Issues with 2WASZ keys on a Model F AT (and cleaning)

User avatar
derzemel

07 May 2015, 00:46

I have managed to get an IBM Model F AT and I have an issue with it:
- the keys 2,W,A,S,Z do not work (connected to PS/2 with a DIN-PS/2 cable).

Do you have any ideas?

As a first idea I took it apart to clean it... maybe this would solve the issue.

Here is the Imgur album with the teardown and starting to clean it.

As a curiosity, it has flippers with 2 shapes. They are randomly placed:

Image


Can this foam mat be reused? I wasn't falling apart, but it is not very resistant either

Image

Image

Can I use 2mm thick foam rubber, like the one below, to replace the Foam Mat? If not, where can I find some Foam Mat?

Image

I checked the strip continuity (hopping that this may be the issue with the 2WASZ keys). none of the strip wires are interrupted:

Image

While Cleaning the PCB with Isopropyl Alcohol, I found 2 scratches:

Image

and

Image

In case the cleaning does not have any effect on the 2WASZ keys, do you have any ideas what may cause this issue?

Thank you!

P.S.: Tomorrow morning I'll finish cleaning the flippers and the barrels and reassemble it (as now is almost 2 AM)
Last edited by derzemel on 07 May 2015, 13:35, edited 1 time in total.

andrewjoy

07 May 2015, 01:17

It could have just been assembled wrong and the flippers may not have been aligned correctly, if its the stock controller it should be fine. They are all on one row however so it must be something to do with that part of the matrix. Trace out where all the ground points should be and make sure they are all connected, poor grounding on a model F can lead to all sorts of nasty stuff.

think it may be a trace on the controller somewhere that's not connected ( i known you tested the ribbon cable). Trace out the controller and see if there is continuity between the pcb and the correct pin on the IC

Failing that it could be something going wrong in the metal can ICs there is loads of analoug voodoo going on in there if i understand correctly

Is the scratch on the pad one of the ones that is not working ?

You can replace the controller with a modern USB alternative

User avatar
Muirium
µ

07 May 2015, 01:20

Good luck with reassembly. Hopefully everything will simply work again. You do know the trick to make the AT's spacebar stay seated, right? Here's my thread which may be useful as people helped me out there too!

http://deskthority.net/workshop-f7/open ... t9791.html

The second crack doesn't look anything serious. That first one though… hmm. If your keys still aren't working tomorrow, that's the place I'd look. Might need to bridge it. Not that I know how. Did you test continuity on both sides of the break?

Your foam looks fine, by the way.

Andy's got some good points. Careful reassembly can fix a bunch of old errors. But I've never seen leaky capacitors on a Model F (they seem to be a Model M2 problem, after many years of IBM going down market). What he means by a modern controller is likely Xwhatsit's Model F controller, which is a nice piece of kit but quite expensive. I doubt your stock controller is the problem anyway.

User avatar
derzemel

07 May 2015, 10:55

I checked the continuity on the both sides of the scratch and everything is ok.

I also checked the controller and the pins... some of them have continuity, some do not, but I hardly know what I am doing so everything may be ok with the Controller and pins on the IC

User avatar
Nuum

07 May 2015, 11:47

Try an active converter like the Blue Cube. I had problems with my AT on a modern PC's PS/2 connector as well, I think it has to do with power consumption.

User avatar
derzemel

07 May 2015, 13:19

I finished assembling it.

DanielT really was not joking when he told me that these things were build like Lego: everything snaps in place :))

This topic really helped: http://deskthority.net/workshop-f7/bm-a ... t9488.html

Sadly, the 2WASZ keys still don't work. I'll try an active converter, as several of you have suggested.

Until then, more images:

The metal plate, cleaned with Contact Spray (to remove a few small rust spots) and then rinsed with rubbing alcohol:

Image

Image

Isopropyl alcohol and a mild brush, for those hard to reach places:

Image

Barrel cleaning:

Image

and done:

Image

and Flippers now:

Image

Cleaning the barrels and flippers took wayyy to long.

