EEVblog #638 - Apollo Saturn V LVDC Testing

[EEVblog]

Dave tests to see if the chips in a 1960's Apollo LVDC  Saturn V Launch Vehicle Digital Computer board still work.

Fran Blanche:
https://www.youtube.com/user/ContourCorsets
http://www.frantone.com/designwritings/design_writings6.html#lvdcproject
http://www.frantone.com/designwritings/design_writings7.html#LVDC2
The Dinosaur Den
https://www.youtube.com/user/TheDinosaurDenShow
http://fran-n-bil.com/

[HP-ILnerd]

Phenomenal.
I love vintage tech, Apollo-era tech in particular.  Those guys were really striking out into unknown territory on every front.
Personally, I can't get enough of this sort of thing.

[EEVblog]

Did they make a functional programming computer from discrete components? I doubt emulation would have been easy in the 1960's.

Yeah, that would be interesting to know!
You would think they would have had to have something to emulate/test on?

[HP-ILnerd]

I may go out to visit him late this summer and pick up what I can and help him clean out his techno-junk ( as he calls it). I will share what I can with the EEVBLOG community and then send the rest off to the Smithsonian
Bob

I hope you have a decent camera!

[G7PSK]

I read that in flight the equipment ran on 28 volts but on the launch pad it ran at 97 volts, and it was running at the high voltage on the pad that caused the insulation on the LOX heater on Apollo 13 to fail and led to the eventual blowing of the tank when it was stirred on the way to the moon.

[digital]

What a great video it was a treat to see the board and the layout it would be great to be able to talk to the design engineers from that time that were involved in the design of that board and why they did not compact the chips further.When you slid the chip out of the connectors i thought you may have used your heat gun to gently burn of the oxidation but hey you did it.thumbs up Dave.Cheers

[max_torque]

On a project as complicated and interwoven as the Apollo system, designers of individual systems were typically just given a set of targets and told to get on with it!  Due to the lead times, and constraints of project management** you were given a mass/space/power envelope and a set of performance requirements and from those you designed your system.  With reliability and flexibility high up the list, much higher than capability / integration.

Because of those factors, the designers of all the Apollo systems chose proven, high reliability blocks from which to construct their systems.  With it's roots in the Mercury and Gemini space projects, a lot of the electronic tech was even at the start of the project nearly a decade old, but when someone says "I want man on the moon in under 10 years" you don't start re-inventing the wheel ;-)

The system of very basic ICs on cards is incredibly flexible, and because each "chunk" is relatively simple, testing and production validation during build / assembly is much easier.  The requirement to provide test hardware early in the program life (a long time before anything actually flew) mean't that it was inevitable that changes to the control logic and execution paths would occur, and that a massively modular architecture enables that to occur without significant re-certification for flight (In effect, you make a few very basic components, certify the heck out of them, and then use them in their thousands, rather than using fewer but more complex and each slightly different components!)


There are a good series of videos uploaed to Youtube that cover the status of the Apollo project through the years.  As someone who thought the project was about building a moon rocket, the amount of support systems and infrastructure engineering and development is simply mind blowing.  Compared to some of the test systems, building the actual rocket was relatively easy:





** imagine trying to project manage a project the size of Apollo today, even with all our computerised management systems it would be a massive PITA, in the 1960's with paper and pencil, just managing the work scope/flow must have been a nightmare!)

[EEVblog]

The system of very basic ICs on cards is incredibly flexible, and because each "chunk" is relatively simple

Sure but I expected at a bit more integration like a maybe latch chip or something?
Some of these chips only have two resistors in them!

[Radio Tech]

Great video of this classic 1960 era design.
I do however find it a bit strange that the chips are slid into holders and not soldered in place. 
I know the chip was a bit tough to move due to years of build up and oxidation.
We know the rocket produced a lot of vibration and thrust, I guess they were not concerned about the chips sliding out of the holders.

[German_EE]

My nomination for the next destination of the board is Mike @ Mike's Electric Stuff, he has an X-Ray machine and should be able to see inside the chips and the board layers.

