“Space age electronics”: Inside a GE thin-film paperweight from the 1960s | Hacker News Reader

“Space age electronics”: Inside a GE thin-film paperweight from the 1960s

righto.com

151 pointsparsecs33 comments

Author here if anyone has questions...

I love that it is a literal paperweight and not simply a relic. Thank you for the fun read.

Thanks for the great write up!

Any chance this thing still works? I imagine it may require some modification either way because of the oldschool battery.

The paperweight worked after I fixed a broken wire and powered it from a power supply, and it generated the beeping sound. However, after I re-assembled it, it switched to an annoying fast-beep. I don't know if a component failed or I messed something up during assembly.

You can hear the beeping here: https://twitter.com/kenshirriff/status/1411032440534028289

>I imagine it may require some modification either way because of the oldschool battery.

It's a 22.5v ANSI/NEDA 215, and they still make them. Alkaline now instead of Zinc-Carbon. They are kind of expensive, at $20+ USD, but available.

A great "tear down" of a really interesting device. I'd be curious how you ended up with this?

I'm sceptical that the satellite is Ariel 1 or Ariel 2 for two reasons: (a) Ariel was the British satellite programme, and I think that General Electric would have based the design on an American satellite. (b) Ariel 2 was build by Westinghouse, a rival to GE. My guess would be that it's a generic design, but I would love to hear an update if you find out for sure.

(I'm interested in the Ariel satellites, particularly Ariel 3, since I've ended up with a few of the original Ariel 3 telemetry tapes from the STADAN tracking stations. Still trying to work out a viable way to read / understand them...)

The paperweight was loaned to me for analysis.

It's pretty cool that you have Ariel 3 telemetry tapes. I agree that the satellite in the paperweight isn't Ariel, but Ariel was the most similar satellite I could find. I'm hoping someone will find a satellite that is a better match. But it could be some obscure proposed satellite that was never built.

What kind of physical format do those tapes have?

They are half-inch wide 7-track tapes. It's the 7-track bit that makes them difficult to read (I have a half-inch machine but it's 8-track). I think it was NASA and a few other government agencies using this 7-track format (or 1-inch 14-track), rather than the 1/2-inch 8-track / 1-inch 16-track used in audio studios.

The workflow for these tapes is described on the Chilton Computing website [1].

(Note: this is analogue 7-track, not to be confused with the digital 7-track format used by early IBM mainframe machines).

[1] http://www.chilton-computing.org.uk/acl/applications/satelli...

There is a Teac 1/2 " that does 8 tracks, you could probably read them one track at the time by judiciously jiggling the height of the head relative to the base of the player. The heads are narrower than a single track so that should work. The analog format is likely simple FSK or some variation on that theme.

It would take 7 passes to get it off assuming only one of the heads lines up with a track at any given time. Keeping the sync between the tracks will be the hard part. I'm in Europe so that probably rules out collaboration on this but if there is some way to get me a sample of such a tape I'll be more than happy to give it a shot.

edit: this machine: https://soniccircus.com/product/teac-80-8-1-2-8-track-analog...

The machine I have is very similar to that (Tascam 38), although the Teac 80-8 has the advantage of being able to run at the correct speed (7.5ips while the Tascam 38 only runs at 15ips).

From reading the papers published at the time (mid-to-late 60s), it seems the encoding scheme was based off a comb-filter, although I don't honestly understand enough to actually describe how it works.

I've had the tapes for around 5 years now, and managed to do basically nothing. It would be good to finally make some progress! I'll send you an email.

The speed isn't that much of a problem, it can be fixed after the signal on the tape is digitized.

Looking forward to your email, I love these kind of projects.

Could you record a clip of the sound?

You can hear the beeping here: https://twitter.com/kenshirriff/status/1411032440534028289

Cordwood was mentioned here, a couple of weeks ago: https://news.ycombinator.com/item?id=27531379

The one in the photo in your article has a much denser collection of components in it than the cordwood gate that I remember. There was a lot of air in that gate.

Impressive how little that battery leaked over all those years. And funny how the transistors have an extra wire coming out of them, but those are all paired and run to the same terminal, I wonder what's up with that.

The beeper is one of those crystal (piezo) earphones with part of the bit that you would stick in your ear cut off.

The two wires are probably for reliability and/or current capacity.

Right, so they're either emitter or collector. Base would then be the single wire left over and the counterpart the big terminal bonded directly to the thin film circuit.

Beautiful bit of history, thank you very much for taking the time to do this one.

But matching the schematic to the picture, the double wires are on the base. It makes me think it's a double-base transistor, but both bases tied together because they don't need two for this circuit.

If it was for reliability, I'd expect the emitter to also be doubled, but it has a single wire.

GE made the 3N36/37, but here's a datasheet for a similar part: http://bitsavers.org/components/ti/discrete/transistors/3N34...

That means it is definitely not for current. The current to the base would be negligible compared to the EC current.

It would have been a zinc-carbon battery[1].

I'm also very surprised that it's not corroded.

[1] https://en.wikipedia.org/wiki/Zinc%E2%80%93carbon_battery

Impressive how little that battery leaked over all those years

It was very likely one of batteries where each 'cell' looked like a block of compressed charcoal, about an inch or two square, with a strip of plastic around the edge. When I was a kid back in the 1960s, we used to pull these apart and throw the cells around like stones.

The most common size radio battery then was a 45 v battery, about 7" x 5" x 2" (IIRC) with a small socket in the top of it. https://www.flickr.com/photos/unclegal/31122957375

I’m frankly shocked that a battery that looks like it is from the 60s still has charge enough to run a beeper.

The battery was dead and bulging open. I powered the paperweight from a power supply.

Why gold wires instead of copper?

Gold wires were used for bonding in integrated circuits too. I think because gold is much more ductile, so you can make thin wires that won't break. It's also more resistant to corrosion, which is a problem with copper wire.

Doesn't copper interact badly with silicon?

I don't know about copper-silicon interactions, but normally you'd bond to a metal pad, not to silicon directly. For a while, the "purple plague" was a problem when gold bond wires reacted badly with aluminum bond pads, forming a purple compound. https://en.wikipedia.org/wiki/Gold%E2%80%93aluminium_interme...

The fact that it is more ductile probably also helps with the welding process.

A lot of high performance chips like hybrids (the chip, not cars) are gold bonded. Mainly because its a good conductor and super inert (metal behave funny in contact in space, oxygen is earth's "memory" after all).

> oxygen is earth's "memory" after all

Could you explain what you mean by this?

They may be referring to ‘cold welding’. Oxygen lets metals ‘remember’ what they are connected to. References:

- https://en.wikipedia.org/wiki/Cold_welding

- https://youtube.com/watch?v=Y2nQ8isf55s

Makes you wonder what's possible in 2021 in the age of light sails and uncommoditized nanoelectronics.

What is SOTA for 2D semiconductors (e.g. MoS2 with bismuth) in production?

Reply on news.ycombinator.com