Cryogen expiry; QE loss from contamination

[roger smith x294]

In reply to Mashal Ezra:	Cryogen Expiry Alarm
				--------------------
We liquify our own Nitrogen on Cerro Tololo, and our dewars have hold times
on the order of 18 hours.  All our telescopes are close together.  So we
have been content, thus far, refilling dewars by hand twice per day. 18 hour hold-
time gives us considerable safety margin for late refill, or vacuum degradation.
Our dewars have fill tubes which extend to the center of the tank so that there
is negligible spillage in any orientation.  Of course this means that only 
half the tank volume is useful, so the tank is simply made large enough to 
compensate.  The dewar is still quite compact and weighs about 7 Kg.  It is 
all Aluminium.  (More on this under the heading of contamination.)

No matter how big you cryogen flask sometime someone will forget to refill it
on time or the vacuum will go bad.  The next line of defence is an alarm.

We use a temperature sensor (1N914 diode, forward biased at 10 uA) to detect
cryogen expiry.  There is a loud beeper built into the CCD controller, and a
new telemetry system will soon provide remote monitoring.  We are tossing around
ideas like having a small graph showing temperature over the last day for each
detector in use, displayed on a workstation in a central office, plus audible 
warning from the workstation there, the observer's workstation as well as the
the controller itself.  By monitoring the correct points one can determine the
time since last fill, hold time of each dewar, etc.

What is the correct point to measure?  The obvious one is the Nitrogen flask.
One can do this, but the rate of change of temperature is lowish, and to get 
prompt warning one has to set the alarm threshold fairly close to the operating
temperature.  One must avoid false alarms at all cost since this will lead to 
the alarm being disabled or ignored.  Component tolerances and the variations 
in temperature of the sensor with dewar orientation force one to increase the
safety margin and thus delay the alarm.  One wants to sound the alarm and get
a response before the detector temperature servo loses control.  My measurements
showed that it would be difficult to do this soon enough with a 
sensor mounted on the flask.

Instead, I put the alarm sensor on the neck tube, about 20% of the way from the
warm end.  When there is no gas flow the temperature profile is approximately
linear (though to be honest I haven't measured it).  However when the vent gas 
is flowing the profile is much flatter, rising to room temperature rapidly near
the exit.  Thus a sensor placed near-ish the exit sees an abrupt temperature rise
when gas flow stops, ie: when the last of the liquid boils away.

Of course the neck temperature varies wildly during nitrogen refill, and it drops
below normal immediately after a fill, due to the higher flow rate.  In spite of all
these variations, I can set the threshold to catch the cryogen expiry within
about 5 minutes, since the temperature rise is so large and fast.
The detector servo will hold its setpoint for about 30 minutes after the alarm 
sounds, and sometimes more.  Thus there is a reasonable possibility of recovery
without loss of observing time, and a very good chance of recovery before contaminants
cryopump onto the detector surface.

---------

In reply to Hugo Schwarz:

All CCD dewars that I know of use getters.

We have seen contaminants condense on the CCD while kept cold.  This is quite
distinct to (an less common than) the cryopumping mechanism during warm-up which
I described last week.  I am curious as to why the getter doesn't soak up the material.
I am even more curious about where the contaminants come from.  These are well pumped
and aparently clean dewars with no leaks.

Presumably, the only surfaces which can outgas are the warm ones.  Our dewar interiors
are polished Aluminium, and while they are explicitly not anodized, they will of course
be oxidised.  Anodized surface scan adsorb/absorb significant amounts of water which is
hard to pump out and will later outgas.   Perhaps we should only open dewars in
dehumidified air!

Does anyone reading this coat treat internal surfaces in any way  to reduce this effect?
Has anyone switched to alternate materials because of it?


QE changes when contaminated:
-----------------------------

What happens to the AR coating performance?...  Though we haven't done a quantitative
study of the effect of contaminants condensing on the CCD, I am told that the loss of QE
is not serious for thin Tektronix CCDs.  Indeed, improvements in QE have been seen at
some wavelengths, presumably when the contaminating film acts as an AR coating!

The serious problem with slowly accumulating thin films is that flat fields become
unstable over time, by as much as 5%.

>From what I read about UV flooding here recently, I would expect UV flooded chips to lose
significant QE in the presence of water vapor.  Is this still true when the detector is cold?




Roger Smith,   CTIO