Author: Tom Droege Date: 971203 Revision: #1 971203 Key Words: instrumentation, documentation
Editor's note: This document describes the use of the "TASSBCD6.EXE" program for diagnostic testing of the camera. Source code is available for those with QB 4.5 or above. For production the units are running TM3GET11 or RT Linux client/server program.
Do the usual things to check if the camera has survived the trip. Look at the camera tube ends. The window should be pushed all the way down against it's stop. Possibly air travel will cause the window to pop out. Just push it back into place.
On the bench, screw the three lenses onto the camera tube ends. Use the #6-32 cap screws provided. You will need a 7/64" allen wrench. I suggest buying a long "T" handle one.
Plug the DB-25 cable into the camera. Plug the Control card into a PC. Plug the DB-25 cable into the Control Card. Nothing special about the order here.
Leave the cover off the PC so that you can observe the LED on the control card. This LED lights for about .1 second every time an I/O command is processed by the control card. This is a big help in checking initial operation. It is also helpful when checking the drift scan timing (if your computer does not have a speaker).
Get the program tassbcd6.exe from the TASS software archive. If you have QuickBasic 7.1 (possibly 7.0 and 4.5 will also work, but QBasic that comes with Windows will not work) then you can run in the environment. Get the files tassbcd6.bas and tass1d.bas.
Note that the camera is powered over the DB-25 cable, so no special power supply is needed at this point.
Turn on the PC and start the program tassbcd6.
Make sure the lens caps are on the lenses, and that the tape that covers a lens hole that admits light has not come off.
Function key 5 will now start data taking. Answer any proper file name of 6 characters or less for the file, and 4000 for the number of lines. (Make sure you have 21 Mb or so available on your disk.)
The upper left corner of the screen will count to 1000 as the program reads out 1000 lines to clear the accumulated charge from the CCD. It will now start reading the three cameras. The West facing camera is at the top, the South facing in the middle and the East facing camera is at the bottom.
You should see vertical lines as hot pixels stick up above the mean value. The program takes the mean of a few pixels in the middle of the line in order to determine how to make the plot.
If you can see vertical lines at this point, the everything has probably survived the trip.
Several items are displayed at the top of the screen. When you start the program, you are asked for a number of lines to take. You can also use the Block function key to specify the number of lines in a block. Each block is output to a separate file. Each camera is also output to a separate file. For example, if you specify 4000 for the Block length (the default) and 8000 for the number of lines, then 6 output files will be produced. These will be named:
name11.dat name21.dat name31.dat name12.dat name22.dat name32.dat
Where the first number in the file name is the camera number and the second is the block number.
The first item displayed at the screen top is the current block being taken and then the current line in that block.
Next is displayed the water temperature, then the ccd temperature, then the value of pixel 250 of camera two, then the second stage thermoelectric cooler current.
The function switches allow the following operations:
Look for vertical lines as above
If there are vertical lines, then a good check is to take off a lens cap. The affected lens should go blank as the CCD saturates. If done on camera two, then the 2Pix250 value should change from a number like -24000 to 32767, indicating saturation.
You need to set the clock to a value that will give a 0.916 second interval between ticks. If you have a speaker on your PC, it will beep at each line. There is no way to delete this from the program. It is designed in as incentive for one of you to write some good code. You can use a stop watch to count beeps. I find I have to count for 100 seconds or so to get a sufficiently accurate setting. A starting number for the VCO value is written on the Control card on the power supply. Time this value and a few hundred counts on either side to get enough information to make a count vs rate plot. Then you can pick a suitable operating point. Note that you will want to be able to set a bad clock value for adjusting the rotation. So make a note of where you will want to be for about a 4% error. Sorry, but the VCO has a temperature drift. You will have to let everything warm up for a half hour or so to find a good value.
Each camera head contains a two stage thermoelectric cooler. One of them (normally the center camera) contains a pair of thermisters. One of these is glued to the water cooled plate. A second one is glued to the CCD. Sorry, but the location is far from optimum. In particular, there is no good place on the CCD to locate the thermister. This means that the CCD temperature is probably measured high, as is the water temperature. A bad choice of value allows a fair amount of self heating.
The first stage of the thermoelectric cooler is driven directly from the voltage applied to the Red(+) and Black(-) binding posts on the back of the camera. With 12 volts applied at the camera terminals, the camera will cool to about 30 C below the water temperature. Note that if the power supply is any significant distance away from the camera, there will be significant drop in the wires. You will then need somewhat more than 12 volts at the power supply. Long experience with physicists (some even with Nobel prizes) indicates to me that a PhD is no assurance that the drop in such power wires will be computed correctly. Sometimes they get it half right, only failing to realize that the return wire has a drop too. I use a heavy duty extension cord with the ends cut off. Cheapest wire around. (Unless you have a stock room full of such stuff.) I need to run my power supply at about 15 volts with a 25 foot wire.
The second stage of the thermoelectric cooler is driven from a servo. It looks at the voltage generated on a thermister load resistor, and drives current through the second stage to match a command voltage. The current supplied is shown on the top line as ITEC and is roughly in amperes.
Before powering the cooling circuits, you will need to supply cooling water. I use a "Little Giant" water pump in a large garbage can. See and earlier discussion, but the general idea of the cooling design is to always have the cooling water above the dew point. This way you will not have dew problems with the camera.
Unsnap the quick water disconnect (over a towel - the camera still has anti-freeze solution in it from production test). You can now use the spare quick disconnect sent to connect tubing to the water pump. A high water flow will not help the cooling much. About 1 foot per second is enough. Just watch a bubble in the tubing to get an idea if there is enough flow.