I was always told that TTL was invented for the Apollo program and it looks like you have proved me wrong as those chips use RTL. It's still an interesting video though and like many others I was surprised by the spring contacts. Weight was a massive problem on the Apollo flights and subcontractors were limited in the amount of solder that they could use on the joints so those clips are a puzzle.

The combination of Health & Safety, MBAs and Death by Powerpoint mean that we probably couldn't build the Saturn V now and I'm not holding my breath for the new launch systems. Sad.

[EEVblog]

My nomination for the next destination of the board is Mike @ Mike's Electric Stuff, he has an X-Ray machine and should be able to see inside the chips and the board layers.

Fran has already done that.

[LabSpokane]

Dave,

The US space program is extremely conservative. Even the Columbia shuttle used Apollo era core memory - at least initially. Two resistors on an "IC" isn't too surprising given the era.

[RupertGo]

TTL certainly built on the modular approach that Apollo spearheaded, but it was quite different in concept otherwise. http://www.embedded.com/electronics-blogs/break-points/4431173/TTL-turns-50--more-or-less--in-2014- is an interesting piece that points out 2014 is the 50th anniversary of the 5400 series of parts. Perhaps Dave could do a Birthday Special on SSI and MSI logic? I cut my teeth on 74/74LS/40XX TTL and CMOS parts..

[SeanB]

Built in the 1960's, using a frozen tech that was probably started in 1961 and held there.


crDSC01484 by SeanB_ZA, on Flickr

This is an example of late 1960's to early 1970's avionics, simple SSI on a large board designed somewhat for being repairable. Has some serial CCD memory as well, in the 2 Mostek chips on there. Made in the late 1970's, but the design was finalised probably a few years before for testing and certification. Most of the chips are now completely obsolete.

[bktemp]

In comparision with the Apollo Guidance Computer, the LVDC looks ancient:
http://en.wikipedia.org/wiki/Apollo_Guidance_Computer

[max_torque]

I don't know for sure, but i would suggest that the basic "rocket guidance and control" tasks for apollo had their roots firmly in Mercury and Gemini, as getting up into orbit is broadly the same task for all conventional rocket systems.  That makes the components and sub systems in the LVDC much more likely to have their ancestry in the early 1950's.

However, the guidance computer, required to leave orbit and get to (and back from ;-) the moon, well, that was a very specific Apollo requirement, and as such could be expected to be made up of more up-to-date technology and processes as it must had to have been developed specifically for Apollo.

[RobP]

Interesting photo of the full LVDC here:
http://upload.wikimedia.org/wikipedia/commons/thumb/b/b5/WernherVonBraunAstrionics.jpg/1280px-WernherVonBraunAstrionics.jpg

[PA0PBZ]

It's a 12 layer board:



The resistors are on the bottom of the module:



I have a lot of info about the LVDC, unfortunately the PDF is 84 MB, just a little bit too much to attach 

[KK]

Phenomenal.
I love vintage tech, Apollo-era tech in particular.  Those guys were really striking out into unknown territory on every front.
Personally, I can't get enough of this sort of thing.

My Uncle ( now close to 90 years old) lives in New Jersey and was the program manager for the LEM ( lunar excursion module) when he worked for Grumman in the 1960's. He still has a letter from LBJ hanging in his living room congratulating him and his team on the successful mission to the moon. He says that he doesn't have much hardware but has some declassified schematics of the cameras used on the LEM. I may go out to visit him late this summer and pick up what I can and help him clean out his techno-junk ( as he calls it). I will share what I can with the EEVBLOG community and then send the rest off to the Smithsonian

Bob

"From the Earth to the Moon" was a great documentary about the Apollo program. They had a specific segment on the LEM and the Grumman engineering team in NJ led by Tom Kelly. There is a scene where he is on the phone with his boss, presumably your Uncle? getting the go ahead for the project. I can't remember that name.

Fine engineering from all involved in that project!

