East Germany, Beige Plastic and 2 5V pins [Solved]

[Aruberon]

Hi DT,

So in my closet lurks the East German Robotron K7673-02 which Ive done nothing which since I got, but always wanted to get some life into.

Allegedly these run XT with a slightly modified protocol so some keys might not work but it should be happy with a standard Soarer converter, so I made one in a pro micro and plugged it in, and he woke up:

photo_2023-06-21_18-41-37.jpg (94.57 KiB) Viewed 7345 times

So at this point, its all good, it even outputs commands, and the correct ones to boot.
Now this is where the problem begins:

There is a massive lagg between pressing the key to it outputting on screen, it also keeps outputting, sometimes it stops after 20 characters, sometimes its going to the moon.

Now this is where my inexperience in electronics and programming becomes apparant, because im stuck at this point and would hope that someone where with some experience with something similar can come to my aid.

My main lead at this point, thanks to KBDbabel for mapping the pinout:

kbd_connector_k7673-02.png (6.56 KiB) Viewed 7345 times

It has 2 5v pins, which im guessing is gonna be a problem on 1 USB slot.
Alternatively, would a pull-up resistor help? Troubleshooting ADB converters was a joy before I realised I needed one of those.

[jsheradin]

Try TMK on your converter before you get too much into it. The XT umbrella has quite a few quirks and a lot of boards are slightly non-standard. TMK has much better support for the weird ones than Soarer's.

https://github.com/tmk/tmk_keyboard/wik ... d-Protocol

[DMA]

It has 2 5v pins, which im guessing is gonna be a problem on 1 USB slot.

Not really. They are most likely connected inside the keyboard - there are two ground pins, too. Look at the other end of the cable, or simply check resistance between those pins - should be less than 1 ohm.

[Aruberon]

It has 2 5v pins, which im guessing is gonna be a problem on 1 USB slot.

Not really. They are most likely connected inside the keyboard - there are two ground pins, too. Look at the other end of the cable, or simply check resistance between those pins - should be less than 1 ohm.

Tested the two power pins and ground pins, both are connected with about 1.5 ohm of resistance. A bit higher than anticipated, are we still within a workable range?

I also tested running the keyboard on either pin as well, it didnt care if it was left or right 5v pin or the left, right or both ground pin, it still worked and behaved the same.

The main thing that bothers me with the dual pins though is its predecessor and successor board, using the same DE-9 connector, but only one set of each pin:

kbd_connector_k7672-01_03.png (6.26 KiB) Viewed 7257 times

I guess we can speculate on the reason why at this point and never get the answer, but there must be an underlying reason, no?

Try TMK on your converter before you get too much into it. The XT umbrella has quite a few quirks and a lot of boards are slightly non-standard. TMK has much better support for the weird ones than Soarer's.

https://github.com/tmk/tmk_keyboard/wik ... d-Protocol

Never worked with it before, but definately worth trying, I've never had issues prior converting some more off-brand XT boards with whacky keys that dont read properly, but there is a first for everything.

I suppose its worth to test the Promicro as well, while it seemingly runs, accepted its protocol, it was brand new and previously untested.

[DMA]

Tested the two power pins and ground pins, both are connected with about 1.5 ohm of resistance. A bit higher than anticipated, are we still within a workable range?

Pretty much. If you want to be double sure - open the keyboard and measure at the PCB (although at PCB end you'll likely see traces between the pins).

I also tested running the keyboard on either pin as well, it didnt care if it was left or right 5v pin or the left, right or both ground pin, it still worked and behaved the same.

Mmkay, that also works

The main thing that bothers me with the dual pins though is its predecessor and successor board, using the same DE-9 connector, but only one set of each pin:

I guess we can speculate on the reason why at this point and never get the answer, but there must be an underlying reason, no?

"Consistency is the hobgoblin of little minds".
most likely reason is 01/03 were interfaced with a different machine type, with a different keyboard socket pinout. DB-9 is, like, _the_ standard connector, after all, but pinouts are different because Reasons (serioiusly though - nope, not everything has an underlying reason. Basically, ganging pins gives you less voltage drop, but that's about it. 1.5 ohm is a bit too much for cable resistance - so it might be that computer for the .02 was using longer keyboard cable and they needed that extra wire cross-section.. or it was "just in case" because whoever specified the cable wasn't sure about power consumption..)

[Aruberon]

So opening the board put a quick end to the connector mystery, behold this end cable connector:

photo_2023-06-22_11-01-07.jpg (99.79 KiB) Viewed 7168 times

Im not sure if I trust the two capacitors either, they arent bloated, but the soldering looks stock, and they both have residue at the bottom:

photo_2023-06-22_11-07-04.jpg (88.47 KiB) Viewed 7168 times

photo_2023-06-22_11-07-04 (2).jpg (137.65 KiB) Viewed 7168 times

To me this looks more like leftover flux, but its only present under the two capacitors.

The resistance looks to be about 1 ohm exactly down here.

[Muirium]

Bear in mind that only (tube shaped) electrolytic capacitors leak. The ceramic ones don't. So there's only one suspect in this case: the tall fella in the middle.

If it is leaking, take it out. The residue slowly rots everything, and the capacitor itself will be worse and worse at its job.

[Aruberon]

Both of the suspect ones are of the electrolytic kind.

To rule out a few more things, I gave the promicro 2x 1k pull up resistors and connected it directly to the PCB instead of via the cable.
No change in behaviour so im thinking capacitor swap next.

That and giving TMK a shot

[Aruberon]

TMK loaded, identical behaviour, so its not the firmware

[jsheradin]

Pull up hid_listen with either controller firmware and see what is actually being sent over the bus. I've come across boards that simply don't send upstroke codes. Modern PCs interpret any press as a held key.

https://www.pjrc.com/teensy/hid_listen.html

[Aruberon]

Sooooo, in prep for doing a hid_listen I redid all the soldering, desoldered the original cable and permanently affixed the promicro to the internals and now, as Tod Howard would say, it just works.

Its still on TMK, nothing else changed since last time, its a bit confused about some of the keys but the vast majority outputs a signal.

[jsheradin]

Sounds like it was just getting starved of power and not getting a chance to send release codes. Glad it was an easy fix!

You can use hid_listen to see exactly what's going on with the wonky keys. You might be able to bodge it with some macro trickery. I've done similar on a board that had extra function keys. They weren't sent as the normal F13-F24 but instead as key combos. A macro on the controller made them behave as expected.

Both Soarer's and TMK support pretty powerful macros but Soarer's are generally easier to configure.

[Aruberon]

Luckily the majority of the wonky keys are down to the "special" XT protocol as the robotron forum hinted at, so using the TMK keymap editor, the vast majority of the keys have been restored to where they should be through trial and error, with a few exceptions:

ESC and Prnt Screen is tied and seemingly unseparable, changing prnt screen in editor changes both prnt screen and esc on the actual keyboard.

ESC in TMK keymap edit is actually grave,

Grave itself does nothing.

00 does nothing

SYS triggers the windows snipping tool

Ill probably look into hid_listen to try and figure out these keys at some point, but I think thats enough for now.

[Aruberon]

For the rest of them, grave was escape by default, tab was ctrl, ctrl (on the right) was shift, which is technically correct but in this case had to be swapped.

Having no native win keys, right alt got sacrificed and became a win key, so it was technically correct, but things are better this way and the key between backspace and enter was tab, its now pause play for conveniences sake instead:

dZRb8Sj.png (49.59 KiB) Viewed 6982 times