And putting it back together(yes, I improvised a support). The spacebar wire was a real pain to attach :

Image

I really struggled to remove the case screws with a flat head screwdriver, but then I realized something and I felt really stupid: the screwdriver's heads are the exact same hexagonal size as all the screws in the keyboard

Image

And my trusty assistant, with the cleaning set-up:

Image

Sigmoid

07 May 2015, 17:08

Hey, just wrote you some tips on GH from my recent experience with these Model Fs. :D Anyway, I'd replace the foam. It's supposed to be springy, not flat and mushy.

Also, I'll repost an interesting piece of data here for later generations to find: You can actually test a model F without the key assemblies themselves, just connect up the bare PCB to a computer, it will act as a capacitive TOUCH sensor. (If you poke the pads of a "key" with your finger, it should register as a keypress.)

:)

User avatar
Muirium
µ

07 May 2015, 17:33

That trick also works on Topre PCBs. Touchscreens are capsense too!

User avatar
Nuum

07 May 2015, 18:09

So you could basically build a custom Model F keyboard with a capacitive touchscreen matrix?
Does a Model f flipper trigger such a touch screen? I'll have to try that once I'm at home.

andrewjoy

07 May 2015, 18:11

Nuum wrote: So you could basically build a custom Model F keyboard with a capacitive touchscreen matrix?
Does a Model f flipper trigger such a touch screen? I'll have to try that once I'm at home.

Yes they do, tested on iPad 2 :)

User avatar
Redmaus
Gotta start somewhere

07 May 2015, 18:24

Hope you get it fixed. Sounds like a capsense problem to me.

User avatar
Nuum

07 May 2015, 18:41

andrewjoy wrote:
Nuum wrote: So you could basically build a custom Model F keyboard with a capacitive touchscreen matrix?
Does a Model f flipper trigger such a touch screen? I'll have to try that once I'm at home.

Yes they do, tested on iPad 2 :)
Interesting, so you could build a keyboard with just the touchscreen part, without the display, provided it can discern presses that are 19mm apart each. You would even get 10 key roll over with modern multi-touch touchscreens.

User avatar
Muirium
µ

07 May 2015, 19:12

FingerWorks did precisely that back in the early 2000s with the TouchStream:
Before Apple bought them, killed their products, and made a little something with the technology, called the iPhone!

I saw one of these next to a DataHand and a μTRON at HaaTa's keyboard meeting a few weeks ago. Bizarrely, my mech keyboard skeptic brother latched right into the TouchStream as his favourite board there, before he even heard the backstory. He hates key travel. The TouchStream certainly optimises for that!

User avatar
derzemel

07 May 2015, 22:23

Sigmoid wrote:Hey, just wrote you some tips on GH from my recent experience with these Model Fs. :D Anyway, I'd replace the foam. It's supposed to be springy, not flat and mushy.

Also, I'll repost an interesting piece of data here for later generations to find: You can actually test a model F without the key assemblies themselves, just connect up the bare PCB to a computer, it will act as a capacitive TOUCH sensor. (If you poke the pads of a "key" with your finger, it should register as a keypress.)

:)
Thank you Sigmoid!

I'll do a test tomorrow evening with an active converter to see if it is a ps/2 decoding issue.

If the problem still persists, then I'll take it apart again in the weekend and try the finger poking test that you have suggested.

Quick question though: can I use 2mm thick rubber foam sheets (these ones) to replace the mat?

User avatar
Mal-2

08 May 2015, 00:47

derzemel wrote: The metal plate, cleaned with Contact Spray (to remove a few small rust spots) and then rinsed with rubbing alcohol:
You were lucky and had no need to paint the plate, but if you did, I wonder if it might be a good idea to tack the replacement foam mat to the back while painting the front. Then you'll certainly know exactly where to punch the holes in the mat!
derzemel wrote: I really struggled to remove the case screws with a flat head screwdriver, but then I realized something and I felt really stupid: the screwdriver's heads are the exact same hexagonal size as all the screws in the keyboard
Standard quarter inch hex drive. It works on most computer case screws too, except obviously the round kind.