[PA0PBZ]

The original drawings of the INV and AA modules. I sent them to Fran, maybe this is what was sent to Dave also:





Examples of the two highest usage blocks are shown in Figures 2-10
and 2-11. Figure 2-10 shows an INV module which contains an inverter with
a permanently connected 1.5k AND and an extra 2.5k AND. Two INV modules
are required to form a latch. The AA module of Figure 2-11 is used to obtain
AND and OR diodes and AND resistors. For example, the AA module may be
used to obtain two three-input 2.5k AND's with two OR diodes, a seven-input
2.5k AND with one OR diode, or just to obtain eight AND diodes. The versatility
of these modules allows just two modules to satisfy all logic connections
with minimum waste of unused components.

[bobwidlar]

I can only guess why they didn't integrate further is because they started the project years before the launch. So, IC technology was relatively new and they didn't order IC's for the project. It's already mentioned above, Apollo guidance computer's first version was using ~4k ICs and the most complex one was 3-input NOR gate (3 BJT + 4 resistor). That was it, that was what possible with the first planar bipolar process. In 1964 we can see uA702 and integration boost coming.

BTW: here's a nice talk about Fairchild history, and around 48:00 they talk a little bit about how tough was to meet apollo program specs http://youtu.be/arIXj8xpf6Q?t=47m45s Later on they say that each chip of dual 3-input NOR gate they sell to Apollo program was 9.9$ but for other customers it was 0.99$

[warp_foo]

Phenomenal.
I love vintage tech, Apollo-era tech in particular.  Those guys were really striking out into unknown territory on every front.
Personally, I can't get enough of this sort of thing.

My Uncle ( now close to 90 years old) lives in New Jersey and was the program manager for the LEM ( lunar excursion module) when he worked for Grumman in the 1960's. He still has a letter from LBJ hanging in his living room congratulating him and his team on the successful mission to the moon. He says that he doesn't have much hardware but has some declassified schematics of the cameras used on the LEM. I may go out to visit him late this summer and pick up what I can and help him clean out his techno-junk ( as he calls it). I will share what I can with the EEVBLOG community and then send the rest off to the Smithsonian

Bob

"From the Earth to the Moon" was a great documentary about the Apollo program. They had a specific segment on the LEM and the Grumman engineering team in NJ led by Tom Kelly. There is a scene where he is on the phone with his boss, presumably your Uncle? getting the go ahead for the project. I can't remember that name.

Fine engineering from all involved in that project!

Spider. Best episode of the series..

[PA4TIM]

I restored a HP9100A and B. They are from the dame eara. That is, one I managed to repair, the other showed some sings of live, (but it was found under a floor, near sea, and showed signs of salt water).  No sevice manuals, a big collection of hand-drawn schematics made by tony, the authority in 9100 land, but very hard to follow and project to actual pcbs. Everything was current driven that made probing not easy. The very rare, short, electrostatic CRT that was broken in the B version. Most common problem with these beasts.

This is a cutting edge calulator, the first programmable sceintific calculator (it even can do complex numbers)

I made a lot of pictures and it is made very interesting. It has hundereds of transistors and diodes. uses core memonrie and, a magnetic card reader. But on the other hand was the first use of teflon as pcb and some boards are multilayer with many layers.

If I understand well the ROM was made by a multilayer pcb which formed a matrix of crossing wires to makle some sort of replacement for core memory.]

If you can lay your hands on one it could be footage for an interesting tear down video. I do not have them anymore. I restored them for a museum.

http://www.pa4tim.nl/?p=3964   The 9100A that was very easy to fix
http://www.pa4tim.nl/?p=3878  The 9100B with lots of detail pictures. I got it to work with a round tube that fitted the cabinet more or less. But it seemed like the rom was gone. The crt showed some activity, it seemed to react to operators but it made no sence, After 2 months I gave up.

The multilayer teflon board :


Some of the > 1000 diodes in this thing

[Homer J Simpson]

Here is the like to the Wikipedia article on the LVDC with some more details.

http://en.wikipedia.org/wiki/Saturn_Launch_Vehicle_Digital_Computer


Triple redundant logic with a voting system.

