Inside a Titan missile guidance computer
righto.comThis is another good read about using automated terrain mapping for cruise missile guidance systems.
https://en.wikipedia.org/wiki/TERCOM
Notably how it fit on older computer systems with limited memory by only sampling the terrain using a predefined path which a plane will travel over with imaging sensors beforehand. For minutemen style systems with predefined targets it would work well as the terrain won’t change significantly and there will be plenty of sampling data to support a high error rate and/or a relatively low resolution while moving at high speeds.
Of course it increased in quality and effectiveness over time (especially as whole topographical maps could eventually be stored, not just the path in question) but as an entirely ‘unguided’ rocket it seems to be relatively accurate for its era.
It knows where it is because it knows where it isn’t.
Isn't it an example of a highly guided missile?
Yes that’s correct, I meant compared to more controlled guidance by wire, radiation, satellite, etc. I phrased that improperly.
Fun tangent - the most unguided missile ever.
> Lethal radius of the blast was estimated to be about 300 metres (980 ft). Once fired, the Genie's short flight-time [12 seconds] and large blast radius made it virtually impossible for a bomber to avoid destruction.
Why can’t you dive or climb just 150m in 12 seconds and be out of the blast range?
I suppose you have to be sure of the rockets detonation point to know if you are moving away in the right direction. I'm not sure that is a thing you could be sure of the moment the rocket is fired (or if you would know).
Also the target (at least initially) was large formations of Tupolev Tu-4s who can only climb 15m/s. So that puts them right on the edge, but assuming more than one rocket is fired or that they simply don't always have time to realize it was fired quickly enough... that probably does it.
Why 150m? It says 'radius' so if the missile is on an intercept course, 150m is not going to help you. Beside the difficulty of knowing the exact launch time and having to move a 50mx50m 180 ton bomber in fewer than 12 seconds.
Fighters and interceptors can, bombers can't.
300m seems like a very conservative estimate for an airburst nuke.
I visited the Titan Missile Museum in Tucson, AZ, where you can see the control room from which the missile could have been launched.
> At launch, orders from the National Command Authority would have specified one of three pre-programmed targets which, for security reasons, were unknown to the crew. ... Target 2, which is classified to this day but was assumed to be within the borders of the former Soviet Union, was designated as a ground burst, suggesting that the target was a hardened facility such as a Soviet missile base. [Wikipedia]
While explaining this, my tour guide showed some punched tape that could be fed into a computer to program the targets. I wonder if the tape was original, and if so, how difficult it would be to decode its coordinates.
> I wonder if the tape was original
I'd be very surprised. We used punched tape for initializing our crypto gear when I was in the Air Force, and it was accounted for carefully and destroyed promptly. Even more carefully than our regular classified documents, though maybe that was just a culture thing. I can't imagine them just giving it over to a museum.
Yes, the Titan museum is very interesting and I recommend it.
I wondered the same thing about the punched tape, if I could extract the coordinates from it. I read somewhere that they were extremely strict about keeping the targets secret; before any maintenance on the computer, a special team came in just to ensure that the coordinates were erased from memory. It would have been a huge oversight if they didn't encrypt the punched tape, so my guess is that it can't be decoded.
Y'all are overthinking it- true lat long "coordinates" would be meaningless to this computer. What was fed into the computer (ahead of time) was a series of firing instructions (per target for 3 targets), calculated to take into account the launch site of each missile. (Core memory values persist long term, even through power outages, hence the need to manually erase values before maintenance.)
So even two missiles/warheads launched from different sites at the same target would have different guidance instructions.
You'd need to do some major reverse engineering and geodetic calculations even after decoding targeting punchcards to figure out what the actual targets were.
The punchcards used to select the target during launch are simply instructions to execute "target stored in memory 2" or something similar. The minutemen would not have had the ability to change the set of targeting options programmed in the missile, only select one of the pre-programmed options.
If you're interested in the history of the Titan missile program and US nuclear program in general I recommend "Command and Control: Nuclear Weapons, the Damascus Accident, and the Illusion of Safety".
I definitely agree with your recommendation. Much of that book describes the Titan missile that blew up in 1980 after someone dropped a wrench socket that punctured the rocket's fuel tank, throwing the 9 megaton warhead 100 feet. (The warhead did not detonate.) What I didn't realize until I visited the Titan museum near Tucson is that it wasn't a regular wrench socket, but a big, fist-sized 8-pound socket. This made it much more understandable how the socket put a hole in the missile.
