Correspondence between Michael Richmond and Tom Droege
Date: Fri, 26 Jan 2001 To: Michael Richmond From: Tom Droege
Rah! That you are seeing something. OK, I started writing up step by step checks, finally I had an Ah! Hah! So search for this. Please make up a new section for the home page. "Mark IV Troubleshooting". You can put this whole message there as "TS-2". I will send you "TS-1" Which I just wrote to Arne. This way, I will only have to write up these things a few times.
Looks like you have a pretty slow computer. I think you want at least 200 MHz. Make sure that there are no other jobs running. At least if you are running a DOS window under Windows.
> Here's the latest news: I am able to read out the two Mark IV >cameras, using the "Take" subroutine. I was having some problems, >so I decided to comment out of the code the calls to the subroutines >which turn the RA drive on/off. What happens now is
I am not sure what you commented out, so I would suggest instead that you unplug the RA drive motor, and put something between the limit switch on the RA drive so that it thinks it is always at the home position. Now you can run as long as you want and can be sure that you will not overdrive the RA.
> - I request a dark exposure, time 0 sec > - "start clearing scan" > counter on screen goes to 61 seconds
It should go to 46 seconds. 61 seconds indicates that some other process is no longer running in the background on your computer. When you start the program you will see a number flash at the top of the screen. This is the number of iterations in a timing loop used to time things. This is taken with nothing else running. (I hope). However if the computer slows down because Windows has decided to monitor what it can charge you or sell you, then the timing loop no longer works. I usually see smaller numbers than 46 indicating the computer has slowed down. You get a larger number which means the computer is running faster. But note it is a timing loop, and is sensitive to how fast your computer is. Look at the code in the subroutine Timex. You may want to change the number of counts in the loop. You might check the time taken to count from 1 to 61 with a stop watch. I bet it will be close to 46 seconds.
> - "clear complete, start exposure" > - "exposure complete, start read-out scan" > counter on screen goes to 61 seconds > > At this point, two lines are printed to the screen, up at the top: >one gives the date and time (I think this is a copy of the the logfile >entry), and the other gives the temperatures of camera 0, camera 1, >and the water.
Ah! Hah!
When the read out is complete, a "Block Done" is sent over the cable between the Mark IV and the memory card. The screen counter is testing for this. My code used to just keep counting if it did not see a block done. Arne added code which stops the count after 60 (61) tries. So it looks like the problem is that you are never getting a block done at the memory board. This means (probably) that the cable between the Mark IV ADC board and the memory board is open somewhere. Make sure the memory board is plugged in firmly, make sure the cable is plugged into the ADC board, and make sure the intermediate connector is mated.
> Then, I wait for about two minutes, and eventually a bunch >of lines are printed to the screen which give some diagnostic pixel >info from the middle of one chip: > > 542 255 255 -1 > 543 255 255 -1 > 544 255 255 -1 > etc.
OK, this shows that all the chips pixels are reading out all 1s. The CCDs are set to a bias level of around -25000 counts. Normal numbers are (I just ran my camera and got):
542 158 180 -24908 543 158 132 -24956 544 158 163 --24955
> After all these lines appear, I have to wait about another two >minutes for the readout to finish. Then, I finally hear the >triumphant music :-) > > Here are the things that I'm puzzling over: > > 1. the RA drive is _still_ turned on, despite my attempts > to comment out all the RA stuff in the "Take" routine. > This is no big deal.
I will worry about this later. You probably missed something.
> 2. it takes about 5 minutes to save the exposures to disk. > This is way too long, right? I don't remember waiting > this long in your workshop.
I just ran my 500 MHz K6 computer with 64Mb of memory.
Clear time 46 seconds Exposure time 0 seconds Read Out Time 46 seconds Rewind Time Not Measured (10-20 Seconds) Disk Save Time 63 seconds
> 3. almost all the pixel values, in images from both cameras, > are saturated (i.e. each byte is 255). I have taken > a 0-sec dark with the lights off in the room, and with > black velvet over the camera heads. Still the chip > appears saturated. I'm not running the water or the > chiller, but that shouldn't be necessary to get some > non-saturated values at room temp, right?
My best guesses are:
Power supply problem. Cable Problem
First check that all the power supply voltages make sense when turning on. Just read the table and see that it makes sense. Then check the voltages at the marked points on the ADC board. This is the one with the two cables going to the cameras and the long cable going to the Memory board. You will see points marked on the bus. The + 20 should read between +20 and +24. Others should read close to the value marked. The +5 will read a little low, say 4.8 and the -5 will read a little high - say -5.1.
If this looks OK, then you probably have a cable problem. First check that the green light on on the memory board blinks each time a number ticks on the screen. This light blinks for each I/O operation to the board. Next the green light should glow pretty steady as you are reading to disk. Your disk light should also blink as data is transferred to it. Go To "Ah! Hah!"
> The final puzzle may simply be a consequence of item 3 -- the pixel >values may not contain any sensible data. > > 4. when I try to convert the raw images to FITS, using > 2037 rows by 2043 cols (and there are the exact number > of pixels for this combination), I don't see a > good-looking pattern; that is, I don't see the very few > non-saturated pixels falling into nice patterns by > row and col. > > I suspect that item 4 will disappear once I figure out how to solve >item 3. > > Any hints? Do I need to clear the chips 5 or 6 times before I read >them out? > > Michael >
Date: Sat, 27 Jan 2001 To: Tom Droege From: Michael Richmond
After bringing my own (fast) computer from home to school, swapping it for the slow computer, and trying to read out the camera, I found the same results as yesterday. No luck.
Then, I started following your list of items to check. I tried to check the voltages -- seemed okay. As I was examining the ADC board, I noticed that there were two connectors with cables attached, and one connector without a cable attached. About twenty minutes later, it occurred to me that I might want to attach a cable to it :-/
Success.
This exercise did teach me one other thing, though: the old computer I have been using -- a 50 MHz 486 with 16 MB RAM -- can control the camera, read out the chips, and save the data ... but it takes about 5 minutes to write the data to disk. My new computer takes only about 1 minute. Obviously, I'll have to find a computer faster than the old, old 486.
Thanks for the help!