Looking inside an old aircraft CDU and (hopefully) powering it up

[D Straney]

Here's an interesting eBay cheap-avionics find: a used aircraft CDU from 1994, made by Teledyne Systems.  (Control Display Unit: basically the main computer UI from what I understand)


From looking up the part numbers, it looks like this model was made for the US S-3 submarine hunter aircraft: https://en.wikipedia.org/wiki/Lockheed_S-3_Viking

After taking off the back cover, you can see a few different cards inserted into (as expected) a very nice frame for conduction-cooling the electronics:


Now, going through the cards in order...

Analog board


This has a mixture of misc. digital logic and resistor arrays (likely as pull-ups/pull-downs), with an analog section in the middle.  The part marked "PMI" (Precision Monolithics Inc., now part of Analog Devices) is an 8-bit DAC with a current output.  The 8-pin DIP to the left of it is a high-speed op-amp, which likely is set up as a transconductance amp for current-to-voltage conversion, and then the large metal can is a power op-amp that can drive up to 200 mA.  I don't know what external load exactly this would be driving: maybe some kind of (very very small) actuator, an unusually large analog gauge, or a reference signal to other equipment?


The resistors (presumably used as feedback for the power op-amp) are interesting too because they're encased in glass, and you can see the spiral cuts on the metal film inside that increase the length of the current path:

I could imagine that maybe these resistors are enclosed in transparent glass, rather than the normal potting with epoxy, so that they could be laser-trimmed after leads etc. are attached, for extra-tight final tolerance (as mechanical stresses from the rest of the resistor manufacturing process likely would shift the resistance somewhat).  I don't really know for sure either way, though.

ICs:

• JM38510/32403BRA = 54LS244: 8-bit buffer with 3-state outputs
• JM38510/37501BCA = 54ALS32: quad 2-input OR gate
• JM38510/32803BRA = 54LS245: 8-bit bidir bus transceiver with 3-state outputs
• JM38510/30903BEA = 54LS157: quad 2:1 digital mux
• JM38510/37001BCA = 54ALS00: quad 2-input NAND gate
• JM38510/30701BEA = 54LS138: 3:8-line decoder (is this doing address decoding here?)
• SNJ54ALS576AJ: octal D-FF inverted-output
• JM38510/11302BEA = DAC08: DAC 8-bit, multiplying 1 Mhz, 85 ns settling, compl. current outputs
• 5962 8962703P H0C9230 = HA7-5147/883: op-amp 100 Mhz low-offset (100µV), stable with gain ≥10
• LH0041G: 200 mA output power-op-amp

Comms board


This board seems to be for external communications, as it has two separate explicitly-labeled MIL-STD-1553 transceivers (a common serial bus used on aircraft and in military applications).  It looks like there's two separate channels, as there's two sets of...

• Pulse transformers (used for isolation of the serial bus)
• Physical-layer transceivers in metal cans
• Manchester encoder/decoder ICs specifically made for MIL-STD-1553 (the 7802901JA)

There's also a very nice, very wide ceramic DIP which turns out to be a 16-bit microcontroller, likely for offloading the communications tasks from the main processor:


There's also two mystery chips with many connections to the bus connector: wasn't able to find anything for the part number so these likely aren't off-the-shelf standard parts, but "62458" turns out to be the CAGE code for Raytheon, so these are probably some sort of custom or aerospace/military-specific parts.  Best guess is that they handle backplane bus interfacing?


ICs:

• M38510/21004BJA: PROM 2Kx8
• 8550201XA (large purple DIP): MCU 16-bit TTL
• (Metal can to its left is most likely 24 Mhz oscillator)
• 7802901JA: Manchester encoder/decoder & synchronizer for MIL-STD-1553
• HE1025-0-G-002: ?
• 5962-8605201 = MM54HC30: single 8-input NAND gate
• JM38510/33001BCA = 54F00: quad 2-input NAND gate

CPU board


This is the only board which is listed as "programmed" on the label, despite the microcontroller on the comms board:


First, the non-processor items: this board has the expected EEPROM (program memory?), RAM, and misc. glue logic - both standard TTL, and programmable, as with the PAL device in the corner and the larger "SEEQ" device, where I was only able to find vague references to the part number.
The CPU itself is an MD8085, which looks like it's just the mil-spec version of the normal Intel 8085 8-bit processor: something with consumer electronics that you would've expected to see only in the 80's, but makes sense with the low processing power requirements, long design cycles, and slower qualified part availability.  There's also a microcontroller though, the MD8748H: no idea what the "division of labor" is here, but best guess is that it handles low-level repetitive tasks that otherwise would steal too much time from the main processor, like keyboard scanning?

