Re: Rochester status report, Jan 2, 2002

[Tom Droege <tdroege@veriomail.com>]

I agree with Chris, don't try to parallel water systems.  A second pump 
might be a cheaper solution.  But then you have more heat going into the 
water if you get a submersible pump.  The CCD list has long discussions on 
pumps.  They use a 12 volt pump designed to pump bilge water.  I paid a lot 
for the last pump I bought because I was tired of fussing with cheap 
pumps.  You might try making the tubing size larger everywhere you 
can.  There is a lot of pressure drop in the long tubes.

I think there is enough capacity in the 12 volt power supply to run a 
pump.  But I would recommend an AC powered unit.  You could just run it 
straight off the tap if there is a drain to get rid of the water.  It 
depends on your feelings about the relative cost of things.  This also 
leaves a problem as to what to do to shut down so the heads do not freeze.

In any case, this is a problem that can probably be left for spring.  When 
it is freezing outside, one needs very little cooling.

I have been running at -17 C.  This is a compromise as to what I think I 
might be able to do during the summer.

I believe the small peak at -12 C is due to the famous Michael Richmond 
effect where the number of noise sites in a pixel is quantized to be 0, 1, 
2, ... bad sites.  To really see this effect run around +5 C.  There I can 
usually see 6 peaks.

Sorry about the V camera shutter.  They are a little fussy.  The idea is 
sound, it is just the implementation.  The problem is that when the 
shutters were glued up, they came out too thick, so the clearance in the 
track is tight.  A little work with a file will make the track fit 
better.  I use a graphite lubricant on the tracks.  The same stuff used to 
lubricate locks.  Note that the shutters are self adjusting and slip at 
both the closed and open position to constantly maintain their position.

I just came down from the roof where I attempted to "fix" the V camera here 
that had a sticky shutter.  You might check that the ring with the heater 
on it is not clamped too tight and is pushing on a track that might be 
sticking up beyond the camera head face.  This can cause the shutter to 
stick.  There is a key that will constantly run the shutter back and 
forth.  Test, MAServo, Test0 I recall.

Use of the old Stamp board instead of the nice new one that fixes several 
problems has been noted.  I think you really want to change the board 
before you waste time trying to understand the covered and overscan 
pixels.  The old scheme was wrong on several counts.  It is a wonder that 
it worked at all.

I guess I should write a maintenance note on the shutters.

Tom Droege

>   - the first frame read out after prolonged inactivity
>          (an hour or more?) usually has garbage in the first few hundred
>          rows -- even if the chip has been read once to clear it

Yep, this is a standard problem with CCDs.  Charge builds up on the 
detector and it takes a few clears to get it all.  I find that once things 
have been dark and cold for a while, a single clear does the job, even 
after several hours.

At 04:56 PM 1/3/02 -0500, you wrote:

>   I continue to prepare the Rochester Mark IV for regular operation.
>Most of the recent developments deal with the cooling system and
>dark frames.
>
>        1. drilled holes in the sides and floor of the shed,
>               so that coolant lines can enter the shed even when
>               the doors are fully closed
>
>        2. added 2 gallons of antifreeze to the coolant, and put
>               insulation around the tubing which is exposed to the
>               outside air (these spurred by discovering a
>               layer of ice 3 inches thick on the top of the
>               coolant a few days ago)
>
>        3. changed dessicant in the air tubing, so that moisture
>               won't build up in front of the CCDs
>
>        4. labelled the switch controlling the air pump with "ON" and "OFF"
>               (I discovered it was on when I thought it was off)
>
>        5. covered the (outdoot) electric outlet to which the coolant pump is
>               connected with an inverted bucket, so that I can
>               keep it plugged in at all times (even when it's raining)
>
>   I ran some tests of dark frames, and discovered a number of things:
>
>        - it takes about an hour for the temperature to stabilize after
>              the TE coolers are turned on
>
>        - the stdev of pixel values in a dark frame drops with temperature:
>
>                   T =  +9 C         stdev = 65 DN
>                   T = -12 C         stdev = 42 DN
>                   T = -20 C         stdev = 14 DN
>
>          The histogram of pixel values shows a small secondary peak
>          at T = -12 C, but it disappears at -20 C.  I judge -20 C
>          to be a reasonable CCD operating temperature.
>
>        - the first frame read out after prolonged inactivity
>          (an hour or more?) usually has garbage in the first few hundred
>          rows -- even if the chip has been read once to clear it
>
>        - the "F" (V-band) camera shutters don't quite meet when they
>          close.  This causes a light leak during the daytime.  At night,
>          with the shed roof closed, it doesn't appear to lead to any
>          perceptible signal.
>
>
>   I spent some time trying to understand the pixels near the edges of
>the frames -- the first and last few columns.  What I found was:
>(I use "active" to mean a pixel exposed to light in the usual way)
>
>        - cols 1, 3, 5  all have the same, fixed value (bias, I guess)
>
>        - col 2         is higher than 1, 3, 5
>                              its value doesn't change with exptime
>
>        - cols 0, 4     are higher than 1, 3, 5, but not as high as
>                           the "active" pixels
>                              their values DO increase with exptime
>
>        - col 6         is a tad lower than "active" pixels
>
>        - cols 7+       are "active" pixels
>
>   And, at the other side of the frame, the edge cols are
>
>        - col 2038      is similar to cols 0, 4
>                              value increases with exptime
>
>        - cols 2040, 2041, 2042  are similar to 1, 3, 5
>
>The behavior of these edge pixels may change with newer versions
>of the electronics and/or STAMP code.  I'm using an old version,
>so Tom will probably chastise me :-/
>
>   I was hoping to find a "dark" column, which would be shielded from
>light, but give a reasonable estimate of the "dark current" for
>active pixels.  I'm not confident that there is such a column;
>cols 0, 4 and 2038 look like the best bets.  One might want such
>a column if the temperature varied slowly throughout a night,
>with consequent variations in the dark current: dark frames taken
>during the day might not have the same temperature as images
>taken at night.  One could use the "dark column" to shift the
>daytime dark up or down as necessary.
>
>   I need to run more tests of the cooling system to see how
>stable the temperature remains over a 24-hour period and day/night
>cycle.  I also need to see if there's a close relationship between
>the values of, say, column 4 and the mean value of active pixels.
>More work...
>
>   Finally, I discovered that my current pump sends only about
>4 gallons per hour through the cameras.  That's not enough to
>keep them as cool as possible -- I discovered that one of them
>can be chilled to only -20 C, while the other keeps going to about
>-30 C.  Tom suggested that the downstream camera is getting
>warmer coolant, which can't carry off enough heat to cool it
>further.  He estimates that 10 gallons per hour is required to
>cool both cameras fully.
>
>   I went to the hardware store today and found that the cheapest
>pump was NOT an improvement over the current one :-(  Rats.
>
>   One of my colleagues here suggested I re-wire the coolant system
>so that it runs through the two cameras in parallel, not serial.
>That would be cheaper than buying a big (1/6 HP) new pump, but
>would take more work outside in the cold without gloves.  Sigh.
>I'll let you all know what happens.
>
>                                               Michael Richmond
>
>
>
>