Misc. SSI computer boards

[D Straney]

Just thought I'd show off some interesting old computer boards I've ended up with, from non-integrated-CPU systems with mostly SSI logic ICs.  There's already the Univac CP-901 flight computer board, and a Honeywell 316 industrial computer core-memory-related board (helpfully identified by PA0PBZ), but now there's also...

Sperry-Univac 7165534-02 from AN/UYK-7



I'm pretty sure this is from the 32-bit AN/UYK-7 computer, which was the US Navy's standard shipboard computer for roughly 15 years through the 70's and part of the 80's.  The reason I'm confident in the source is because https://vipclubmn.org/Artifacts.html shows a similar board (IMG_1873.JPG, bottom-right corner) with a directly adjacent part number (7165535 vs. 7165534) and identical construction, which it identifies as being from the AN/UYK-7.

As for what this board actually does...

...I'm not totally sure.  The array of magnetics in the middle suggests either ferrite core memory or transformers for driving or receiving external signals of some kind.  However, having a group of 3 transistors on one side, and a group of 6 transistors on the other side, with 18 cores total, strongly implies that they're electrically arranged in a 6 x 3 matrix, which leans heavily towards the "core memory" theory.  I did manage to find two (1, 2) technical descriptions from Bitsavers, but there's no direct clues there.  Besides the 32-bit word size, this would be an impractically low-density way of implementing the main memory, so this could be some kind of auxiliary storage.  It explicitly mentions that the "control memory" is provided by ICs, so it can't be a register:

...memory used during the execution of instructions and input or output transfers, to capture, maintain, and provide status information when needed; and to provide various controlling  addresses and data as dictated by the operating programs.
...Eighty-two integrated-circuit random-access registers of appropriate size serve as the central processor control memory. The various registers are grouped into stacks according to their use, addressing, and relative size.

However, the address length is also 18 bits, so this is probably storing something related to addressing.

There's another one of these available if anyone else is interested.

[D Straney]

Marconi-Elliott 6743-00191B03 Navigation Computer Board


From looking up the part number (equiv. NSN 5998-00-480-6599), the applications info I found on this was "FSC APPLICATION DATA:AIRBORNE INERTIAL NAVIGATIONAL SYSTEMS" and that it's from an A-7E aircraft.  According to that article, it has a fancy projected map display from the navigation computer (one of the first systems of that kind supposedly), tied into the inertial navigation system, which also could update its absolute location (both as an initial reference, and to correct for drift over time) from multiple external sources of position info.  So it sounds like this could be either part of the listed "AN/ASN-90(V) Inertial reference system" (if it has its own digital logic), or the "AN/ASN-91(V) navigation/weapon delivery computer".

The particular board is marked "Clock & Checkout", and has a 6 Mhz crystal, so at a minimum it generates clocks for the rest of the logic.  Because of the timeframe (the date codes are 1972-1973, and the NSN record was created in 1970) I wasn't able to find any info on the ICs, as they have specialized possibly-custom part numbers, but the info on the board lists its contents as "RESISTOR 63,CAPACITOR 31,PANEL 1,TERMINAL POST 2,DIODE 13,TRANSISTOR 9,RETAINER 5,FLIP FLOP 13,AMPLIFIER 6".  I'm guessing the many chips at the bottom edge are the flip-flops, which probably divide the clock signal and split it up into multiple phases for different parts of the computer, similar to the way that old integrated CPUs like the 8086 require multi-phase clocks to schedule instruction decoding / memory bus access / etc. separately.


There's also a mystery diode(?) down there:


Some of the gold chips with the Teledyne logo at the top edge are probably the "amplifiers" listed in the board contents.  There's also some discrete transistors there, mounted upside-down in heatsinking clips.  I'm not sure what the purpose of the analog section is: would expect a little bit of circuitry next to the crystal to be the oscillator circuit that's tuned by the crystal, but no idea what the rest of it would be for.  I'm also not sure what the "checkout" function of the board refers to, so it could be related to that.



One feature mentioned in the AN/UYK-7 technical description was how it would generate an interrupt if the power input failed, at which point it would have 250 µs to "clean up" and put things in a safe state; another useful feature in a "high-reliability" or "high-availability" computer would be generating a backup clock source if the crystal failed to start.  So it's possible that some of the analog circuitry here is a combination of power rail monitoring, to inform the stored program if something isn't working right for diagnostic purposes, and also as a clock supervisor circuit, possibly with a backup RC oscillator (like many microcontrollers), or maybe even also implements a hardware watchdog timer.  Any extra info appreciated.

