Google-apper
Hovedmeny

Post a Comment On: Ken Shirriff's blog

"Inside a Titan missile guidance computer"

19 Comments -

1 – 19 of 19
Blogger Josh O said...

Ken's nonchalant breaking the tamper stickers warrants a "Deal with it" gif with the glasses sliding down over his eyes. Did you remove the service tag too? I hope not!

March 21, 2020 at 2:12 PM

Blogger Ken Shirriff said...

Josh, I left the service tag on rather than risk a year in prison. How would I explain that to my cellmates?

March 21, 2020 at 2:33 PM

Anonymous JRD said...

Ooh, a real 1970s/80s nuclear missile guidance package. Of course, this is the computer, not the gyroscope, but I wonder if the power supply has a constant frequency or phase-locked loop?

The reason I'm asking is that I immediately thought of of Hal Hardenberg's story "The Missile that Couldn't Fly Straight." It tells the hilarious story of why the difference between a phase-locked loop and a constant frequency in the guidance package can send a ocean-bound test missile "heading straight for Kansas City."

It's in issue #23 of his "DTACK Grounded" 68000 programming newsletter from 1983. (http://www.easy68k.com/paulrsm/dg/dg23.htm, scroll to "Page 17, Column 1") It claims to be fiction, but since Hardenberg had long experience in electronics and Southern California defense companies, I wonder.

March 21, 2020 at 2:42 PM

Blogger Ken Shirriff said...

JRD: I haven't looked at the Titan IMU (yet), but the computer's power supply is DC-to-DC, so frequency isn't an issue there. On the other hand, the 1201/1202 alarms that almost aborted the Moon landing were due to a phase difference between the computer's power supply and the radar's power supply. (Due to complicated reasons that usually aren't explained well, but CuriousMarc will make a video at some point.)

As far as the entertaining missile guidance story that you linked to, I don't know enough about guidance systems to tell how much of that is truth versus fiction. Autonetics did make gas-bearing gyroscopes for the Minuteman II, which matches the story's pseudonyms. But I couldn't find missile failures that match the story and the phase-locked loop business seems a bit implausible.

March 21, 2020 at 3:18 PM

Blogger Unknown said...

Have you ever designed a PLL? Doesn't sound implausible to me at all. Without the perfect loop gain, PLL's will lock to any old harmonic they can find in the system that is close enough to the target frequency. Sometimes you need a bit of synchronized noise in the system to make them lock. Finicky things.

March 21, 2020 at 6:07 PM

Blogger DHess said...

The VMX board looks like a set of Royer inverters which might be used to produce floating supplies.

March 21, 2020 at 10:29 PM

Blogger DHess said...

On second thought the transformers are missing a pin but they could still be multivibrator driven inverters which would make more sense for lower noise.

March 21, 2020 at 10:34 PM

Anonymous Corne said...

I suspect IOC stands for Input Output Control as it is the interface between the computer and the outside. This also explains as why it contains the clock as this is an input to the computer.

March 22, 2020 at 2:14 AM

Anonymous Anonymous said...

When you commented on the unit being pressurized with nitrogen, you said the fill valve was a "Schrader Value".....

It might have some value, but not much... ;-)

March 22, 2020 at 1:12 PM

Blogger Klimax said...

Hello,

maybe Bitsavers.org still have some Signetics catalogs unscanned.
(They already got some catalogs and other documentation scanned: http://www.bitsavers.org/components/signetics/)

March 22, 2020 at 2:35 PM

Blogger Unknown said...

You may be interested in Soviet space electronics. There are many photos on this site.

http://offtop.ru/dustyattic/v20_722878_24.php

March 24, 2020 at 7:40 AM

Anonymous Bill said...

The overall construction looks very similar to guidance equipment I worked with at Litton in the early '80s. We had all the ICs marked with our own part numbers, even if they were standard off the shelf military grade parts.
There is probably no existing record of what the parts actually are, unless maybe in a dusty file cabinet somewhere.

