Lead wire selection (was RE: CCD-world: Socket for the THOMSON TH7899 CCD)

[Tom Droege]

The following was posted to CCD-world:

George,

I do not fault anyone who actually goes off and does a computation.  You
have done that, and while you come to a slightly different conclusion than
I, you have thought about it and now we are down to checking computations,
and that can be discussed.  It is amazing how people will argue over this
without doing any look up work in the literature.

Looks like I did not find the same numbers that you did for
Phosphor-Bronze.  It is probably not a practical material to use for wire??
  I recall Gold being one of the better materials by your (and my) figure
of merit.  I looked at a lot of elements once I started it was a hunt to
find something better than copper.  I just sat there with the Goodfellow
catalog and worked down the list of elements.  Gold is actually a
semi-practical solution as you might ball bond connections.  

I much agree with your plan of attack.  I would add solderability (or
crimpability) to the list.  

Clearly CCD-World is a more thoughtful list than the CCD list where this
discussion generated lots of "I used xx wire because Sam used it" sort of
discussions. 

Tom Droege 

At 07:24 PM 12/13/99 -0800, you wrote:
>Thanks, Tom, for setting me off to think about the tradeoff between thermal 
>and electrical conductance, and also the link to KKA Assoc. web site.
>
>I compared 5 materials, looking only at electrical resistivity and thermal 
>conductivity. If your application involves magnetics or radiation it may be 
>too simplistic. For lead wires to a cryostat, you would like to have LOW 
>thermal conductivity and LOW electrical resistivity. For a figure of merit, 
>I calculated the inverse product of the two properties, i.e., 
>1/(th.conductivity*el.resistivity). This way, higher is better with both in 
>the denominator, i.e., maximum electrical conductivity per watt of heat 
>gain.
>
> 
>
>
>Constantan is necessary for many thermocouples, hence it is often 
>conveniently lying around waiting to be used. As a cryostat lead material, 
>it is about the same as Nichrome in terms of heat gain per ohm. So, current 
>density must be kept extremely low to avoid self-heating.
>
>The optimum process for wire selection might be like this:
>1. Set a minimum wire gauge for practical handling (subjective).
>2. Set a maximum of either voltage drop or heat gain (may be arbitrary 
>parts of larger budgets).
>3. Find the material that satisfies both 1 & 2, in the following order of 
>precedence:
>      Phosphor-Bronze > Copper > Manganin > Constantan > Nichrome.
>
>For most wires, you also need thin insulation so they can be effectively 
>heat sunk. If the leads are inside a circuit board, you can tailor the 
>conductor cross-section to whatever you need. Note that the overall thermal 
>conductivity of a circuit board will be a strong function of component and 
>trace density.
>
>Some of these wire materials may also be difficult to solder. Maybe someone 
>else on this list can address that issue.
>
>-George Tylinski  Portland, OR
>
>
>-----Original Message-----
>From:	Tom Droege [SMTP:droege at wwa.com]
>Sent:	Friday, December 03, 1999 10:58 AM
>To:	CCD-world at astro.ku.dk
>Subject:	Re: CCD-world: Socket for the THOMSON TH7899 CCD
>
> [snip]
>
>This method of mounting CCDs has been the subject of controversial debate
>between myself and those advocating constantan wire and the like to make
>connections to the CCD.  If you do the computations, you will find that for
>the same resistance, copper wire connections are superior to materials like
>constantan.  i.e. the ratio of electrical conductivity to thermal
>conductivity is better for copper.  What constantan allows is the use of a
>larger wire size (which is then more rugged and has lower impedance) to
>make the socket connection.  Putting the wire on a printed circuit
>substrate solves the "rugged" problem and allows using strip line
>techniques which are difficult to implement with wire.  With strip lines
>and thin printed circuit boards one can make low impedance connections and
>put the clock line edge shaping where you want it without ringing problems
>from the interconnecting leads.
>
>I hope this does not start a fist fight.  In other groups I have found an
>"attachment" to constantan for the connections.  I have been unable to
>persuade anyone to make computations.  It is not a clear cut decision.
>There are trade offs.  This group should be better at evaluating such
>considerations.  I am also taking the liberty of answering more that what
>is asked.  One selecting a socket should also be thinking about the thermal
>problems.  ;^)
>
>Tom Droege
>[snip]
>
- -- CCD-world -- --
CCD-world is fully moderated. Send posts to CCD-world at astro.ku.dk
Standard replies will go to the list; address personal replies manually.
For more information, please go to:  http://www.not.iac.es/CCD-world/