD World, in order to cope with the power consumption of FIERA and to possibly reduce size and weight of it's power supply we have made experiments with switching power supplies. The present experimental design is built around Vicor second generation DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is made from FARM3 and MiniHAM modules to meet CE EMC rules (see www.vicor-power.com for details on the modules used). The differential noise measured at the outputs (+5, +/-15V, +30V) under load is pretty low (a few mV at 20MHz BW) but there is still significant noise visible as soon as a CCD is connected and read out. Also, grounding seems to be extremely critical. Small variations in the grounding scheme create big variations in noise. We think there is still a substantial amount of common mode noise involved but did not succeed yet to get rid of it. One of the difficulties are the high currents involved (20A @ 5V, 10A @ 15V). Are there any ideas or experience around how to reduce power supply CM noise? Kind regards Roland Reiss European Souther Observatory From cdm at ast.cam.ac.uk Tue Feb 22 10:37:20 2005 From: cdm at ast.cam.ac.uk (Craig Mackay) Date: Tue, 22 Feb 2005 13:37:20 +0000 Subject: CCD-world: Switching Power Supplies In-Reply-To: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> References: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> Message-ID: <421B3590.7010301@ast.cam.ac.uk> Dear Roland I have been experimenting recently with devices known as common mode chokes, components that I find a number of my colleagues do not know. I think they may have something to offer you with your power supply problems. Basically these devices are inductors with two windings that are essentially identical. Current that goes out through one winding and is returned via the other sees a very low impedance because the magnetic fields cancel out, and the net impedance is the difference between the two impedances. Common mode current, however, sees this as having an impedance equal to their sum rather than the difference. These devices are fairly widely available and are not expensive. They have application in quite a wide range of situations. Best wishes Craig Mackay. From atwood at phyas1.mps.ohio-state.edu Tue Feb 22 13:40:52 2005 From: atwood at phyas1.mps.ohio-state.edu (Bruce Atwood) Date: Tue, 22 Feb 2005 11:40:52 -0500 Subject: CCD-world: Switching Power Supplies In-Reply-To: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> References: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> Message-ID: <421B6094.4090701@ohstpy.mps.ohio-state.edu> Roland, I too have used common mode chokes (transformers) to break ground loops at high frequency. As Craig points out the DC current goes in one winding and out the other leaving the core with no net magnetization. This means that each side of the choke can have a large number of turns and still not have the core saturate from the DC current. I'll bet a beer in Sicily that uRAM modules already have common mode chokes. To me the more interesting comment is "Small variations in the grounding scheme create big variations in noise" . My experience agrees entirely in the sense that one can get switching supplies to work in a low noise system, by clever cut and try techniques, but the system must then be absolutely frozen. The smallest change can move you back to square one. I think the best solution is to extend the notion of Absolute System Schronisity (ASS, means all the digital activity is EXACTLY the same on every pixel, right down to the phasing of the clock in your transputers, should you be so inclined) to the power supply. If the system clock were also the switching supply clock one could imagine a configuration where the "noise" from the power supplies was the same on every pixel an therefore not really "noise" at all. When Roger and I talked about this some time ago (I'm sure Roger will post any minute now) I think our conclusion was that one would need a 10 MHz clock for a CCD system (and perhaps even higher for a mid IR system). To my knowledge there are small 2 MHz switchers but still no 10 MHz ones. Even if the basic clock were locked with the system clock most of these supplies regulate with PWM, which would still couple into the signal chain. bon chance Roland Reiss wrote: >Hi CCD World, > >in order to cope with the power consumption of FIERA and to possibly reduce >size and weight of it's power supply we have made experiments with switching >power supplies. > >The present experimental design is built around Vicor second generation >DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is >made from FARM3 and MiniHAM modules to meet CE EMC rules (see >www.vicor-power.com for details on the modules used). > >The differential noise measured at the