While running the program, turn on the power supply and adjust for about 12 volts at the camera terminals. You should see the CCD temperature fall. If it increases, then the polarity is probably wrong. Nothing should be harmed if the polarity is wrong, but it voids the guarantee. ;^)
The power supply should draw about 6 amps at this point. Just to make sure that the second stage cooler is not working, use F2 to set the temperature command to 1.0 volt.
The CCD temperature should be observed to be decreasing, and the water temperature will increase somewhat. A 30 C indication is not unusual for the water temperature with water that is closer to 20 C. It is just a question of placement of the thermocouples. In ten minutes or so the CCD temperature should get 30 C or so below the water temperature.
Now use F2 to command a lower temperature. A value of 0.5 is good for a start, but may produce no effect. Watch ITEC on the top line of the display. Slowly decrease the command voltage until this value starts to increase. About a change of 0.05 every minute. Depending on how fast the command is changed, the current will probably overshoot and go to full scale of the secondary cooler current, about 2.5 amperes.
I find that a 0.3 volt command gives a regulated temperature of -10 C with about a 1 ampere current in the second stage cooler. This is a good place to run as -10 is cool enough that the sky brightness is the dominant noise term. When operating correctly, the temperature servo will hold constant to 0.01 C RMS. Note that it takes some time of steady state operation to become this stable.
If the display is observed while the camera is cooling, the vertical lines from hot pixels will start to disappear. They disappear slower than might be expected from the temperature since the drift scan mode of read out means that the full CCD is not cleared for about 8 minutes after the cooling starts. To speed up the process you can use the F1ClrCCD function key to run the clear routine.
If all is working correctly, then when the CCDs are cooled to -10 C there should be a pretty blank screen. There may be a hot pixel. That is life.
If the above tests give the expected results, then you are ready to look at stars. Order a clear sky.
The serial number on the camera body, and the camera tube letters should point north. This means that the flat face of the camera body should point at the north star. A fixture will be required to angle the camera according to the latitude of the location. Note that there are two screw holes on each end of the camera body intended for mounting. I just drill two holes into plates at the appropriate angle and mound the plates to a base with angle stock. When mounted properly, the cameras should be pointed at the equator, and should be pointing 15 degrees west of south (#1), south (#2), and 15 degrees east of south (#3).
If you run the camera at dusk, be aware that the cameras will saturate. This saturates the amplifiers and can give funny results. The 2Pix250 reading may show positive full scale, negative full scale or something in between while it is still light. This is because the amplifiers can saturate at different points, and the difference can be anything. Also be aware that the multiplex operation works on a chip with a 5 volt range being driven by amplifiers with a 15 volt range. This can result in other funny results until the signals come into range. Sometimes, for example, saturated CCD signals can cause the ITEC current reading to be in error. All recover when it gets dark.
Eventually it should get dark. Always later than you think as this is a sensitive device. But eventually you may start to see stars if you just keep running the program. There are three adjustments to make on each of the three cameras. It will pay to keep detailed notes to minimize the number of trips out to the camera. One must focus each camera, rotate it to line up the CCD so that stars travel directly e-w down the columns, and the time must be adjusted so that the shift rate matches the drift rate of a star down the CCD column.
Set the lenses so that 2.5 meter mark is at the fourth white line to the left of the red focus mark. The 3 meter mark is now to the right of the lines to the right of the red focus mark.
Loosen the two screws at the base of each camera head. The nuts behind these screws are just glued to wires so any hard tightening will break them loose and you will get to figure out how to do this better. Note that you only have to tightening these once! Now use the knobs to rotate the lenses so that the square base is roughly lined up east-west. Things are pinned internally, but it is not perfect.
Set the VCO timing to be about 5% off the sideral rate. Say about 0.95 second period instead of 0.916 second.
I hope one of you will write a focus program. Here is what I do. I set the Block length to 50 with F4. When the program starts I set the number of lines to 50. This will then cause the program to run for 512 lines at the VCO rate to build up the star images by the drift scan process. It will then clear the screen, and take 50 more lines of data. This is a good number to do a star count. So I just count stars (by some definition of an obvious star like it shows red or brighter). Then I change the focus and do it again. Eventually I get enough information to plot star count vs lens focus position. Of course if the milky way happens to come by in the middle of a focus run it poses some problems in interpretation. So programmers get busy with a better way.
When the focus is good enough to see stars, don't fuss to much to get a perfect focus, just get star tracks. After the full scan of 512 lines, the stars should be streaks if the VCO has been adjusted to be in error by 5%. Note that 5% of 512 pixels is 25 pixels, so the stars should be nice long streaks. They may or may not be vertical. If not perfectly vertical, try rotating the knob a turn. Note that the anti- backlash scheme is not perfect. The base is greased. You should observe that the rotation actually tracks the knob movement and help it along with your hand if it appears to be sticky. After a trial rotation, do another run and observe what has happened to the streaks. If they are more vertical the direction was correct. Remember you just have to do this once! ;^)
Once the rotation is lined up correctly, then gently tighten the screws at each corner of the camera tube. There is just a nut glued to a wire on the other side, so if you break it loose you will have to figure out how to tighten it.
Once the rotation is correct, then attempt to change the streaks to points by adjusting the VCO. Best to go only part way at first to establish the direction. 0.916 seconds does seem right to me and matches the 135mm focal length and the 15 micron pixel. But creep up on the right value because there is nothing to tell you the sigh has changed if you go past the correct position.
Once the rotation and drift timing are correct, a final focus can be fussed with. I find it to be not too critical. But perhaps it is and I just don't know yet how to measure it.
Back to Index of Technical Notes
Back to TASS home page