I have come into possession of a KAF-0400C CCD Sensor

It's a camera sensor from the late 90's from what I can tell, and it's got big fat 9μm pixels. I want to see if I can turn this into a telescope camera like this guy did. I found the datasheet here and I have confirmed it's the same chip by looking at how the traces are connected on the board I desoldered it from.
For anyone curious, I fished this chip out of a Kodak DC50 Zoom camera that I grabbed before the local school district threw it out while I was still working there.
Mainly at this point, I would like any help anyone is willing to offer with understanding the datasheet. I know enough to start figuring out how it says to use the chip but I'm not confident enough in my datasheet reading abilities to go ahead and design a circuit with a microcontroller etc. right away, especially since they don't make these anymore, and the next generation (KAF-0402) are like $350. I really don't want to fry this thing.
Thanks to anyone who helps out here! Even if it's telling me that this is totally infeasible

The short version is that it acts like a pair of shift registers; you clock 794 pixels out on a row by twiddling the horizontal clocks in opposing phase (in lockstep, one rises and the other falls) to get analog pixel values on the output, and between rows you clock the next row into the output register in a similar way using the vertical clocks. Section 2.5 is pretty clear on how to operate that. Understanding section 1 of the datasheet will allow you to better understand why things need to be done as awkwardly as they need to be.

There are going to be two or three main challenges in using this device: it takes weird voltages on the inputs (see section 2.4; the vertical clocks are -8 to 0V, and the horizontal are -4 to +6V) so you won't be able to directly interface to a microcontroller. It also seems to require three different power supplies: well-controlled 10V and 15V for the output amplifier and reset, plus 7V on the guard ring. Finally, the onboard amplifier is not very powerful- section 3.1 says you can load it with no more than 10 pF which probably means you need an external amplifier to redrive it into an ADC. Today's amplifiers are much better than whatever they had in the 90s so this shouldn't be too hard, but it's more parts to incorporate.

I could probably work on puzzling this out more myself, but that's starting to stray into "building it for you" territory which I'm not really interested in doing. Have a go at it and continue asking questions.

I note that your source has schematics available, though it is a rather archaic design (nearly 20 years old now!). Referring to that will likely be enlightening.

Thanks so much! It was definitely section 2.4 that I was most confused about just because of all the weird voltages. Essentially This becomes "Can I design amplifiers that can send those weird voltages from a 5V logic level signal" which I do think I have the knowhow to do. Unfortunately, I don't see any information about the Vout signal other than on the actual timing diagram in 2.5 that seems to suggest it goes from "reference" (GND?) to reference - 450mV. Luckily that is an output so it can be found out experimentally.

Alright, this is definitely the weirdest chip I've ever worked with. I've come up with a basic design concept that I think should essentially convert all its inputs to 0-5 volts and the output to 0-4.5V so that I can use a microcontroller to talk to it. Basically it's just a zillion op-amps and a giant voltage ladder. I went with the voltage ladder because the voltages I need are -8V, -4V, -2V, 0.5V, 3V, 5V, 6V, 7V, 10V, and 15V. The icing on the cake is that the pitch of the pins is 0.07".
The circuit now just takes +-15V as source voltage and then has 5V logic level inputs and outputs using this circuit here:

All the op-amps are single circuit SOT23-5 op-amps from digikey (576-1314-1-ND) and all the resistors are 0402 (the smallest I can solder) so that I can keep this thing fairly small. I may also add a small section of 0.1" spaced vias just in case I need to make any bodges, but this is the basic idea.

Please leave thoughts and suggestions! Maybe about how I'm going to get the + and - 15V to power this thing. I can't really use AC power since it's going to be used in the field with no outlets...

That seems like it would work, but it's not how I would approach this.

Looking at the AUDINE schematics, they also use +-15V supplies and op-amps to level-shift the clocks, so that seems to make sense. To generate all the other voltages they seem to use various Zener diodes which will tend to burn some power but is quite stable.

The main issues with your design here are that you need pretty precise resistor values (probably 1% or maybe even better), and any noise on the +-15V supplies will be passed directly onto all the other rails. Since the resistors are all chained you might need better than 1% tolerance on the resistors (0.1% maybe? I haven't done any math on that) to ensure the rails are all in spec. That's probably expensive and awkward.

For power stability, the reality is that this approach cannot fix propagating ripple; all you can do is add filter capacitors to mitigate the potential for issues.

To generate power rails I would prefer to use a variety of LDOs, which will have similar stability to the Zener diodes in the AUDINE design (very stable) even in the face of input ripple, but should generally burn less power and be easier to set up. The physical size should be similar to what you've got. For level-shifting since stability isn't important it seems reasonable to do what they've done and use op-amps in appropriately trimmed noninverting configurations to generate the clocks from the 15V rails.

Good note about LDO's. I did look into doing a zener diode approach but it was the amount of power consumption that turned me off of that idea, especially since I would need 10 of those zener regulator circuits and that would just eat power. They also didn't make zener diodes in all of the voltages I need which I found odd (No 8V zener diodes but all the 8.2V zener diodes you could want.) So yeah, going with a variety of voltage regulators could be the way to go.
Would it be true though that for the voltage ladder approach, the resistors wouldn't have to 0.1% within 1k, rather 0.1% with respect to each other? Probably not going to go that route anyway but just out of curiosity.

I'm still trying to figure out how to get +-15V in the first place without any outside power. I've looked into supplying 24V, and then using a circuit like this to generate -24V (noisy with any kind of load) and throwing that into a -15V regulator with the proper decoupling caps. According to my simulations, this would allow me to draw quite a bit of current with big enough capacitors before the regulator would start dropping out. Still, 24V is not usually a power source that comes in a small package in my experience.

Any suggestions on powering this thing still very welcome.

Handwaving a bit, if you use 1% resistors and have 30 of them the whole stack could be up to 30% out of spec. If they're all the same value it doesn't really matter, but resistors aren't sold that way. You could bin them yourself to ensure values are consistent (buy 10% or 1% or whatever then measure the values yourself and only use the ones that are close enough), but that sounds tedious.

Your suggestion for power supply is a charge pump, which could be done but tends to not have large power capacity (but then, I don't think you need much). It seems reasonable, especially if you get a dedicated charge pump controller to do it.
Alternatively you can simply generate ~30V and use a pair of linear regulators to place the circuit's virtual ground between them, like this simplified schematic:

I chose 36V here because that's conveniently generated from 4x9V batteries, but it could also be generated from some other source like a lithium battery with a boost converter to step up to a bit over 30V.

Would you also have to connect the net in between the two 18V sources to ground? I do like that though. Probably less noise than the charge pump. I have very little knowledge on boost converters, so I can read up on those. 9V batteries would definitely be the easier approach but I'm not exactly in love with the idea of replacing batteries in this thing all the time.

Gonna work on taking these ideas and putting them into motion in a new schematic. I'll post that here when it's done!

Anything to report here, or did the project fall by the wayside?

---

I was paging through a copy of Analog SEEKrets recently and in reviewing this topic the chapter on tolerancing seemed very relevant to your earlier resistor-chain-as-power-supply concept.

There's also some useful commentary on building virtual grounds out there that's probably more useful than my quick sketch that may have fatal flaws. And here's another TI app note on rail splitting that may be informative.

It did fall a little by the wayside since I actually got a real astrophotography camera, but I still eventually want to get it to work so that I can use it with a microcontroller for other projects. I am now a semester and a half older and wiser so probably at some point I'll pick this back up. Definitely agree that using individual op-amps to create all the required levels is a way better solution that even allows for fine tuning. Once I do I'll be sure to post here.