987y or oiuy

[IanJeffray]

What's with the stupid subject title? Well, try typing either of those sequences rapidly on any (?) Archimedes.

On a selection of machines tested here - A3000, A3010, A420/1, they all show the same fault - that it's as if the Return key has been pressed!

I just tripped on this when repairing a faulty A30x0 keyboard, and thought the keyboard matrix must be at fault, but as it can be replicated on any machine, even an external keyboard, it's got to be a flaw in the 8051 keyboard controller.

I find it interesting because it's not related to holding multiple keys at the same time, it takes a rapid sequence of presses to trigger the issue. It's also interesting because I've used these machines heavily since '87 and never noticed this sort of keyboard issue before.

I'm just amused, and now forewarned. (And have an A30x0 keyboard that still has a duff '6' - grr!)

[steve3000]

I'm sure I've seen something like this before on an Archi but not narrowed it down...However I've spent last 5 minutes banging oiuy and 987y at varying speeds into my A4000 fitted with an A310 keyboard, and no unexpected returns... just a long line of 987y987y987y987y... or oiuyoiuyoiuyoiuy...

I wonder if this only affects the later membrane keyboards, as my A310 keyboard is the older fully switched type?

[richw]

This reminds me of the Chocks Away instructions. They said something about newer Archie keyboards not picking up the keypresses properly in two player mode, so there was a utility or recommendation to change the keys.

[IanJeffray]

I wonder if this only affects the later membrane keyboards, as my A310 keyboard is the older fully switched type?

Archimedes keyboards were never "fully switched" - the A310-style keyboards are just foam pads on to a PCB - with the same 8051 controller in use. (And still feel great after 30+years of abuse!) I've definitely replicated the issue on these. You do have to be pretty rapid with the typing - I first noticed it when just quickly dragging my hand over the keys to check they all worked, and then narrowed it down to the specifics of four keys in sequence. It's obviously no issue in real use - I was just amused by the hidden bugette and I suspect there'll be more key groups that act this way.

Crappy video demo : https://youtu.be/BLKsycW3J0Y

[steve3000]

Crappy video demo : https://youtu.be/BLKsycW3J0Y

Ah ok, that's a wee bit faster than I was banging them. Will have another bash tomorrow

[philpem]

This sounds a lot like key matrix ghosting.

What happens if you press the keys in the sequence together, in pairs or sets of three?

I'm not entirely sure the 8051 keyboard processor is smart enough to do key roll-over ( https://en.wikipedia.org/wiki/Rollover_(key) ).

Speaking of which... does anyone have one of those 8051s kicking around? I'd like to see if it's possible to dump the ROM, for preservation purposes. On some of the machines it's quite close to the battery.

[IanJeffray]

This sounds a lot like key matrix ghosting.

What happens if you press the keys in the sequence together, in pairs or sets of three?

I was unable to invoke the issue by anything other than the procedure described - I was quite surprised that the specific sequence and timing was important, which is probably why it's not a real issue

Speaking of which... does anyone have one of those 8051s kicking around? I'd like to see if it's possible to dump the ROM, for preservation purposes. On some of the machines it's quite close to the battery.

I can pull the one from the A3000 I've just about finished poking. I have a TL866II here, but the software doesn't seem to know about 8051's or anything similar. I fully expect all A30x0 ROMs to be the same, and all A5000/A4xx/A3xx ROMs to be the same - but the A4 is certainly going to be different as it supports PS/2.

[steve3000]

I still haven't been able to replicate this on my original A310 keyboard, after multiple repeat attempts.

However, I dug out my A5000 keyboard this morning, plugged it in and with this I can reproduce the phantom return effect pretty much every time. So I do think it's down to the membrane keyboards.

Tapping a few more random combinations of keys on the A5000 keyboard found another sequence producing the phantom return: tyui

...and a sequence which produces an alternative incorrect character: 7ytr produces 7yur

I've switched back to my A310 keyboard now, and tried one more time - but I can't reproduce the above...

[IanJeffray]

I still haven't been able to replicate this on my original A310 keyboard, after multiple repeat attempts.

However, I dug out my A5000 keyboard this morning, plugged it in and with this I can reproduce the phantom return effect pretty much every time. So I do think it's down to the membrane keyboards.

Well well well well well. I didn't know what you were saying there. Then I investigated. This was a big surprise. And many lessons learned:

1) This phenomenon only affects A30x0 and A5000 keyboards. NOT A3xx/A4xx with the shorter curly leads.
2) A5000 keyboards are the same (exactly?) as A30x0 keyboards internally. Wow. This was a real revelation. I wonder why they bothered making this change?
3) Never post anything on the Internet - someone will always point out it's wrong. (Never heed anything you read on the Internet - it's all wrong). [Always post rubbish on the Internet - someone will always come along and correct your stupidity for you. ]
4) Be more precise in testing / understand the test subject more completely. Post less casually.

Some pointless photos of an A5000 keyboard guts (top) vs A420/1 (bottom). I've opened DOZENS of these Axxx keyboards over the years, to clear out the gunk that gets between the PCB and the foils - I'm almost boggled that I never actually opened an A5000 one until today.

Apologies for wasting your time.

[BeebMaster]

This is probably a fluke, because I could only get it to happen once on a Master with Native ARM running:

I'm pretty certain I didn't stray onto the 9 key. The K is much further away of course.

[BeebMaster]

Just did it again... 7Y9, or maybe 97Y9, puts in a K occasionally.

[IanJeffray]

...and because Phil asked, here's the eprom contents. It's a 8031 on the A4xx - with separate Eprom. (NMC27C64Q)

A4xx_Kbd_NMC27C64Q.zip

(1.62 KiB) Downloaded 18 times

[IanJeffray]

Just did it again... 7Y9, or maybe 97Y9, puts in a K occasionally.

I have opened a can of worms. I'm sorry

[steve3000]

Some pointless photos of an A5000 keyboard guts (top) vs A420/1 (bottom). I've opened DOZENS of these Axxx keyboards over the years, to clear out the gunk that gets between the PCB and the foils - I'm almost boggled that I never actually opened an A5000 one until today.

It's one of those not-so-obvious changes that you'd never expect unless you'd opened the right type of keyboard.

The A4000 keyboard is also the membrane type, as is the A540's keyboard (at least my A540's keyboard is...). I suspect the change to this type happened around the time of the A3000, which makes sense (change manufacture process so only one type of keyboard is needed for all computers).