Author: Tom Droege Date: 960704 Revision: #1 960704 Key Words: CCD, instrumentation
OK, everybody, somehow I did not get the word out. There is an elaborate circuit built into the camera to regulate the temperature. The first stage runs flat out, and will lower the CCD temperature about 30 C below the water temperature. The second stage is driven with a variable current and is controlled by a thermister. One enters the voltage you desire to appear on the thermister load resistor, and the servo works to hold it there. Some rough tests indicate about 100/1 temperature gain, so a 100 C water temperature change will give a 1 C CCD temperature change. It is probably better than this.
The program function key line shows a temperature setting. Here a voltage is entered that commands the temperature. If it is too high, > 0.5 or so, the second stage just turns off, as the circuit is not set up to heat, just cool. If you set the command temperature voltage too low, then both stages turn hard on, and one actually cools less than at some intermediate setting.
I recommend running with about 2 amps in each of the first stage coolers. To find this point, set the command voltage to 1 volt. This will turn off the second stage cooler. Now adjust the power supply voltage until the power supply draws 6 amps. This is now two amps per first stage cooler.
Now set the command voltage to zero. The second stage cooler will now turn on hard. By watching ITEC and having an ammeter on the whole mess, you can calibrate ITEC to the second stage current. I try to do this at the plant, but the current measuring circuit is not great.
Now starting at say .5 volt, gradually decrease the temperature command voltage until ITEC starts moving away from zero. You will see the CCD temperature start to decrase, and if you patiently plot temperature vs time you will see that it overshoots and settles with a damped transient. Taking the command voltage down in 0.1 volt steps, then 0.01 volt steps as one nears the desired point, you will be able to find a command setting that requires about 1 ampere of ITEC. This is a good point to operate. The current can increase a lot from here without going out of regulation. I get about a 40 C drop operating here. It can also decrease a lot to compensate for the camera cooling down during the nights exposure.
This will get you to -10 C using 30 C water. The water can warm up to 38 C or so before regulation will be lost. Note that the tendency is for the cooling water to heat up during the night, but the outside temperature tends to go down, so the two cancel to make the needed delta t to be pretty constant.
Note that -10 C is cool enough for the Kodak chip. The dominant noise is the sky brightness. If you happen to find a location where the sky is so dark that -10 is not coo enough, then you can use the refirgerator trick to use water just above the dew point and get cooler. I predict that no tass camera will find dark enough sky that dark current at -10 C is a problem.
A basic idea of this design is that relatively warm water cools the camera. The design is such that the camera body parts come to the cooling water temperature while the CCD is much colder. This means that there is no problem with dew, as the camera body is above the dew point.
So you do ***not*** want to use ice water. Just water that is above the dew point, just room temperature or above. This is the easiest kind of water to get. ;^)
On Wed, 3 Jul 96, richmond@astro.Princeton.EDU wrote: > > Since Nick asked for suggestions on keeping constant the >temperature of water used to the cool the TASS camera, >let me give some. > > 1. My experience with about 15-20 gallons of water sitting > in an open trashcan, in the middle of a room at ~ 25 degrees C, > is that running a TASS triplet causes the water temperature > to rise by 0.1-0.2 degrees C per hour. This is slow > enough that it won't make much difference in the images > produced. Naively, one expects the noise to be _very slightly_ > higher at the end of the night, due to the increasing water > temperature, but, on the other hand, the ambient air will > usually have dropped, which may cool the water back down.
You should be able to regulate to -10 C +/- 0.01 C rms with the design. Some day I will take the time to look at an actual run and compute the hourly variation and sigma to go with.
> > ---> running with 20 gallons in a bucket should be fine. > Filling the bucket with ice will help, but is > not necessary.
It will hurt if the camera gets below the dew point. Actually an ice water solution is much less stable than one without. You can see cold "lumps" of water circulating through such a system. My bet it that it will increase the noise more due to temperature variation than without.
Those with cameras can try this experiment. Put lens caps on the cameras and use the temperature command to set a temperature somewhere in the middle of the range. Say around one ampere ITEC. Note the pedestal. By clearing the CCD and then running for a full drift scan exposure, you can get some idea of the dark current component. By checking the noise and comparing to the measured dark current you can convince yourself that you know the number of dark electrons accumulated during a full drift scan.
Now change the temperature - say 5 C. Cark current should double or halve depending on the direction. In fact, that is what you will see if you clear the ccd as above. The unexpected thing you will also find is that the pedestal after clearing changes - a lot. I have not been able to find what causes this, I assume that it is something like a change in resistance of the reset switch. The data sheet says nothing. This is a very strong reason for holding constant temperature even if you are at a temprature where the dark current is negligible.
> > 2. For those who want to control the water temperature, and > cool it down (to make camera colder), I suggest buying > a little "cube" refrigerator. Get 50 to 100 feet > of tubing, and coil it all up inside the refrigerator. > Connect this coil to the tubing running to the camera > and pump, and voila! Cold water. Alternatively, one > might place a big container of water into the refrigerator, > and run the tubing through the container.
This is a nice trick to get a cheap chiller. Refrigerators almost never die, and old ones are easy go get for the effort of moving them.
> > I'm going to try that method myself this weekend. > > 3. For those with a floor with a drain, place a large bucket/ > trashcan under a cold water faucet. Make sure the bucket > sits next to a drain in the floor. Turn on the faucet to > fill bucket, then restrict the flow to a trickle. Let > the bucket overflow onto the floor. If you can deal with > the water bill, this should work well -- just experiment > to find the lowest flow rate that maintains a steady > temperature. Of course, this assumes the cold water > supply _does_ have a steady temperature itself ...
The problem with this is that often in the summer, the water mains will be below the dew point, and that will be a problem.
Sorry for this long diatribe, but I do want to get my design idea out.