Model F has lost its little mind (sending spurious keystrokes), why?

Red_October · 20 Jul 2018, 21:07

So I've got a Model F ("XT" version as on a 5150) that I had been using with a Soarer's Converter. Well, recently, I had some trouble with the converter, culminating in it ceasing function altogether. I ultimately replaced it with one I assembled by hand from a Teensy microcontroller. I had previously thought that these malfunctions were related to the failing converter, but it turns out that it appears to be the keyboard.

What happens is, spurious keystrokes are sent. Usually it is either a rapid pulsing of the "alt" key, and sometimes it is function keys (I get a lot of instances of the "Caret Browsing" pop-up confirmation in Firefox caused by pressing F7) and lots of the Windows "Ding" noise that comes from pressing an invalid key. I found this really, really weird as it was a new converter and one I had designed to not be subject to any stresses on the leads (DIN side is a panel-mount in an enclosure, board itself is held to the bottom of the enclosure with double-side foam tape, USB lead is held to the inside of the enclosure with Kraft tape) so I thought that the only thing that was the same in this system (error has occurred across multiple computers) was the keyboard itself, so I swapped out the Model F for the only other XT keyboard I had handy, a Honeywell/Bull T.I.60164180-001 Rev. J (that's the "Customer Part Number" off the back, closest thing it has to a model number), a huge affair with 5 LEDs (all local toggles) and a shitpot of unlabeled keys, which, while it only sort-of works with the programming for the XT, didn't produce the error at all. I'm wondering what can cause this malfunction in an F, and what the solution is. I do have another F-XT around here, a sort of "parts unit" that's had a really rough life, it's missing one of the "wings" for the feet, etc, but I could swap the controllers or even the entire guts without too much trouble, if it comes to that, but I'd still like to learn what causes this failure. Thanks guys!

mark201200 · 20 Jul 2018, 21:27

did you open it up? maybe the PCB is a bit dirty.

snacksthecat · 20 Jul 2018, 21:31

Something similar was happening to me and it turned out to be a voltage problem. Do you have a multimeter? If so, check the voltage on the board to make sure it's adequately high. Mine was around 4.2V when I experienced the problem.

Sometimes it can be a simple fix like making sure ground is properly connected (tightening the screw that connects the controller to the plate)

Red_October · 21 Jul 2018, 09:28

mark201200 wrote: did you open it up? maybe the PCB is a bit dirty.

So I opened it up to have a look at this... it was a little dirty, on a couple of contacts. I also think I found out why the F7 is frequently prone to spurious keypresses: the spring is significantly longer than all of the others! I also found out that you can't put one of these back together with the keycaps in place! So I've got it held together with rubber bands right now, (so I don't lose the goddamn little contactors) and I'm going to put it on the workbench tomorrow and see to it properly, with clamps and pliers and so on and so forth. Honestly, having a Beamspring unit open to this level was infinitely easier, at least on my Displaywriter keyboard, all you do is install the module upside-down and presto, it's mounted securely to be worked on. On the F, there's no way to do anything like that, and, there's no apparent quick way to pop it apart and then back together. Frankly, I'm not sure if it was ever meant to come apart, and it simply can, unlike the Model M, which is clearly meant to be a non-serviceable assembly, it's only in latter days that we've learnt how to recondition them. Hopefully I haven't fucked things too bad, because I broke a few of the little "feet" off of the contactors, and one now has no "feet". Maybe "ears" is a better term. I'm not worried, I have a parts unit (if the broken feet prove problematic, I plan on just swapping the modules and seeing if that sorts it out) and a really raunchy "Blue Switch" "Bigfoot" F that could donate contactors if need be, but I suspect the other module should work straight away, these things are seriously robust and I'm honestly surprised to have had any trouble at all with one.

fohat · 21 Jul 2018, 15:46

If the spring was lower it probably indicates that the pivot plate had fallen out of place.

Yes, if you never picked up on it from reading the guides, all key stems must come completely out and go completely back on for each iteration, probably the most time-consuming part of the process. And as I have said countless times - expect to do this more than once every time you crack one of these open and try to get it back together.

