An astonishing old calculator

Outstanding! Acoustical wire memory- I'd never heard of that and I'm an obscure tech geek.

I suppose you could call mechanical reverb a related tech. I wonder if that was an inspiration.

Mark L. Fergerson

Yes, that is a cool calculator. I had heard of delay line memory in some early computers, but I assumed that meant electrical delays like in a long coax, but maybe it was acoustic delay lines.

--

Rick C 

Viewed the eclipse at Wintercrest Farms, 
on the centerline of totality since 1998

Mercury was the first medium, because there was good acoustic coupling to piezo transducers.

0.25ton of Hg was used in EDSAC

So it acts like a serial shift register/circular buffer where the bits go around and around and around. What I wonder is how whatever has to read the bits out "figures out" at what position it's supposed to read data out?

I'm assuming it's only possible to read out from the start/end position of the line. Is there some kind of header sequence that triggers a counter so if you want say "address" 100 offset from the header it counts 100 bits and then reads out for X bits into a register?

How long could one store a pulse train in it for and have it re-amplified as it goes around the line before one started to get corruption?

There were also mercury delay line memories, and solid acoustic delay line crystals. And of course computers that used a magnetic drum for program and data storage.

Some early computer used a rotating drum with capacitors, each cap holding one bit, accessed by brushes.

Whirwind (I think) used charge spots on the face of a CRT as RAM, the Williams tube.

Some early video terminals used IC shift registers as the display memory, maybe out of tradition. I guess shift registers were cheaper than RAM once.

Then came core,

originally with tubes.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

CMOS bucket-brigade delays and bubble memory were variations on the theme

Then came the mines, then came the ore Then there was the hard times, then there was a war Telegraph sang a song about the world outside Telegraph Road got so deep and so wide?

ABC?

Many did, I guess it was fairly popular (but I don't remember them by name, sorry).

There were a few other CRT and tube related methods, which were tried but never caught on. Most notoriously selectron tubes -- too big, too expensive, too late!

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

Global timer clock I would think. Synchronous machine, just a few more cycles before it loops around. :)

It was amplified every cycle. Which would also be somewhat prohibitive to analog data, if you wanted to use it as a feedback reverb tank, say.

Now, I don't remember if it was unidirectional (data flows from one end to the other, gets recycled), or if they used a reflector and duplexed the piezo. Probably not the latter, that'd be tricky to get right? (I mean, it's just telephone circuits at the heart, nothing they didn't know back then -- but getting it /just so/ maybe wasn't so easy.)

I'm sad that there are so few design documents about these old machines. It's rare enough to see a logic gate schematic (that is, how a gate was implemented), let alone a technical overview. There are zillions of high level, introductory videos, of course.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

There might have been some in the HP archives, past tense. :-(

an

f

in some

Fascinating! Thanks, Lasse.

There is a lot of information available online.

The Hg delay lines were temperature sensitive, and it was a problem.

Both times I've been at TNMoC in Bletchley Park, while

*listening* to the first computer I used (a 39bit Elliott 803) I've talked to one of the loitering staff.

He's whipped out the circuit diagrams, and discussed them with me.

Now *that's* my idea of a museum.

J. V. Atanasoff, U. of Iowa, reported on one such computer, which had

32, 50-bit numbers per 8" diameter by 11" long bakelite cylinder. He also reported the names of grad students, who soldered in the 1600 paper capacitors, 0.0015 uF each. That was 1940.

It makes interesting reading:

an obscure tech geek.

f that was an inspiration.

g

used ultrasound propagating through troughs of mercury. As a beginning grad uate student. back around 1964, I used to take my programs (on IBM punched cards, to be taken down to the universitie's newly bought IBM 7040/44 which was then still down-town at the IBM office) to the CSIRAC room over in the Physics Department.

--
Bill Sloman, Sydney

A 16x19 array. Maybe an 18 bit word with parity.

I had to work on General Dynamics dynamic logic circuits. I'm thinking it used coils for delay. My memory is weak, but a pulse was a one and no pulse was a zero. Also had early computer crt terminals using some kind of rotating shift registers for memory.

Greg

In the British Elliott 803 of the 1960's, there were several registers built of nickel-line serial delay lines. The logic was also unconventional of today's standards: it used mostly ferrite-memory cores.

The delay line registers made the thing quite sensitive of master clock frequency.

--

-TV

--------------------------

** Almost certainly yes.

Hammond first produced a torsion driven spring reverbs (like modern ones) in 1960. These had magnetic torsion transducers at each end of long coil springs and permitted weaves to travel in both directions to get long reverb times.

The Friden calculator of 1963 used a long, plain steel wire the same way but likely permitted only one way travel.

..... Phil

Everything about the 803 is unfamiliar by today's standards! - 39bit words - 8kWords architectural max - bit serial - 576us instruction cycle time (2kIPS) - logic gates contained a pulse transformer - 5 channel paper tape - magnetic film storage (35mm wide with sprocket holes) - a groundbreaking Algol-60 compiler, by Tony Hoare

See one working at The National Museum of Computing, in the next room to the world's oldest working computer and a replica Colussus.

And discuss the 803's circuits with the attendant :)