I camera fixed?

[Tom Droege <droege@wwa.com>]

It appears that I have found the problem with the I camera.  I am now
getting nice flat (dark) images on both cameras.  To explain the "fix" one
has to go back to last September/October when I was trying to get ARNE
ready to ship out to AZ.  

Warning!  The following will be as boring as Andrew's spline fits (to me)
to someone not interested in electronics.  

There was a noise problem due (I believe) to the signal to noise on the ADC
board.  To reduce the noise, I put more gain in the camera head and less in
the following stages.  Under certain circumstances, the common mode signal
from the camera head could overload the differential amplifier that
received the signal from the camera head.  I have been working on the right
combination of grounds to minimize this problem.  Eventually I changed the
amplifies (they were very fast amplifiers which ran on +/- 5 volts) to a
somewhat slower amplifier which ran on +/-15 volts.  These are then more
tolerant of the common mode voltage.  This required a lot of cuts and
pastes on the PC board.  Eventually I may redo this board.  This problem
really relates to the failure of the clamp circuit in the camera head to
work.  This would have removed the common mode from the camera signal.
Well, it worked, but it was a little slow due to the capacity of the switch
used.  I made use of a left over channel in the switches used to generate
the clock voltages.  So when I redid the camera pc board, I used the switch
channel instead to allow cutting the power of the output amplifier during
data collection to reduce the corner hot spot.  (Not yet implemented - prom
code).

During all this, I was experimenting with different values of coupling
capacitor between the pull down resistor that connects to the amplifier in
the CCD and the input of the gain of +3 buffer amplifier in the camera
head.  The normal value is 0.22 uf and 1 meg, giving a time constant of .22
seconds.  This is just a few lines at readout speed (20 ms per line).

Now the turn on transient that I have been observing has a time constant of
order 25 seconds.  This is a factor of 100 away from what I thought was in
the interstage coupling.

It turns out, that during last September/October I was experimenting with
this capacitor size.  I tried smaller values (could cause a slight loss of
gain) and much larger values.  

In one case I tried a 47 uf capacitor.  In another I tried a 1 nf.  

It turns out, that my Xerox technicians Dan and Mike got hold of the camera
head with the 47 uf capacitor as a sample.  They made all the new heads
with the 47 uf cap in them.  Meanwhile, the V channel I have been testing
ended up with the 1 nf cap.  

Eventually I switched the camera heads and the electronics and found that
the trouble went with the camera.  None the less, it took about 5 more
openings of the camera before I noticed that the V camera had the 1 nf
instead of the expected .22 nf.  This suggested that a fix would be to
replace the .22 uf with a 1 nf to fix the I camera.  I then took the I
camera apart (yet one more time) and found the 47 uf.  This caused me to
look at the other cameras and I found there was 47 us caps used there also. 

OK, I now have the two cameras with the same circuitry, (.22 uf) and they
are behaving the same.  Sigh!  Now I can see that there are still a few
problems left.  But progress is being made. 

Much of this was caused by working a little too frantically giving it the
old college try to get a Mark IV to AZ before the snow.  I don't regret the
attempt.  This one just missed.  I have won a lot of these kinds of efforts
so one is expected to lose from time to time.

I have solved three major problems since last October.  Any one of the
three would have kept me from shipping out the camera.  One is this turn on
transient.  Another is the RA drive.  The third is that I discovered how I
have been breaking the TECs.  There are still some problems left.  But I
seem to be back in success mode instead of break everything mode.

Tom Droege