outputs (+5, +/-15V, +30V) under load >is pretty low (a few mV at 20MHz BW) but there is still significant noise >visible as soon as a CCD is connected and read out. Also, grounding seems to >be extremely critical. Small variations in the grounding scheme create big >variations in noise. > >We think there is still a substantial amount of common mode noise involved >but did not succeed yet to get rid of it. One of the difficulties are the >high currents involved (20A @ 5V, 10A @ 15V). > >Are there any ideas or experience around how to reduce power supply CM >noise? > >Kind regards >Roland Reiss >European Souther Observatory > > > >-- CCD-world -- >CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu >Standard replies will go to the list; address personal replies manually. >For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world > > > -- -- Bruce Atwood Research Scientist Imaging Sciences Laboratory Astronomy Department The Ohio State University 140 W. 18th Ave. Columbus, Ohio 43210-1173 (614) 292 6279 (V) (614) 292 2928 (F) -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050222/8d104d5a/attachment.html From atwood at phyas1.mps.ohio-state.edu Tue Feb 22 13:52:36 2005 From: atwood at phyas1.mps.ohio-state.edu (Bruce Atwood) Date: Tue, 22 Feb 2005 11:52:36 -0500 Subject: CCD-world: (no subject) Message-ID: <421B6354.5070803@ohstpy.mps.ohio-state.edu> I am thinking about abandoning my long standing love affair with LN2. Cryotigers have caught my eye as a candidate for a fling, as I know they have caught the eyes of many of you. I think there have been posts WRT cryotigers in the past but I would appreciate an update on how they are doing. Questions that come to mind are. 1. How have the teething (startup) problems been.? 2. What is the coldest temperature in the Dewar (not the CCD temperature)? 3. Are you using charcoal or zeolite to maintain the vacuum, which, how cold is it.? 4. Are you using ORings to seal the Dewar? 5. What is the total length of ORing in you system (proportional to the rate that gas is entering the Dewar)? 6. What is the ration of ORing length to Dewar volume (proportional to the rate of pressure rise)? 7. How often do you pump? 8. How long have you kept your Dewar cold without pumping, how long do you NORMALLY keep it cold? 9. How long are your lines? Did you fabricate any part of the lines? 10. Do you have your own supply of gas and filling manifold? Your comments on any or all of these will be appreciated. cheers -- -- Bruce Atwood Research Scientist Imaging Sciences Laboratory Astronomy Department The Ohio State University 140 W. 18th Ave. Columbus, Ohio 43210-1173 (614) 292 6279 (V) (614) 292 2928 (F) From jean-luc.gach at oamp.fr Tue Feb 22 12:45:48 2005 From: jean-luc.gach at oamp.fr (Jean-Luc Gach) Date: Tue, 22 Feb 2005 16:45:48 +0100 Subject: CCD-world: Switching Power Supplies References: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> <421B6094.4090701@ohstpy.mps.ohio-state.edu> Message-ID: <00a601c518f5$9777c470$86d4ddc3@portablejlg> Hi Roland, Voltage noise is NOT the only source. You must measure current loops at high frequency which produce comon mode voltages through parasitic resistances (cables, solders, connexions). You should trace them with a high frequency current probe and block them with comon mode ferrites or open the loops. But there is no real method, you must try different solutions. A good method to eliminate the noise is to make the power supply synchronous with the CCD clock (for example with the pixel clock at low speed (up to 1MHz) or line clock at higher speed) regards ------------------------------------------------------------- Gach Jean-Luc Observatoire de Marseille 2, Place Le Verrier 13248 Marseille cedex 4 FRANCE phone : +33 (0)4 95 04 41 19 fax: +33 (0)4 91 62 11 90 http://www-obs.cnrs-mrs.fr/interferometrie/interferometrie.html Roland Reiss wrote: Hi CCD World, in order to cope with the power consumption of FIERA and to possibly reduce size and weight of it's power supply we have made experiments with switching power supplies. The present experimental design is built around Vicor second generation DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is made from FARM3 and MiniHAM modules to meet CE EMC rules (see www.vicor-power.com for details on the modules used). The differential noise measured at the outputs (+5, +/-15V, +30V) under load is pretty low (a few mV at 20MHz BW) but there is still significant noise visible as soon as a CCD is connected and read out. Also, grounding seems to be extremely critical. Small variations in the grounding scheme create big variations in noise. We think there is still a substantial amount of common mode noise involved but did not succeed yet to get rid of it. One of the difficulties are the high currents