[vcf-midatlantic] vacuum tube computers, was Re: Museum Report 2017-11-26 November 26, 2017

Dave Wade dave.g4ugm at gmail.com
Mon Nov 27 04:41:41 EST 2017


> -----Original Message-----
> From: vcf-midatlantic [mailto:vcf-midatlantic-
> bounces at lists.vintagecomputerfederation.org] On Behalf Of Dave McGuire
> via vcf-midatlantic
> Sent: 27 November 2017 01:51
> To: Jeffrey Brace via vcf-midatlantic <vcf-
> midatlantic at lists.vintagecomputerfederation.org>
> Cc: Dave McGuire <mcguire at neurotica.com>
> Subject: [vcf-midatlantic] vacuum tube computers, was Re: Museum Report
> 2017-11-26 November 26, 2017
> 
> On 11/26/2017 04:54 PM, Jeffrey Brace via vcf-midatlantic wrote:
> > The parents love the explanation of the progression of technology. I
> > need to learn more about how vacuum tube computers work to give them
> a
> > comparison.
> 
>   It's important to realize that there's really no difference.  Vacuum tubes are
> analogous to FETs (field-effect transistors) and are simply the active
> switching element of tube-based digital computers, where transistors are
> the active switching element of solid-state computers.
> The logic...the instruction set, buses, etc...can be identical.  The only real
> reason they're not is because computer architecture has grown up alongside,
> but largely independent of, electronics technology.
> 
>   Now if you're talking about tube-based analog computers, that's a different
> animal...and the same point applies there.  Analog computers are (mostly)
> composed of op-amps, and those op-amps can be built with tubes or
> transistors. (with or without multiple transistors being "integrated" into one
> "circuit", i.e. using integrated circuits)  Again, the architecture is (or at least
> can be) the same.
> 
>   The takeaway here is the importance of understanding the difference
> between architectures and implementations.
> 

So to take the simplest Valve computer, perhaps the Baby or SSEM that Kilburn and Williams built in Manchester, in architectural terms its architecture is pretty standard

32-bit words
Single Accumulator
Fixed instruction formats, one address, one op code


Two quirks:-
The conditional test only branches to "+1" or "+2"
No ADD, only Load Negative, Subtract and store

Its implantation is however unique or at least unusual to modern eyes in many places. 

The adder and subtractor are implemented as Kirchhoff analogue components. You can see the circuit of the subtractor here:-

http://www.cs.man.ac.uk/CCS/SSEM/volunteers/subtractor.html

as this allows the adder/subtractor to be built using fewer valves than an entirely digital line. In the same way several of the clock dividers use phantastron dividers

http://www.cs.man.ac.uk/CCS/SSEM/volunteers/clock1x2split.html

also described here:-

http://www.radartutorial.eu/17.bauteile/bt52.en.html

which were originally developed for Radar (Williams and Kilburn were Radar experts who built the Williams tube because they wanted to electronically process Radar Sign

but again, analogue circuits adapted for digital use, probably because Kilburn and Williams were familiar with them, but also they used fewer valves that the equivalent digital circuits...

>                -Dave
> 
> --
> Dave McGuire, AK4HZ
> New Kensington, PA

Dave G4UGM
Manchester, England
Now without gall bladder so may be able to get affordable travel insurance for USA




More information about the vcf-midatlantic mailing list