There's two similar-looking Raytheon(?) chips here too, with different part numbers but again maybe for bus interfacing?  The fact that these appear on both the CPU board and the comms board convinced me that the two copies on the comms board didn't correspond to the two MIL-STD-1553 channels, and that they weren't related directly to the serial connections.


ICs:

• MD8748H: MCU 8-bit MOS
• 7136799-1 / DM2817A-250: programmable logic?
• PAL...: programmable logic, erasable
• (Metal can is likely 12 Mhz oscillator)
• CY7C186A-45: SRAM 8Kx8
• 7901001QA / MD8085AH: CPU 8-bit 3 Mhz 8085 https://www.cpu-world.com/CPUs/8085/Intel-MD8085AH-B%20(7901001QA).html
• JM38510/22402BYA: EEPROM?
• HE0904-0-G-002: ?
• JM38510/34001BCA: quad 2-input AND gate, "F" family
• JM38510/33002BCA = 54F04: hex inverter
• JM38510/32403BRA = 54LS244: 8-bit buffer with 3-state outputs
• JM38510/32401BRA = 54LS240: 8-bit inverter with 3-state outputs
• JM38510/32803BRA = 54LS245: 8-bit bidir bus transceiver with 3-state outputs

Power supply




This (shielded) power supply most likely steps down the aircraft power to run the 5V supply that drives all the old power-hungry digital logic.  There's a UC1845-equivalent PWM controller, a single MOSFET (helpfully marked "G D S"), then two dual-diode(?) packages and two transformers or inductors.  With the single MOSFET, my guess is that it's a forward converter driving a transformer (left-most magnetics) with the 2x dual diodes used as a full-wave rectifier, feeding a buck inductor (right-most magnetics).  The additional diodes could be for snubbers, or there could be an additional voltage rail that it produces.


The arrangement of 4 individual diodes though does look suspiciously like a (lower-current) full-wave rectifier:

...and some of the metal cans could be transistors used as linear regulators (couldn't find what the larger can is, the smaller one with the Linear Technology logo seems to be a voltage reference):

The passives "standing on their head" against the metal I'm guessing are small inductors used for output filtering.

The DIP-14s are just LM139 equivalents, quad comparators, probably used to do UVLO/OVP/"power good" indication or sequencing of multiple power rails.


Power supply / EMI filter
This one's the chunkiest one by far, and responsible for most of the weight.




When I looked up the part number of the big metal box on this board, it was listed as an EMI filter, however I don't think that's the full story.  From the weight of this assembly I originally thought it was a transformer, and the only power supply onboard.  Unless it had some really, really serious low-frequency-suppressing chokes or had to deal with massive currents I would never expect an EMI filter to be this heavy; plus, it lists output voltages that don't appear on the input (70VAC and 18VAC).  The 28VDC inputs and outputs are unconnected, but the 115VAC input and 5VAC input/output are used: so it looks like it gets 115VAC & 5VAC externally - I've heard (but don't remember where from) that 5VAC is a common aircraft lighting thing?  So maybe that provides the backlights for the keypad?

There's a bridge rectifier and bulk capacitor hidden at the corner, but otherwise that's it for this board:


There are some unused footprints, which may be for inrush control or something like that:


This board also acts as a pass-through for a bunch of external I/O from the circular connector(s) directly to the backplane.  I do like the aesthetics of multi-layer boards where you can see traces buried at different levels inside, like mountain ranges slowly fading into the distance:


Powering up?
As everything looks intact, I was hoping I could turn it on and check that the display works at a minimum, maybe overwrite the internal program (if the CPU situation was nice and simple...which it's not) or figure out the display/keypad connections to swap out the CPU card as wall art and hook up something of my own to use the whole thing as a UI for something else.
Would've been very helpful if the "EMI filter" was really only an EMI filter, as in that case I could provide a DC supply directly to the output of the bridge rectifier on that board, and power up the system that way: however, because of the step-down action implied by the labeling, and the fact I couldn't find continuity from the 115VAC input terminals to any of the output terminals, it looks like that part really is a transformer.  So that means that further installments are going to have to wait until I can put together some kind of simple 400 Hz AC source to actually drive the input power connections properly.

[jpanhalt]

Can't help on that device, but for 400 Hz, an inexpensive option would be one phase from an inexpensive VFD ( variable frequency drive) for 3 phase motors.  A quick check for 400 Hz inverters showed prices higher than low power VFD prices.

[RoGeorge]

Nice pics! 
Looks so well built/expensive.

I've read boats also use 400Hz, maybe there are on ebay cheap but working 400Hz UPS/generators.

[D Straney]

Success!