involved (20A @ 5V, 10A @ 15V). Are there any ideas or experience around how to reduce power supply CM noise? Kind regards Roland Reiss European Souther Observatory -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world -- -- Bruce Atwood Research Scientist Imaging Sciences Laboratory Astronomy Department The Ohio State University 140 W. 18th Ave. Columbus, Ohio 43210-1173 (614) 292 6279 (V) (614) 292 2928 (F) ------------------------------------------------------------------------------ -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050222/9b46f2ae/attachment.html From dbc at saao.ac.za Tue Feb 22 13:04:20 2005 From: dbc at saao.ac.za (Dave Carter) Date: Tue, 22 Feb 2005 18:04:20 +0200 (SAST) Subject: CCD-world: Ccryotigers In-Reply-To: <421B6354.5070803@ohstpy.mps.ohio-state.edu> Message-ID: Bruce we're just coming out of the learning curve, have had a system running on SALT almost without a break for about 15 months. My comments are in-line below. cheers Dave +-----------------------------------------------------+ Dave Carter Electronics Lab, S. A. Astronomical Observatory. e-mail : dbc at saao.ac.za Tel ( Nat. ) : 021-4470025 ( Int. ) : 27-21-4470025 Fax ( Nat. ) : 021-4473639 ( Int. ) : 27-21-4473639 +-----------------------------------------------------+ On Tue, 22 Feb 2005, Bruce Atwood wrote: > 1. How have the teething (startup) problems been.? Not too bad in restrospect. Initial interactions with Polycold were very difficult - just getting answers to questions etc > 2. What is the coldest temperature in the Dewar (not the CCD temperature)? ABout 110K - BUT (here's the first painful lesson) it varies with ambient temperature. For us the variation is around 11 degrees C summer to winter. So your CCD temperature servo has to have a wide enough dynamic range to cope with a heatsink that varies over a much wider range than an LN2 tank > 3. Are you using charcoal or zeolite to maintain the vacuum, which, how > cold is it.? charcoal, sits on the cryotiger cold end at ~110K. We went for charcoal to avoid having to remove zeolite from the cryostat to regenerate at the 300 degrees C or so that is necessary during maintenance pump cycles > 4. Are you using ORings to seal the Dewar? viton O-rings > 5. What is the total length of ORing in you system (proportional to the > rate that gas is entering the Dewar)? Not sure, at a guess about 1100mm > 6. What is the ration of ORing length to Dewar volume (proportional to > the rate of pressure rise)? Very roughly 2300 cm^3 - (12x12x16cm) - but most of the internal volume is occupied by the cryosiger cold end and focal plane assembly. > 7. How often do you pump? Two cycles: we use an ion pump tuned to grab noble gases (special option available from Varian) which runs about 4 hours a day, and maintains the vacuum at the 1 to 4E-7mbar range. Our design spec is to warm up & pump every six months, we have comfortably exceeded that as long as the ion pump runs every day. The ion pump head has a limited operational life, so we run only as long as necessary to get the pressure down to around 1E-7mbar (cryostat best pressure in cold winter conditions seems to be around 8E-8mbar) Useful device in that it gives a pressure reading while it's pumping. > 8. How long have you kept your Dewar cold without pumping, how long do > you NORMALLY keep it cold? answered above > 9. How long are your lines? Did you fabricate any part of the lines? Roughly 44meter x 2. the compressor sits at the base of the telescope & we have a combination of solid copper & stainless flex lines running up to the tracker/payload platform. We purchased all the lines from Polycold. Polycold will not guarantee the system to operate OK with such long lines, but in fact it seems to operate more efficiently. > 10. Do you have your own supply of gas and filling manifold? We purchased a recharge kit & gas cylinder which we've had to use a couple of times. The procedure in the book was complicated & didn't work so well, we devised our own which works for us. Thing to remember - if you transport the equipment by air you have to flush out the PT30 and recharge it at the destination. Done that also, went reasonably well. Other comments: 1. We had a couple of hose sections that leaked, had to be returned/replaced. finding the leaks was quite a rigmarole. 