Edit: swapped in better photos

[helius]

In computers of that era it was common to have "margining" controls on the operator console. The operator could dial the power supply and the clock frequency up or down 10 or 20 % from the nominal value for testing. If no errors occur at extreme operation points it increases confidence that there are no latent faults.

[D Straney]

Ahh good to know, that makes sense both for the "checkout" label, and the concentration of analog parts!

[Alex Eisenhut]

Whoah 

[D Straney]

Martin-Marietta mystery digital board
I'd originally thought this was part of a computer built from individual SSI ICs, but after looking into it more, I think it's more likely part of a hard-wired digital system.  More details below.




This mystery board turned up on eBay for very cheap, and conveniently even had a tag already attached to it:


There's some passive components sprinkled around:




A very thick waxy-looking coating covers most of the board, but in a few places it's either flaked off by itself or been removed, over a few ICs and part of a resistor - this was helpful for reading the IC part numbers:

The ICs mostly have Fairchild logos: the majority are marked "SL 4373 99B11-1", with a couple "SL 4372 99B13-1", and a couple more "SW 946-1F".  I couldn't find "4373" or "4372" as standard Fairchild part numbers, from reading through the bitsavers copies of 1969 and 1970 databooks.  The markings and part numbers also don't match up to the DTµL-series parts that were on the Honeywell 316 card.
However, maybe I was looking too late - after all, the "overhaul date" on the tag is 1970, which means this board was probably designed then built well before that.  Fairchild's first logic ICs were the Micrologic (µL) series, which all were packaged in 8-lead metal cans (as described in an EDN article).  These were even used in the Martin-Marietta(!) MARTAC 420 computer - which turned out to be a red herring, as its packaging and construction are completely different from this board.
The µL ICs had different letters for different functions, and "S" was for a half-shift(?) register: it's possible that the "SL" in the marking here refers to the same thing.  The "S" pinout in that old EDN article shows that it uses 6 signal pins + 2 power pins, and the flatpacks on this board have 14 pins, with most of them used, so it's possible that these contain 2 separate "S" logic elements with shared power pins: 2*6 + 2 = 14.

Knowing now that the board may or may not be a long string of shift registers, we can make some guesses about its end use from the info on the tag.  From looking at the Wikipedia page for the Glenn Martin aircraft company, and Martin-Marietta, it looks like they were in the aircraft business earlier on, and made some experimental missiles throughout the 50's, but by the 60's were primarily focused on the Titan series of space launch vehicles / ICBMs.  So while it's possible that this was part of some upgrade for older equipment, or a special one-off project, there's a good chance it was used for something related to the Titan.

Now, there's just as much electronics on the ground involved in a launch vehicle (the Titan checkout and control systems are supposedly what led to the MARTAC 420 general-purpose computer, according to the brochure linked earlier), so we wouldn't necessarily assume this was flight hardware.  However, the number on the tag ("5821 624 3132LF") seems to match the format of an old-style National Stock Number (NSN).  I couldn't find any matches from parts supplier websites on this number itself, but we can get some info from the first 4 digits: these correspond to a general category of supplies.  5998 is PCB assemblies, for example, and 5962 is for ICs (you'll see a lot of "5962-..." markings on photos of NASA electronics).  "5821" is "RADIO AND TELEVISION COMMUNICATION EQUIPMENT, AIRBORNE"; there's a separate code for non-airborne radio equipment.  This makes a good case for this being part of a telemetry or radio-control system on a Titan launcher.  Having a long chain of shift registers would be useful for either telemetry or radio control, to either serialize a lot of digital telemetry data to send over a radio channel, or to de-serialize a lot of control data received over a radio channel.

The "F04606-70-C-0467" on the tag turns out to be a government contract/bid number, which shows that it was a 1970 contract from something military-related.  If I could find a record of this particular contract, that would give some much more solid info about whether this came from a Titan rocket, or was part of some unrelated one-off telemetry project for someone else's vehicle(s).  There doesn't seem to be a public online database of these contract numbers, and the Library of Congress has some in-person documents that look promising, but that's way too much effort for my level of curiosity here.

Anyways, hope this was interesting.

[SeanB]

Just thought I'd show off some interesting old computer boards I've ended up with, from non-integrated-CPU systems with mostly SSI logic ICs.  There's already the Univac CP-901 flight computer board, and a Honeywell 316 industrial computer core-memory-related board (helpfully identified by PA0PBZ), but now there's also...