Got tired of waiting for time that I wasn't going to have, to build a 400 Hz inverter, and decided to trace the first level of power connections to see if I could go about powering this up in a simpler way.  The big-ass transformer/EMI filter on the same board as the connectors (see photos above) has 3 outputs, helpfully listed on its label:

• 5 VAC: supposedly it's a standard for aircraft auxiliary lighting, probably backlight for keypad?
• 18 VAC, center-tapped
• 70 VAC

With a combination of visually following traces, and continuity checks, I found out that...

The 5 VAC goes directly to the backplane, as does the 18 VAC.  The 70 VAC goes to a bridge rectifier, rectangular inductor (previously mis-identified as a film cap), and bulk capacitor, all on the same board: this DC output then exits to the backplane.

On the SMPS board, the rectified-70VAC goes to a large pair of bulk caps that very much look like the input caps for the SMPS.  The 18VAC connects to this board too through the backplane, which worried me at first, but after examining the SMPS board from both sides with a very bright flashlight I was pretty confident that the 18VAC pins on its backplane connector didn't actually have any traces running to them (as in, the 18VAC enters the SMPS board but doesn't actually connect anywhere on it).

The final thing to check was the display, as I thought it might have an EL backlight or something similar that tapped directly off the 400 Hz power.  After going deep in the case I found the appropriate screws and was able to unplug the display, but luckily continuity checks showed that there was no 5VAC and no 18VAC on the display connector.  So overall, this meant it was likely nothing critical actually used either the 18VAC or the 5VAC.  I proceeded to snip the 70VAC wires from the transformer, strip them, and connect a set of leads, so I could feed the bridge rectifier and everything downstream from my DC supply instead of the transformer output.  70 VAC = 100 Vpk, but with some amount of averaging rather than peak-charging done by the onboard inductor, and to be a little cautious, I gave it 80V.  It drew 170 mA, but there were no signs of life as I tried pressing or holding all the different keys on the front.  Eventually though, after a full minute, the current draw jumped to 185 mA and the display turned on.

Had fun playing around with the test modes, as you can see in a couple quick videos here:

It took a couple tries to realize I was severely overthinking the display test instructions:


Nothing more I can do with the powered-up unit without the rest of an aircraft around it, though.  Next step is to make a breakout cable for both the screen and keyboard to scope all their signals during the test modes to figure out how to interface to both of them, with the goal of re-using the UI for one of my own projects.

[peter-h]

Interesting to see one 1993 date code on one chip, and the 1994 firmware date, while the general technology (through hole components etc) is much older - 1970s/1980s.

I thought those glass encapsulated resistors are flash tubes for EMP protection. They may be resistors as well, possibly of high value.

[D Straney]

Yeah it's interesting, because there's an (insanely expensive) Teledyne board, same CAGE code, from 1987 that's all surface-mount:
https://www.ebay.com/itm/125159995768
...and another surface-mount one from the same mid-90's time period as this (seller kindly looked into the background and told me it's from an aircraft engine controller, when I called to try and talk my way into a significantly lower price):
https://www.electronicsurplus.com/teledyne-systems-ps200-85004-519-military-circuit-board-assembly
So I'm assuming they went with all through-hole here because the circuitry fit very easily within the expected form factor of a CDU, and there was no need to make it more compact.

Interesting about the flash tubes: these are all definitely normal-value resistors but can see how those would have identical-looking packaging.  Reminds me of some EKG/ECG front-ends (see attached image; not mine) which look like they use neon lamps for surge protection (against a defibrillator rather than an EMP in this case).

Also meant to mention before: I appreciated the advice from jpanhalt and RoGeorge on 400 Hz sources.  I didn't find anything at the time that fit what I was willing to spend to power this on ($30 or less) but definitely going to remember the VFD and "boat inverter" approaches if a more serious application comes up.

[calzap]

Interesting chunk of gear.  Thanks for showing it.   What's the meaning of the caution sticker with a circle and 3 arrows near the two round connectors with infinite pins?   My guess is to warn that plugs are oriented correctly before pushing them in, but I don't know.

Mike

[peter-h]

from 1987 that's all surface-mount:

I designed a product in 1990 which was mostly SMT. And SMT is excellent for shock or high G. But yes the mil people tend to be way behind. OTOH SMT was going big c. 1980, in things like Japanese stuff.

There were continued issues back then with SMT soldering reliability. The Japs were very good at sorting it - even using 0.3mm pitch chips - but the West struggled for a while. Then later they struggled some more with lead-free

I didn't find anything at the time that fit what I was willing to spend to power this on

400Hz has been standard on aircraft because magnetics are 8x smaller on cross sectional area (64x less weight... maybe not quite) than 50Hz ones, and a 400Hz alternator is a lot smaller for the same power as well as easier because 50Hz requires a very low RPM (1500 RPM for a 2  pole rotor and you can't have less than 2 poles).

[harerod]

@calzap: Check out the text in several other pictures - ESD sensitivity warning.