2. The low capacity of the cryotiger cold end (vs an LN2 tank) means you have to totally rethink your approach to the thermal path between the CCD cold plate & the cryotiger. The balance between getting the CCDs to a sensible natural (unheated) temperature and the time to cool down to operating temperature takes some fiddling to optimise. > > > Your comments on any or all of these will be appreciated. > > cheers > > From pjm at wairau.as.utexas.edu Tue Feb 22 14:14:21 2005 From: pjm at wairau.as.utexas.edu (pjm at wairau.as.utexas.edu) Date: Tue, 22 Feb 2005 11:14:21 -0600 (CST) Subject: CCD-world: Switching Power Supplies In-Reply-To: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> Message-ID: Hello Roland, Thanks for your interesting post. All the McDonald Observatory and HET CCD systems since 1999 have used Vicor DC-DC converters for their power supplies. I am extremely pleased with the performance of the power supplies, and can tell you about their implementation and performance as an example of a system with this type of power supply. The performance of our controller (which we call the McDonald Version 2 controller, or V2 controller) is very dependent of both the Power Supply Module and the power distribution architecture. For reference, the performance of the V2 controller with this power supply is: * 18-bit analog processing, with 80,000:1 linearity, and 200,000:1 signal-to-noise ratio at 100 kpix/s * noise statistics in data frames appear to the eye, and measure statistically to be purely Gaussian. We do not see interference patterns of any nature. * using E2V 2kx4k at 15um CCDs as a benchmark, we get about 2.65 electron readout noise at 100 kpix/s with 0.7 electrons per data unit and a full scale of about 180,000 electrons. The cross-talk between the two sides of a single CCD is about 1.5 electrons for a saturated image on the other half of the CCD. The Power Supply Module produces +36 V, +16.5 V, -16.5 V and +6 V. These are produced by first generation Vicor parts, including the VI-ARM-C12, the VI-J6V-CY (the 6V converter, for example), and VI-RAM-C1. The Power Supply Module is an 8 layer printed circuit board with a total parts cost of approximately US$1500. This power passes out of the Power Supply Supply module onto an 8-layer printed circuit board backplane module. All other modules (but one) plug into the backplane. Those modules include the DSP system controller module, the Analog Signal Processor Modules, the Clock Driver Module, and the Temperature Controller Module. All the other modules are either 8 or 6-layer printed circuit boards, with multiple ground planes. Each module has on-board, low drop out, linear regulators that make the power supplies needed within. That power is +33 V, +15V, -15 V, and +5 V. One module makes -5 V from the -15 V supply (it's a very low current supply). The one module of the V2 controller that does not plug into the backplane is our so-called Penthouse module. It mounts on the Cryostat, is a 12-layer printed circuit board, and has the pre-amplifiers and CCD bias supplies and gates. It is run on a cable from the backplane (I had to do it this way to be able to upgrade our old systems). In my view, the critical thing about this power architecture is the 'current supply and return path architecture' for every signal as a function of frequency. For all parts: * at high frequencies, the current comes from the ground plane at the part via the parts ground pin, and via a bypass (and possibly a ballast) capacitor from the ground plane to the power supply pin. * at low and intermediate frequencies, the current is drawn from the ground plane at the in-module regulator via the regulators ground pin, and via a large (1000 uF) ballast capacitor from the ground plane to the regulator input pin. * at very low frequencies, the power comes from outside the module, and specifically, from the Power Supply Module. It really just trickle charges the ballast capacitors of the in-module regulators. The overall goal is to keep the current path for every signal as localized and isolated as possible from all other circuits and modules. In conjunction with the above, every signal has uninterrupted ground on an immediately adjacent layer for the entire signal path from source to destination. The uninterrupted nature of the ground is very important. As a result of all this, the signal currents basically come from the ground at the part sourcing the signal, pass along the signal conductor to the destination part, return to ground at the destination part, and return to the source part through ground tightly coupled to the signal. The same applies regardless of the signal being the current in an Op Amp feedback loop, or a digital signal. This minimizes common impedance coupling between signals, minimizes both radiation of, and susceptibility to inductive coupling, and the ground layer shields the signal from electric fields. In summary, I've found that I get predictable, stable, and acceptable system performance with DC-DC converters and the above architecture. It's important to note that a single oscillator provides the timing for the entire controller with the exception of the DC-DC converters. They have there own variable oscillators. The VI-RAM parts do