Pericom keyboard

[MMcM]

This keyboard is from a Pericom graphics terminal.

Pericom-case-top.jpg (398.29 KiB) Viewed 3096 times

It's possible that its model number was 2506. Or maybe that's just the part number for the plastic top of the case.

Pericom-case-back.jpg (343.58 KiB) Viewed 3096 times

The bottom of the case is sheet steel, so the keyboard weighs almost 2.5kg. Plus it has a huge thick coiled beige cable, as one did.

Not much info has survived about these. It has the same distinctive grillwork as the one shown on a Pericom (UK) anniversary page, but even in that not very good photo one can make out a seventh row for the right sections, so it's not the same.

https://chambermk.co.uk/news/milton-key ... nniversary

It was made at the beginning of 1990, but I am going to say it's still from the '80s. The layout is basically the VT-220, that is, LK201. And many others, such as Wyse, copied this layout around then. But the LED placement and layout is more like the VT100, with an L5 added. It swaps which of the function keys have P in front, perhaps because more of them really were programmable. Some of the dye-sub key caps have second function legends in red. I imagine the simplified shape of the RETURN key was a restriction of what was available.

The switches are ITW low-profile. The seller didn't bother to take a photo with a key cap off. I suppose in fairness you need a tool for that. But they did show the PCB itself and four contacts per switch and an LM319N voltage comparator over in the decoding section is kind of a magnetic valve give away.

Pericom-PCB-top.jpg (480.06 KiB) Viewed 3096 times

Pericom-PCB-bottom.jpg (409.32 KiB) Viewed 3096 times

The keyboard has a DIN 41524 connector. There are a few protection diodes in there, but I still bet powering like an XT would be a really bad idea. I admit that was probably a lot less likely when it was designed.

After figuring out which of the four signals was which, the output is clearly idle-low 1200 baud serial. It sends a unique 7-bit code for each key with the 8th bit break versus make, which is nice. Only special case is CAPS LOCK, which maintains the LED locally and sends make and break on alternate key presses and nothing on key releases, so that it acts like another shifting key. This can be quickly turned into a QMK converter.

Pericom-converter.jpg (291.35 KiB) Viewed 3096 times

Which leaves input to the keyboard for changing the LED row and ringing the bell. The circuit is about as simple as could be: state comes from DB0-7 on the controller through a 7406 inverter. Which means that all the LEDs come on and it squeals loudly for a long time when powering on until that bus can be set high. There are actually only 7 LED signals: On Line and Local are complementary. I just imagined the simplest thing the 8748 could do, namely when T0 -- where the input goes -- sees a start bit, read eight more of them and set that as output. And that was exactly right. There is no more protocol. Low 7 bits for the LEDs in order (1 = on) and high bit turns on the transducer until you clear it.

[EnthusiastDude]

That's an interesting keyboard, thanks for taking the time to share your findings. Do you have an opinion on the switches themselves?

[MMcM]

Do you have an opinion on the switches themselves?

I think I like something with a little more travel. I've got an older ITW in the quarantine queue (that's 75% heavier) that I may end up liking more.

The ones on this board also feel a little scratchy. I don't know whether that's because they need a more thorough cleaning or the plastic itself is doing something.

[EnthusiastDude]

Do you have an opinion on the switches themselves?

I think I like something with a little more travel. I've got an older ITW in the quarantine queue (that's 75% heavier) that I may end up liking more.

The ones on this board also feel a little scratchy. I don't know whether that's because they need a more thorough cleaning or the plastic itself is doing something.

Is the travel noticably shorter than Alps and MX? Please share the older ITW too, there aren't that many opinions/reviews of these boards and I find them rather interesting. I've been on the hunt for ITW for a while but these seem to avoid me

[DMA]

Tools are not needed to pull the cap off from a standalone row. In low-profile ones you'll pull out more than a cap though (easy to push back in, so no worry), and in full-height ones you risk breaking the cap's stem (fixable by putting all pieces back and filling the cap with poxypol).
Good thing yours come with mostly-sensible protocol - older ones usually only provide make code for alnums, or don't provide any codes for modifiers, sending different make codes depending on modifiers' state.
In that case MCU replacement might work if PCB is intact: no water damage (water creates relatively low-resistance paths between pins), no trace corrosion (higher resistance of inter-switch paths) - scan pattern is relatively simple (time-sensitive tho).

If PCB is screwed - be VERY careful when desoldering switches, unless it's a full-height ITW. Those ferrite cores are _very_ fragile, make sure all pins moving sideways freely before you pull.