Electronics

Andy found some driver ICs that can drive the wierd voltage levels needed by the CCD.

Input/Output card alternatives:

- a parallel IO card driving many signals out to the head where a good ADC and level shifter sits
If we use a paralell IO card we need to make our own ADC but we can put it near the CCD to reduce interference.
- the IO card I listed where we have the ADC sit in the PC
Alternately we can use the IO card with the ADC and risk picking up noise in the cable (hard to say if it would happen or not).

(Getting a device like an ADC working well is a tricky job so it might be better left to the professionals.)

Computer alternatives:

- desktop
might do more than run the CCD (maybe do some auto pointing) and would be an easier to maitain base.
a 24 bit parallel IO card suitable for heart of the CCD electronics for $100 (desktop machine PCI slot).
- laptop
more portable but might box us in.
a 48 bit parallel IO card for a laptop for $200. I'll donate the laptop - its a 486 with windows 3.1 and no battery but working txformer.

Timing:  Someone needs to buy the IO card,  and write low level code to
make the timing signals in the binder for the CCD and those required by
the ADC and read it in to a file.  This is relatively tedious but not
difficult.  Assembly is best or C program with embedded assembly - you can
disable interrupts in assembly so the computer wont go answering video
interrupt in the middle of a cycle.  There also needs to be a long cable
to the head and power supplies for  the head (linear).

There may or may not be a noise problem sending the low level signal back from the head
 - hard to say at this point.
But the log amp could be a possible patch.

Here's how the log amp might work:

data             Voltage of CCD output         Voltage of log amp output
bright star         1 volt                       3.5 volts
medium star         .1 volts                     3.0 volts
dim star            .001 volts                   2.0 volts
noise floor         .0008 volts                  1.9 volts

So .2 millivolts of noise would completely obscure any low level stars on the CCD output,
 but it would take 100 mv of noise to wipe out even the dimmest stars at the output of the log amp.
Its an 8 pin chip we can get cheap if needed.
The output of the log amp is a factor of 100 in voltage in = 1 volt out.

We also need a driver board near the CCD.
I found good chips for this, just have to add regulated 8,-4,-8, +4 supplies (one chip each).


There's plenty of Linux boxes around - I dont think we need a donation if you want to go that approach.
C++ can easily embed assembly for the timing generation
 and we definitely need some kind of display software to manipulate our final product.

What I envisioned as the design goes a little like this.
We take the timing diagrams in the book and turn them sidewise,
 making each signal a bit in a bus and turning it into a stream of binary numbers.

These numbers go out the IO card as digital signals to the head.
We translate to the wierd levels used by the CCD (not a big problem),
 then the ADC feeds still more bits of the IO card.
At the right time in the cycle we trigger the ADC and record its data.
Now we have a simple assembly program (so we can clear all interrrupts)
 that just writes a data file to the IO card (several meg words) and reads the input lines at the appropriate times.

Tomanipulate the timing we just change the file content.
Circuitry is simple, we have control at the software where we want it.
Also if the ADC is out by the head we have little potential for noise to interfere with the read.
What do you think?  The laptop I was going to donate is way
underpowered for linux.


(Andy)