an excellent job of filtering that, and no doubt the in-module linear regulators help too. If you're interested in seeing the hardware, I have a photo gallery at: http://rotoiti.as.utexas.edu/CCDgroup Look at the link "McDonald V2 Controller". There is a photo of the Power Supply Module there, and you can see the regulators in the various modules. Regards, Phillip ________________________________________________________________ Phillip MacQueen pjm at wairau.as.utexas.edu The University of Texas at Austin Tel: (512) 471-1470 McDonald Observatory Fax: (512) 471-6016 1 University Station C1402 Office: RLM 17.334 Austin, TX 78712-0259, USA CCD lab: RLM 17.336 On Tue, 22 Feb 2005, Roland Reiss wrote: > Hi CCD World, > > in order to cope with the power consumption of FIERA and to possibly reduce > size and weight of it's power supply we have made experiments with switching > power supplies. > > The present experimental design is built around Vicor second generation > DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is > made from FARM3 and MiniHAM modules to meet CE EMC rules (see > www.vicor-power.com for details on the modules used). > > The differential noise measured at the outputs (+5, +/-15V, +30V) under load > is pretty low (a few mV at 20MHz BW) but there is still significant noise > visible as soon as a CCD is connected and read out. Also, grounding seems to > be extremely critical. Small variations in the grounding scheme create big > variations in noise. > > We think there is still a substantial amount of common mode noise involved > but did not succeed yet to get rid of it. One of the difficulties are the > high currents involved (20A @ 5V, 10A @ 15V). > > Are there any ideas or experience around how to reduce power supply CM > noise? > > Kind regards > Roland Reiss > European Souther Observatory > > > > -- CCD-world -- > CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu > Standard replies will go to the list; address personal replies manually. > For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world From robert.w.bumala at lmco.com Tue Feb 22 12:46:26 2005 From: robert.w.bumala at lmco.com (Bumala, Robert W) Date: Tue, 22 Feb 2005 07:46:26 -0800 Subject: CCD-world: Switching Power Supplies Message-ID: <573562C6FDA9564A8EEE66D899BC190B0C4B3546@emss01m10.us.lmco.com> Roland, The common mode noise is a standard problem of switching power supplies. It occurs because of the capacitive coupling from the primary to the secondary in the switcher. There is a really good app note from interpoint at: http://www.interpoint.com/datasheets/Output_Noise.pdf What I did for a low noise application was to get a toroid from Ferroxcube, and wound wire-wrap wire around it as the inductor. It has the added benefit of doing some differential mode filtering as well. Bob. -----Original Message----- From: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] On Behalf Of Roland Reiss Sent: Tuesday, February 22, 2005 3:02 AM To: CCD-world at ctio.noao.edu Subject: CCD-world: Switching Power Supplies Hi CCD World, in order to cope with the power consumption of FIERA and to possibly reduce size and weight of it's power supply we have made experiments with switching power supplies. The present experimental design is built around Vicor second generation DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is made from FARM3 and MiniHAM modules to meet CE EMC rules (see www.vicor-power.com for details on the modules used). The differential noise measured at the outputs (+5, +/-15V, +30V) under load is pretty low (a few mV at 20MHz BW) but there is still significant noise visible as soon as a CCD is connected and read out. Also, grounding seems to be extremely critical. Small variations in the grounding scheme create big variations in noise. We think there is still a substantial amount of common mode noise involved but did not succeed yet to get rid of it. One of the difficulties are the high currents involved (20A @ 5V, 10A @ 15V). Are there any ideas or experience around how to reduce power supply CM noise? Kind regards Roland Reiss European Souther Observatory -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world From dbl at tyrvos.caltech.edu Tue Feb 22 14:32:38 2005 From: dbl at tyrvos.caltech.edu (Lang, Daniel B.) Date: Tue, 22 Feb 2005 09:32:38 -0800 Subject: CCD-world: Switching Power Supplies Message-ID: <132CC123DCB88A4D8BA1ADAD250E264F29B1DF@olympos1.tyrvos.caltech.edu> If one can assure that the switching intervals are exactly the same on a frame by frame basis, subtracting the dark frames should also subtract most of the switching noise. Thus, you should make sure that the switching supply and the start of the frame readout are synchronized to the same clock which can be lower than the CCD readout frequency. The