_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I've noodled ideas for relay logic in the past (I was trained by the BBC, and an awful lot of signal switching (1970s) was done with 48v relays) and I see that relays in small bulk are available for around 20 cents each for SPST types. It's always a trade-off between voltage and current - most seem to be around the same power (for similar types) so I'd probably work with 12v types to keep the current manageable.

A little lateral thought might tempt you to make a clocked system which has a higher voltage at the start of the clock pulse and then a lower holding voltage, to save power - in many cases, the holding voltage is only about 60-70% of the pull-in voltage. Designing such a circuit might be, um, interesting...

The other main thought: do you use a single type of relay throughout (spst?) or do you experiment with dpst (or more poles) or with latching relays? Do you use a single logic 'gate' type, or do you get creative with routing voltages (and directions!)? The opportunities are endless.

Neil

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Yes, that's always the trade-off - simplicity against efficiency.

Neil

Do you have a link? The cheapest I have seen is $0.99. I have enough for this project, but if it goes well I might do something crazy (e.g. The world's worst math coprocessor for the 6502). Even a modest decrease in cost would make it more approachable.


Most of the relay projects I have seen online standardize on a single relay to ease PCB design and construction. That way you don't get confused and wire something wrong because the relay had a different pinout. Latching relays would seem a natural fit for building registers, but I haven't seen them used.

Harry Porter's machine standardized on four pole double throw relays because he could use two relays for the enable logic to control the bus. He also had a standard wiring that allowed a single relay to have multiple logic function outputs. I imagine that came in handy because sometimes it's handy to have both the AND, NOT, and OR simultaneously available.

True story. I discovered his machine because of a typo. Years ago I was trying to search for Harry Potter, and replaced a t with an r. I then spent hours studying his machine, and never redid my original search. I emailed him and he got a laugh out of it.

@Garth, wow, that capacitor coupled with a resistor is a really flexible circuit element.

Try this: https://www.lcsc.com/product-detail/Sig ... 02426.html

There are other voltages in the same series.

Neil

@Neil, thanks. Wow, I thought I had a good price, but I could have brought over a hundred for what I paid.

I have been reading the design documents for the Mercia computer and it is interesting. He has a diode ROM, whichever isn't too crazy. But he has two types of RAM, static RAM built from relays for page zero, and dynamic RAM using capacitors with relays to charge and read values. The later is really interesting as I have never heard of it before.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Core memory is still available and the Mercia design docs say he considered it. But the circuitry requirements were too complicated.

As I understand it, core memory has a destructive read, so a write back after read is required. The process of reading and writing is also difficult because it is essentially a form of magnetic media.