Sperry-Univac 7165534-02 from AN/UYK-7



I'm pretty sure this is from the 32-bit AN/UYK-7 computer, which was the US Navy's standard shipboard computer for roughly 15 years through the 70's and part of the 80's.  The reason I'm confident in the source is because https://vipclubmn.org/Artifacts.html shows a similar board (IMG_1873.JPG, bottom-right corner) with a directly adjacent part number (7165535 vs. 7165534) and identical construction, which it identifies as being from the AN/UYK-7.

As for what this board actually does...

...I'm not totally sure.  The array of magnetics in the middle suggests either ferrite core memory or transformers for driving or receiving external signals of some kind.  However, having a group of 3 transistors on one side, and a group of 6 transistors on the other side, with 18 cores total, strongly implies that they're electrically arranged in a 6 x 3 matrix, which leans heavily towards the "core memory" theory.  I did manage to find two (1, 2) technical descriptions from Bitsavers, but there's no direct clues there.  Besides the 32-bit word size, this would be an impractically low-density way of implementing the main memory, so this could be some kind of auxiliary storage.  It explicitly mentions that the "control memory" is provided by ICs, so it can't be a register:

...memory used during the execution of instructions and input or output transfers, to capture, maintain, and provide status information when needed; and to provide various controlling  addresses and data as dictated by the operating programs.
...Eighty-two integrated-circuit random-access registers of appropriate size serve as the central processor control memory. The various registers are grouped into stacks according to their use, addressing, and relative size.

However, the address length is also 18 bits, so this is probably storing something related to addressing.

There's another one of these available if anyone else is interested.

18 transformers, on what is an 18 bit system, looks a lot like a bus driver, as these kind of systems often enough had parts spread out over a few racks, so needed to have a pretty robust bus to transfer data around. 18 what looks like autotransformers could be a bus driver, 18 high power drivers for them, and then 18 silver mica capacitors, probably for peaking response, means it was probably used as a driver, and having it on a single board makes field repair easy, as there likely were a good number of places this would be used, so making it a module that would fail if something goes wrong in the wiring (given how sailors are kind of rough on equipment) to protect the rest, and keeping a chunk of spare modules in stock on board, is common.

[D Straney]

Good call - "address bus driver autotransformers" makes a lot of sense for the magnetics, esp. with caps for helping out the edge rate like you say.  Going to assume that's the case until a Univac part number reference happens to magically turn up.

Also found out you can send queries to the Library of Congress online: I asked about the contract number on the Martin-Marietta board, and a very helpful librarian directed me to a website with some records.  It looks like the original contracts all are considered not worth keeping and eventually get shredded, but I did find that this contract number covered 2 separate contract records:

Code: [Select]

DATE OF THIS ACTION 70I September, 1970
EST[IMATED] COMPLETION DATE 70J October, 1970
...
CONTRACTOR NAME MARTIN MARIETTA CORP MARTIN MARIETTA CORP
...
FSC [FEDERAL SUPPLY CLASS] DESCRIPTION RADIO / T V COMM EQUIP AIRBORNE RADIO / T V COMM EQUIP AIRBORNE
SYS [SYSTEM] OR EQU [EQUIPMENT] DESCRIPTION F-4 PHANTOM II F-4 PHANTOM II

Code: [Select]

DATE OF THIS ACTION 70B February, 1970
EST[IMATED] COMPLETION DATE 70J October, 1970
...
CONTRACTOR NAME MARTIN MARIETTA CORP MARTIN MARIETTA CORP
...
FSC [FEDERAL SUPPLY CLASS] DESCRIPTION RADIO / T V COMM EQUIP AIRBORNE RADIO / T V COMM EQUIP AIRBORNE
SYS [SYSTEM] OR EQU [EQUIPMENT] DESCRIPTION C-121 CONSTELLATION C-121 CONSTELLATION
So, looks like unfortunately this board has nothing to do with space or launch vehicles.  Since I was trying to procrastinate actually important work, though, I was curious and did a quick search for things in that time frame that happened with the C-121 aircraft, and the only one that fit the timeline perfectly is this "Operation Kingpin" during the Vietnam war: https://en.wikipedia.org/wiki/Lockheed_EC-121_Warning_Star#Operation_Kingpin
The fact that this contract also covered F-4 aircraft, which were involved with that mission, makes me more confident this is related - I'm guessing that other copies of this board were on the "airborne terminal" side of the two-way digital radio comms for "tactical data display", which actually failed to work in the field (gotta do better testing, guys).