8-bit acorn hardware • Re: BBC Plug lopping

- just how many Beebs can you power on a B type RCD?

Slightly fewer than Acorn had on their stand at the 1982 Personal Computer World show!

I got to 15 and the power went off, well below the current trip level, but above the RCD trip level. Of course, some BBCs were worse than others in this respect.

Then I got into measuring the earth loop resistance from the chassis of the PSU to the earth in the plug - 4 ohms - that's too high. So I picked another at random, 2 ohms - then I picked one where I had to rewire the PSU - 0.02 ohms...

Are you theorising that these have deteriorated with time and we should therefore be testing/fixing this as routine maintenance?

500uA across 4R is only 2mV so not a great concern from the Econet point of view. But 230V across 4R is 57A, so not adequate for safety in a circuit without RCD (won't trip 32A breaker fast enough).

It is not a theory, I cut open the pins on a couple of the plugs with poor readings and they are crimped internally, there was corrosion on the crimped area and the copper wires were loose.

I did notice that there was some current flowing in the ground of the Econet wire and I suspect this is the reason. I wonder how much current has to flow through the Econet ground before something gives up? I do have a large isolation transformer (3kW continuous load), but it's so big and heavy and I don't like it as part of "normal" use (builds up a good bicep though)

I think the isolation transformer is more likely to cause trouble than help. If you ground the neutral on the isolated side, you are in exactly the same situation as the original setup. If you don't ground the isolated side, then it's unacceptable on safety grounds for a large/permanent setup: it becomes safer on first fault, but you no longer notice a first fault, and a second fault then has no protection for the users (so fault monitoring is mandatory). Even if you ignore that, I'm not sure how much advantage it gives as the L/N will still not be isolated from ground - all those filter capacitors in the BBCs (or anything else plugged in) will tie the L/N to some voltage relative to earth - maybe 115:115V if it's all symmetrical.

Yeah, the isolation transformer is only something I use for testing purposes. I had the neutral and earth linked at the isolated end and that only was "safe" if everything was on the transformer. That stops the nuisance trip of the RCD, but does create an issue if something is connected to the real ground.

The actual problem with Econet is usually not so much the actual current flowing through that GND but the resulting voltage when it isn't connected (and the pulse of current when if you do then connect it).

Experience BITD was that the 75159s are fairly tolerant of steady-state conditions, but three scenarios caused problems:

1. individual BBCs with no connection to mains earth. These work OK if you have good earth continuity in the Econet cable, but as soon as you plug in a machine that's already powered up and the Econet earth doesn't necessarily contact first, you are now discharging a substantial capacitance that's been charged to 100V or so through the protection diodes of the line drivers - so much more than that steady-state 500uA. This scenario was rarely well documented BITD, but I suspect it accounts for a fair number of the 'random' failures observed. Dodgy 4-way extension leads were probably a prime cause.
2. Whole groups of BBCs in an area with no mains earth, connected to other parts of the network that are earthed. Now you have very substantial current in the Econet earth, a permanent difference of several volts at low impedance between parts of the network, and hence significant currents when the line drivers turn on and try to drag the line up to their local 0V. One I personally debugged was Acorn User magazine's offices in Long Acre: this was wired with steel conduit providing the earth for the 13A sockets, and the whole of one side of the offices had no continuity to the sockets on the other side, with obviously about half the BBCs plugged into each. Fixing that with a heavy bond wire joining the mains socket earths took them from blowing up 159s on a regular basis to none at all. There were lots more cases that I was aware of, but did not personally attend, lots more similar examples; there was for example a school in west London somewhere that had built a new portakabin-style computer room, and that room was not earthed at all - reportedly sparks flew when they connected up their Econet cable to the rest of the school. In all these cases, there was a blatent fault in the electrical system, and fixing that made the problems go away. In later days, we would always check out the mains earthing before starting an install.
3. Multi-building installs, even where everything was correctly wired, in the event of thunderstorms. Although under normal conditions the mains earth and Econet earth are all connected together, with the extremely high currents flowing in the ground, and the buildings anchored to their particular piece of ground, significant voltage differences appear momentarily between buildings which the cables are unable to control and again the line drivers take the strain. These incidents tended to blow dozens of line drivers in a single event. Knowing what I know now, opto-isolated bridges would have been the solution, but we didn't realise that at the time. Attempts to fix it with transorbs (the SJ "zap box", formally "Transient suppressor box") were ineffective, and may actually have made things worse given that the fitting instructions told you to obtain a solid earth such as an earth rod.

Transorbs mounted locally to the machine might help - but they are only going to be effective if all machines have them (they will otherwise just deflect the pain to other machines, forcibly anchoring your D+/D- within range of your machine's earth and so putting the whole of any voltage difference across whatever other machine(s) are plugged in).

An opto-isolator is an interesting idea, the clock is unidirectional and the data bi-directional. There's a TI equivalent using some magic that doesn't require an opto-isolator https://www.ti.com/lit/ds/symlink/iso1500.pdf Using fibre optics is another possibility, looking at fibre solutions for RS485/RS422 they're not cheap!

Statistics: Posted by maniacminer — Wed Mar 26, 2025 2:52 pm

---