March 25, 2020 at 5:11 AM

Blogger ka1axy said...

I have some experience with "house numbered" ICs. There were two kinds that I know of. The first kind are parts that are made under contract for a buyer, who specifies the number to be put on the part. These can be custom parts, or identical to parts sold on the market, but the numbering on the parts is specified by the customer. Data General did this to prevent third party service companies from being able to repair their computers. Most of the DG parts are off the shelf TTL, custom marked with DG numbers. I used to have a list, but in a fit of house-cleaning, threw it out...regrettably.
The second type of house numbered marking can appear the same as a commercial IC. I once worked on an electronic tachometer for an MG. The part inside was a TI part, with a part number one higher than a tachometer part listed in their catalog. I called the sales office, and they said it was a custom, in spite of the numbering. They then went to the engineer's desk and pulled a couple of parts from his desk drawer and sent them to me! As far as I know, the tach is still working.

March 25, 2020 at 5:52 AM

Anonymous Anonymous said...

I suspect the Angstrohm module is a precision D/A converter.

March 25, 2020 at 6:06 PM

Blogger Unknown said...

Ken: The Apollo 11 1201/1202 alarms stemmed from poor integration. The rendezvous radar used resolvers to track it's orientation. The resolvers were referenced to a 800 Hz clock. When the radar was moved, pulses were sent to the Lunar Module Guidance Computer (LGC) to increment/decrement position.

The problem was the Lunar Module (LM) was built by Grumman, while the LGC was built at MIT. Each had their own reference. If the radar was in LGC mode, the 800 Hz reference came from the LGC and all is well. In AUTO or SLEW mode, the reference came from the Grumman system. The two references were not synchronized, and the LGC was always connected to the resolvers. Therefore, in AUTO or SLEW mode the LGC got a huge barrage of meaningless pulses; each pulse triggering an interrupt. Normally the LGC could handle the extra load, but in the computationally intensive landing phase it was enough to overload the system.

This was only noticed twice in about 100 pre-flight integration tests. It all depended on what part of the cycle the Grumman signal was in when the LGC powered on. Apollo 11 was just lucky enough for the two signals to be almost completely out of phase. MIT's George Silver had discovered the problem the year before, and wrote up a hardware fix to synchronize the signals. However, the fix was declined on grounds of budget and schedule.

March 26, 2020 at 10:40 AM

Blogger Electron12 said...

Really wonderful to have knowledge of This Computer , date back 70's
Thanks all

April 15, 2020 at 8:52 PM

Blogger Amaury said...

it sure looks like those boards were using vias. in your closeup with 3 chips, you can see lines stopping in the middle to what looks like a small circle.
also, there are plenty of those lines seemingly going nowhere, but you can make what looks like continuation lines on the next layer down, in particular they are very visible on this one
https://photos.google.com/share/AF1QipM5GGA8K7bCldhCmk1cb4VmuAQhkSuugSWaGfWxEBqAg0rjA6k-Gt4dJraT0bnZOQ/photo/AF1QipPEBaWIAhA19leUg6cH7E-DvNx3cY7dttN8SGxT?key=WktYZTdXZmlvZGw4MzhmZWV0eFhTdzNWbGNzTm13

October 12, 2020 at 3:30 AM

Anonymous Anonymous said...

>On the other hand, the 1201/1202 alarms that almost aborted the Moon landing were due to a phase difference between the computer's power supply and the radar's power supply. (Due to complicated reasons that usually aren't explained well

This is explained well on Don Eyles page: https://www.doneyles.com/LM/Tales.html

March 22, 2021 at 3:28 PM

Blogger FlorinC said...

This technical documentation will be very useful in a post-apocalyptic world.

March 23, 2021 at 2:54 AM

You can use some HTML tags, such as <b>, <i>, <a>

Comment moderation has been enabled. All comments must be approved by the blog author.

You will be asked to sign in after submitting your comment.
Please prove you're not a robot