Driver array for -20V, low current -- recommendations?

[ebastler]

I am looking to add a microcontroller or FPGA based module to an old vacuum-tube computer. This is to replace the original magnetic drum memory, at least temporarily, or maybe permanently if it turns out to be beyond repair.

The vacuum tube logic levels are 0V and -20V. I need on the order of 16 output bits and 16 input bits, which have to be level-converted to 3.3V or 5V (to be decided). I have not had the opportunity to measure inside the old machine, but would expect its logic output levels to be reasonably clean: They are limited to 0V and -20V by clamping diodes at each gate output. I have also not measured any signal currents yet, but expect these to be in the mA range. (Driving tube grids and resistors in the 10k range.) Digital switching would be all that is required, of course, at pedestrian speeds -- the clock rate is 125 kHz.

I am looking for a compact and simple solution for level-conversion (in either direction). In my preliminary search, I keep ending up with oversized power drivers, H bridges etc.. Are there any nice driver arrays, say 8 gates per package, that you would recommend? Or other obvious solutions I am overlooking?

Thanks for any advice you might have!
Juergen

[jbb]

I thoroughly recommend using 3.3V logic rails. It will give you a bigger range of microcontroller / FPGA options.

Why not reference your 3V3 support my to the -20V rail? That might simplify matters. An isolator chip can be used to connect the debugger or comms.

Regarding your logic signals: if they're expecting a 10k load, would that be pulling up to 0V or pulling down to -20V? You should be wary here, as the designers of the original equipment probably used many design tricks to reduce component count and the exact behaviour could be quite important.

For input signals, a bank of voltage dividers and comparators is a good candidate.

For output some signals, you need more info. Are they push-pull or pull-up or pull-down type? Are they always outputs, or do they need to support tri state (I.e. no-output)? Also, your driver should not blow up if short circuited.

[ebastler]

Why not reference your 3V3 support my to the -20V rail? That might simplify matters. An isolator chip can be used to connect the debugger or comms.

Yes, that might be an option if I can tap into the original -20V supply. I want to try and keep this retrofit "minimally invasive", using existing plug connections into the machine if possible. Supply voltages are on a separate connector, so it may be cleaner to just connect to GND and generate -20V locally.

Regarding your logic signals: if they're expecting a 10k load, would that be pulling up to 0V or pulling down to -20V? You should be wary here, as the designers of the original equipment probably used many design tricks to reduce component count and the exact behaviour could be quite important.

For output some signals, you need more info. Are they push-pull or pull-up or pull-down type? Are they always outputs, or do they need to support tri state (I.e. no-output)?

Good point. The main logic board uses diode-resistor logic: pullup resistors to positive potential, and signals pulling down via diodes. It is difficult to verify that there are pullups for all logic equations however; there are about 1400 diodes and the schematic gets a bit complex... The original tube flip-flop outputs have 10k or 15k pullups also. So providing these pullups and a single-ended transistor output to pull down to -20V should be adequate for the new output stages.

[danadak]

You might search net for VF drivers, some have the logic translation for
- going signals.  Or octal comparator as another possibility.


http://www.chinaeds.com/zl/laoli%5CT%5CTexas%20Instruments%5CLM%5CLM339X2.pdf


Regards, Dana.

[ebastler]

Thanks, Dana! I like the idea of differential receivers; seems like a clean and compact solution.

Sorry -- I'm blanking on the "VF" acronym. If I Google for "VF drivers", I find plenty of interesting but not-so-relevant information, because "forward voltage" is specified in so many datasheets... But you probably meant something else? Thanks for a hint!

EDIT: Aahh, now it clicked -- vacuum fluorescence, right? I will have a look at these drivers now...

[danadak]

vacuum fluorescence, thats correct.

Regards, Dana.