remaining problem is that at least one of the switching supply transitions is also a function of the input voltage and load current. Ideally, you could make sure that this occurrs during a parallel transfer or a pause between readouts. Very few converter manufacturers make externally sychrnonizable supplies. One is the BetaDyne powerwatt 50 series which is available with a shield: http://www.beta-dyne.com/products/dcdc50.html In addition to the common mode chokes, you should add conventional LC filtering to remove high frequency differential noise. Most linear regulators have poor high frequency ripple rejection. Daniel Lang -----Original Message----- From: Bruce Atwood [mailto:atwood at phyas1.mps.ohio-state.edu] Sent: Tuesday, February 22, 2005 8:41 AM To: Optical & SWIR imager development for professional astronomy Subject: Re: CCD-world: Switching Power Supplies Roland, I too have used common mode chokes (transformers) to break ground loops at high frequency. As Craig points out the DC current goes in one winding and out the other leaving the core with no net magnetization. This means that each side of the choke can have a large number of turns and still not have the core saturate from the DC current. I'll bet a beer in Sicily that uRAM modules already have common mode chokes. To me the more interesting comment is "Small variations in the grounding scheme create big variations in noise" . My experience agrees entirely in the sense that one can get switching supplies to work in a low noise system, by clever cut and try techniques, but the system must then be absolutely frozen. The smallest change can move you back to square one. I think the best solution is to extend the notion of Absolute System Schronisity (ASS, means all the digital activity is EXACTLY the same on every pixel, right down to the phasing of the clock in your transputers, should you be so inclined) to the power supply. If the system clock were also the switching supply clock one could imagine a configuration where the "noise" from the power supplies was the same on every pixel an therefore not really "noise" at all. When Roger and I talked about this some time ago (I'm sure Roger will post any minute now) I think our conclusion was that one would need a 10 MHz clock for a CCD system (and perhaps even higher for a mid IR system). To my knowledge there are small 2 MHz switchers but still no 10 MHz ones. Even if the basic clock were locked with the system clock most of these supplies regulate with PWM, which would still couple into the signal chain. bon chance Roland Reiss wrote: Hi CCD World, in order to cope with the power consumption of FIERA and to possibly reduce size and weight of it's power supply we have made experiments with switching power supplies. The present experimental design is built around Vicor second generation DC-DC modules and Ripple Attenuator Modules (?RAM). The mains front end is made from FARM3 and MiniHAM modules to meet CE EMC rules (see www.vicor-power.com for details on the modules used). The differential noise measured at the outputs (+5, +/-15V, +30V) under load is pretty low (a few mV at 20MHz BW) but there is still significant noise visible as soon as a CCD is connected and read out. Also, grounding seems to be extremely critical. Small variations in the grounding scheme create big variations in noise. We think there is still a substantial amount of common mode noise involved but did not succeed yet to get rid of it. One of the difficulties are the high currents involved (20A @ 5V, 10A @ 15V). Are there any ideas or experience around how to reduce power supply CM noise? Kind regards Roland Reiss European Souther Observatory -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world -- -- Bruce Atwood Research Scientist Imaging Sciences Laboratory Astronomy Department The Ohio State University 140 W. 18th Ave. Columbus, Ohio 43210-1173 (614) 292 6279 (V) (614) 292 2928 (F) -------------- next part -------------- An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050222/75da9473/attachment.html From rsmith at astro.caltech.edu Tue Feb 22 14:43:30 2005 From: rsmith at astro.caltech.edu (Roger Smith) Date: Tue, 22 Feb 2005 09:43:30 -0800 Subject: CCD-world: Switching Power Supplies References: <002a01c518cd$f43e19c0$501dab86@ads.eso.org> <421B6094.4090701@ohstpy.mps.ohio-state.edu> Message-ID: <421B6F42.E56416B9@astro.caltech.edu> An HTML attachment was scrubbed... URL: http://www.ctio.noao.edu/pipermail/ccd-world/attachments/20050222/aeaf7055/attachment.html From Les.Saddlemyer at nrc-cnrc.gc.ca Tue Feb 22 15:07:59 2005 From: Les.Saddlemyer at nrc-cnrc.gc.ca (Saddlemyer, Les) Date: Tue, 22 Feb 2005 13:07:59 -0500 Subject: CCD-world: (no subject) Message-ID: We've been using Cryotigers for about 4 years now, and very happy. Observers are even happier! The only nuisance is the coolant lines. Comments below. Cheers, Les Les Saddlemyer | Tel/T?l: (250) 363-0060 | Fax: (250) 363-0045 | Instrumentation Group | Groupe d'instrumentation Herzberg Institute of Astrophysics | Institut Herzberg d'astrophysique National Research Council Canada | 5071 West Saanich Road, Victoria BC V9E 2E7 Conseil national de recherches Canada | 5071, chemin West Saanich, Victoria (C.