2.048 MHz.

Reliability of 99.6% over 250 hours of operation.

and.........water cooling.

Not bad for 50 years ago.

KT

[EEVblog]

If you can lay your hands on one it could be footage for an interesting tear down video. I do not have them anymore. I restored them for a museum.

I've had the keyword on ebay watch for some time! No joy 

[Richard Crowley]

It appears that a major limitation on integration levels were those "huge" thick-film resistors on the bottom of the (presumably?) ceramic substrates.  Clearly there was adequate space available on the top for the "flip-chip" transistor and diode dice.

Interesting that modern CPU chips with 1.4 billion transistors and LGA "packages" are in many ways similar to that 50-year old technology. The LGA "package" is simply a small epoxy-glass PC board with printed pads on the bottom ("land-grid array") and on the top, matching the "pads" on the inverted die. Much simpler than when we used to do individual bond wires to each pin along the edge.

[KK]

Hardening against cosmic radiation was a requirement for the systems on Apollo. Might have had some influence on the pcb coatings and component spacing.

[bobwidlar]

Hardening against cosmic radiation was a requirement for the systems on Apollo. Might have had some influence on the pcb coatings and component spacing.

I'm not sure, but definitely PCB coating and packaging of the ICs required radiation protection. Even airplane circuits requires these kind of protection in a lower level of course.

[VK3DRB]

In the mid 1980's, I was a specialist on a machine that the architecture was spawned from the computer used in Apollo 11 - the Series/1. The Series/1 was used in a wide variety of applications in Australia. For example I worked on machines that:

- Monitored the conductivity of flesh of sheep used in experiments (the place stunk)
- Counting soiled laundry from hospitals (the place stunk worse than the sheep)
- Interfaced an array of 3101 VDU's to the System/7 for Telecom's Directory Assistance
- Interfacing the major banks to the overnight money market (Tandem operation for ultra reliability)

As for DTL, my first job out of uni was working on a machine that used DTL boards at IBM. They were used in the IBM 129 Card Punch. The cards were very similar to that Dave has been testing, but by 1971 the IC's were encapsulated in aluminium covers. It was amazing the low level logic used in that machine. As I recall, the 129 had no microprocessor, but had 80 columns of memory for verifying punch cards and a digital display - it was electronic poetry in motion. Here is a peek as to the innards of the 129 (check out the DTL board and the ratsnest of wirewraps): http://sturgeon.css.psu.edu/~mloewen/Oldtech/Keypunch/IBM129.html

I was very lucky to have been trained at IBM to debug much older equipment than what was used in Apollo, like the IBM 557 Alphabetic Interpreter - an immensely complex electro-mechanical programmable machine that was full of relays, commutators, motors, cams and selenium rectifiers. Quite challenging to debug. On such equipment, memory was achieved with valves (vacuum tubes) that had on-board resistors and capacitors INSIDE the valve. These were integrated circuits long before the semiconductor IC was invented. http://www.columbia.edu/cu/computinghistory/557.html

The oldest machine I ever had to fix was made in 1922 called the 011 electric hand punch made by the Computing-Tabulating-Recording Company (CTR) which became IBM in 1924. I only saw on once but it was a privilege to fix that for a customers. http://www.columbia.edu/cu/computinghistory/011.html

For newbies who don't know about what these machines were used for, check this out...

[dgtl]

1) It looked from the video like some of the metal contacts had left impressions on the bottom epoxy. Thus, the metal contacts are probably put on before the chip is covered with the epoxy. Finally the ready-made chip is soldered to the board.
2) Are the top and bottom sides electrically connected on the chip? Maybe the metal contacts connect the tracks at the top and bottom sides and when the chip is removed from board with the contacts ripped off from it, the connection may be broken and thus the resistors should be probed on the bottom side and diodes-transistors at the top side?

[BMac]

OK Dave now you have a working Saturn V Rocket diode, you need to feature it in a project!

BMac

[EEVblog]

OK Dave now you have a working Saturn V Rocket diode, you need to feature it in a project!

I would if you could reliably connect to the thing...