User avatar
wcass

08 May 2015, 02:16

EVA "art foam" is not as soft as i like, but it works. I prefer this.

There is a thread here on DT where i have posted foam sheet templates - just print the PDF template at work of office supply store on 11x17 "Tabloid" paper (centered and do not shrink-to-fit), tape that to your foam, punch the holes, then trim the edges (this cuts off the tape).
http://deskthority.net/keyboards-f2/rep ... ml#p192651

User avatar
derzemel

10 May 2015, 08:48

One DT member was kind enough to lent me his Techkeys active converter, which he uses with an SSK.

Sadly, the 2wasz keys still do not work, so it's not a PS/2 decoding issue.

Do any of you have any Idea if/how I can test if the PCB sends signal to the controller from the 2wasz keys?

Can I do it with a multimeter?

andrewjoy

10 May 2015, 13:04

not that i know of.

As mu said the chances of it being the controller are quite low , but that does not mean its not the controller. only thing i can think of is switching the connection with a working column to see if its the pcb or controller. Hope its the controller because they can be replaced. The PCB less so , if you had an XT i would have you covered but AT unfortunately i have no spare PCB :(

User avatar
OleVoip

10 May 2015, 13:39

The controller scans the matrix with a high frequency. When the controller drives the 2wasz line and one of its keys is pressed, it takes a moment till the change has been registered. If the controller moves on to the next line quicker than the signal propagates, you loose the key press. Combined with the capacitance of the trace, resistance values in the milliohms range can already lead to a signal delay that is too long, even though the connection seem alright when you measure the steady-state resistance with your multimeter.
You could try and run a wire directly from the "2" key pad to the corresponding IC pin.

User avatar
derzemel

26 May 2015, 22:26

I listened to OleVoip (thank you!) an run 2 wires from the PCB to the controller. I first soldered them to corresponding pins to make sure they work an then I crossed them to check if the controller has problems reading the 2WASZ column and that is the problem.
With them crossed when I preesed the 2 pad on the PCB the controller received signal and registered it as the 3 key. But when I pressed the 3 pad nothing happened.
This makes me think that the controller is at fault here.

Does anybody have a spare Model F AT controller laying around and would like to part with it? :D
Or should I just get Xwhatsit's USB Controller?

Here's an image of what I did:

Image

andrewjoy

26 May 2015, 22:33

i had a feeling it was the controller, i have a spare XT PCB, is that a mask rom or a micro-controller ? possibly we could swap em if they are the same chip ?

I think i have a 122 PCB somewhere possibly you could use that and a converter ?

User avatar
derzemel

27 May 2015, 09:55

andrewjoy wrote: i had a feeling it was the controller, i have a spare XT PCB, is that a mask rom or a micro-controller ? possibly we could swap em if they are the same chip ?

I think i have a 122 PCB somewhere possibly you could use that and a converter ?
wait... does the XT and the 122 controller work on the AT pcb? aren't they different things altogether?
Anyway, I have a DIN to PS/2 cable that came with the keyboard which does it's job quite well, so no need for a converter if I can get my hands on an Model F AT controller.

I wanted to post in the "want to buy" section, but I have no idea what I should pay for an original one or if I may be better of with Xwhatsit's USB Controller

andrewjoy

27 May 2015, 10:23

the 122 one will work. but it does not speak AT protocall so you would need a terminal > USB convterter .

The XT is fixed to the board but you cold salvage components possibly

User avatar
derzemel

27 May 2015, 13:15

andrewjoy wrote: the 122 one will work. but it does not speak AT protocall so you would need a terminal > USB convterter .

The XT is fixed to the board but you cold salvage components possibly
thank you for the offer and for the info, but I'll try to find an AT controller first. I would like to keep it as original as possible.

andrewjoy

27 May 2015, 13:40

thats the best option :)

Post Reply

Return to “Workshop”