-B.) V9E 2E7 Government of Canada | Gouvernement du Canada Leslie.Saddlemyer at nrc-cnrc.gc.ca -----Original Message----- From: ccd-world-bounces at ctio.noao.edu [mailto:ccd-world-bounces at ctio.noao.edu] On Behalf Of Bruce Atwood Sent: Tuesday, February 22, 2005 8:53 AM To: CCD-world at ctio.noao.edu Subject: CCD-world: (no subject) I am thinking about abandoning my long standing love affair with LN2. Cryotigers have caught my eye as a candidate for a fling, as I know they have caught the eyes of many of you. I think there have been posts WRT cryotigers in the past but I would appreciate an update on how they are doing. Questions that come to mind are. 1. How have the teething (startup) problems been.? > Only with icing in the lines. Use of scrubbers have totally solved this problem. (symptom: > the system cooling to 0C and sitting there (+/- 1 degree) as, presumably, ice was clogging > and melting at the cold head. I'd not consider the system without a scrubber/drier. 2. What is the coldest temperature in the Dewar (not the CCD temperature)? 3. Are you using charcoal or zeolite to maintain the vacuum, which, how cold is it.? > We run down to -135C and -110C for each of our two CCDs (SITe 1024^2 and SITe2kx4k respectively). > Regulated with heaters, with plenty of headroom. We've switched to activated charcoal to avoid > the problem with the required re-activation temperature of zeolite. However, we use zeolite > to help initially dry the dewar. I found > http://www.astronomy.ohio-state.edu/~isl/PAPERS/Aadsorb.htm to be a very useful reference. 4. Are you using ORings to seal the Dewar? > Yes 5. What is the total length of ORing in you system (proportional to the rate that gas is entering the Dewar)? > Raw, we see about a 30mTorr rise/day in our ~1 litre dewars. There are two O-rings about 120mm > in diameter (and a couple of smaller ones). As these are the prime source of pressure rise > (from what I've read), I'd like a better solution. 6. What is the ration of ORing length to Dewar volume (proportional to the rate of pressure rise)? > Dewar is about 1 litre 7. How often do you pump? > Once we've dried the dewars, we only pump about once every 3-6 months. We monitor the heater > supply, and schedule a warm-up and pump once the headroom is dropping. If the dewar isn't opened, > we tend to get progressively longer periods between pumps. 8. How long have you kept your Dewar cold without pumping, how long do you NORMALLY keep it cold? > 3 - 6 months. 9. How long are your lines? Did you fabricate any part of the lines? 10. Do you have your own supply of gas and filling manifold? > Lines on one telescope are 30m (lines run to Newtonian mount) and 10m on the other (coude'). > We've not yet serviced/changed the coolant. We'll be looking into whether we should perform > a 'regular service' after 5 years of use. > Another note: on one telescope we're considering running two heads off of one compressor. > This we'll not try for about another 6 months or so. Your comments on any or all of these will be appreciated. cheers -- -- Bruce Atwood Research Scientist Imaging Sciences Laboratory Astronomy Department The Ohio State University 140 W. 18th Ave. Columbus, Ohio 43210-1173 (614) 292 6279 (V) (614) 292 2928 (F) -- CCD-world -- CCD-world is fully moderated. Send posts to CCD-world at ctio.noao.edu Standard replies will go to the list; address personal replies manually. For more information, please go to: http://www.ctio.noao.edu/mailman/listinfo/ccd-world From pjm at wairau.as.utexas.edu Tue Feb 22 15:08:17 2005 From: pjm at wairau.as.utexas.edu (pjm at wairau.as.utexas.edu) Date: Tue, 22 Feb 2005 12:08:17 -0600 (CST) Subject: CCD-world: (no subject) In-Reply-To: <421B6354.5070803@ohstpy.mps.ohio-state.edu> Message-ID: Hello Bruce, It's hard to give up on LN2, it's so nice in many ways, but not having to think about a CCD system for a year or so at a time has its advantages too! 1. How have the teething (startup) problems been? Oh yes. But that was in 1998 and 1999 when CryoTigers were a brand new product. They aren't much of a hassle at all now. 2. What is the coldest temperature in the Dewar (not the CCD temperature)? It's about -191 degrees C (88 K), basically liquid nitrogen temperature. That's the temperature of the CryoTiger cold end. It varies about two degrees with ambient t