American Experience (PBS) did one of their documentaries on this event. It is very well done.
Prior to the socket being dropped, there was a change order issued to use a new type of socket that would not dislodge from the wrench. As I recall, the two technicians tasked with the repair didn't have the new type of socket, but proceeded into the silo anyway.
https://smile.amazon.com/dp/0143125788/ (Amazon)
https://archive.org/details/commandcontrol00eric (OpenLibrary)
The History Guy did a great video on this as well: https://www.youtube.com/watch?v=jDcog2ZP684
The fascinating thing is that the Titan was completely inertially guided, i.e., no GPS or ground-based navigation. The high precision IMU + gyro send the computer acceleration and angular velocity and you integrate that over time to figure out where you are now. I'd think this would be very challenging and error-prone for inter-continental travel especially when the thing is planning to detonate nuclear war head. I wonder what algorithms they used. Typical code to do dead reckoning from IMU+gyro is fairly short but drifts usually is huge killer.
"Inventing Accuracy" is a fascinating history of ICBM inertial guidance. https://mitpress.mit.edu/books/inventing-accuracy
GM/Delco were very diverse in manufacturing, so I'm not surprised that they also made missile components at one point. At one time they owned Frigidaire too.
Motorola made "MTTL II" which had 20xx part numbers for commercial temperature range and 21xx for industrial/military, so these could be Signetics' variant of that. In the late 60s through the 70s semiconductor companies made lots of different logic families.
I've always found it a bit sad that the ultimate purpose of things like this was for destruction.
Thanks for the part number info; that's a good lead and I'll see if I can track anything down.
As far as your last point, I basically agree. This computer, though, is not quite as bad as it could be. It was a later, 1970s model originally designed for the Titan IIIC rocket to put satellites into space, a more positive role. It was also closely based on the Carousel IV commercial navigation system used for aircraft such as the Boeing 747, which is a nice heartwarming application.
Edit: the Motorola 21xx parts have the wrong pinout, e.g. ground on pin 10 vs 11. And many of the numbers don't exist as Motorola parts. Unfortunately, I don't think these are the right parts, although it seems so close.
I was just thinking, imagine trying to use a modern-day smartphone for this. So much complexity how could you be sure the missile would even get off the ground. There is a big advantage to systems that engineers can mostly "keep in their heads". Very few systems are that way these days.
A Teensy 4.0 just as an example which is much more like this. Has a bunch of general purpose I/O; it's quite powerful; and you can essentially keep it in your head.
I can't imagine trying to discover & isolate faults on something like this though.
THAT is not a computer. THIS is a computer.
I find it fascinating how much those grids resemble RAM chip die photos. I know the principle is different but in a sense we have just been shrinking things for over 70 years.
Of all the details, I particularly enjoyed the keying method with the half hex nuts. I'll file that away for future projects.
I'm surprised they opted to use a passively cooled power supply.
I imagine it was much heavier than an equivalent active system.
I guess MPG doesn't matter much on a Titan missile.
I'd guess the opposite, actually. Mass is extremely important for a rocket, so there would be strong incentives to make the cooling system as light as possible.
But a passive system has a few major advantages - it's maintenance free, unlike an active system (imagine having to fill up the coolant tank before launch), and perfectly reliable. Sure, a phase-change coolant might be more efficient - but with flight times in the hours I'm not sure that the difference would be enough to compensate for the complexity.
Once the liquid electrolyte was injected into the batteries and the missile powered up at launch time, it was only going to be operating for 30 to 40 minutes. So it didn't need a lot of cooling capacity.
Remember this system (including power supply) replaced a really ancient system that weighed twice as much. If anything, the heavier PS was helpful because it + the support truss brought things back to the mass that the missile was originally designed around. Otherwise it would be necessary to extensively retest an old missile with limited inventory, possibly even changing the guidance equations.
Great read! Intriguing to find out that earlier computers were using rotating drum memory systems in space flight and then delicate wiring in cores, I really hadn't considered that. I wonder about the reliability of those systems and whether there was software compensation for that issue.
Some engineers designed this - used big part of their life for the work and the system was never used.
Sure it was used--as a credible deterrent.
And we're all the better for it...
I love reading about avionics. New and old.
Always scares me to see this displayed for the public. Enemies watching.