Having been through this exercise many dozens of times (estimating average 2.5 attempts each for 2 dozen Model F refurbs) I have only ever broken maybe 2-3 "ears" off of pivot plates, so maybe you have been especially rough. I seriously doubt that they will work properly without both ears, but I could be wrong. Those ears lie in slots that are critical to positioning them and allowing them to pivot.

Lastly, if this is not a 122-key terminal, the space bar and its "undercarriage" are also quite fussy. I use a thread and a weight to keep them in place as the rest goes together.

Red_October · 21 Jul 2018, 21:12

Thanks a bunch, Fohat, I was a bit rough with it at first because I thought the contactors (I don't think that's right... it's capacitive not conducting...) were metal and I could just bend them back if they got smooshed. Well, you live and you learn. I just thought "I've had a Beamspring apart on my lap before, how hard could a little ol' Model F be?" Well if you watch "Top Gear", you know the perils of that question! So happy I have a parts unit now! And it's not a 122-key, my 122-key is working fine, touch wood. But I do know quite well the perils of the Model F spacebar and not to fuck with it unless absolutely necessary.

Red_October · 22 Jul 2018, 03:16

OK I am now 100% officially lost. I swapped out the modules for the one from the rough keyboard, and plugged it in, and started getting the same fucking errors! Random, spurious, constant keypresses, F5 is common (prints date & Time), lots of Xs and Qs, backslashes, forward slashes, etc. So I changed the cable, as that was the only thing I had not changed (I changed the cable from the nasty one to the good one years ago when I first started getting this problem because it seemed to be related to the cable and that's always the first thing you check anyway) and I kept getting the same problem! I have no clue what's going on here now, and I'm going right now to take both cables downstairs and see if there's some kind of short going on, and maybe even do up a replacement cable.

snacksthecat · 22 Jul 2018, 03:44

W̶h̶a̶t̶ ̶t̶u̶r̶n̶e̶d̶ ̶o̶u̶t̶ ̶t̶o̶ ̶b̶e̶ ̶t̶h̶e̶ ̶i̶s̶s̶u̶e̶ ̶w̶i̶t̶h̶ ̶m̶y̶ ̶v̶o̶l̶t̶a̶g̶e̶ ̶p̶r̶o̶b̶l̶e̶m̶ ̶i̶s̶ ̶t̶h̶a̶t̶ ̶t̶h̶e̶ ̶d̶i̶n̶ ̶5̶ ̶c̶o̶n̶n̶e̶c̶t̶o̶r̶ ̶I̶ ̶w̶a̶s̶ ̶u̶s̶i̶n̶g̶ ̶d̶i̶d̶n̶'̶t̶ ̶h̶a̶v̶e̶ ̶t̶h̶e̶ ̶s̶l̶e̶e̶v̶e̶ ̶c̶o̶n̶n̶e̶c̶t̶e̶d̶ ̶t̶o̶ ̶G̶n̶d̶ ̶o̶n̶ ̶t̶h̶e̶ ̶m̶i̶c̶r̶o̶c̶o̶n̶t̶r̶o̶l̶l̶e̶r̶.̶ ̶ ̶I̶ ̶p̶r̶e̶s̶u̶m̶e̶d̶ ̶t̶h̶a̶t̶ ̶y̶o̶u̶ ̶o̶n̶l̶y̶ ̶n̶e̶e̶d̶e̶d̶ ̶t̶o̶ ̶h̶o̶o̶k̶ ̶u̶p̶ ̶t̶h̶e̶ ̶V̶c̶c̶,̶ ̶G̶n̶d̶,̶ ̶D̶a̶t̶a̶,̶ ̶C̶l̶o̶c̶k̶ ̶a̶n̶d̶ ̶d̶i̶d̶n̶'̶t̶ ̶t̶a̶k̶e̶ ̶i̶n̶t̶o̶ ̶a̶c̶c̶o̶u̶n̶t̶ ̶t̶h̶e̶ ̶C̶a̶s̶e̶ ̶G̶n̶d̶ ̶t̶h̶a̶t̶ ̶c̶o̶m̶e̶s̶ ̶t̶h̶r̶o̶u̶g̶h̶ ̶t̶h̶e̶ ̶d̶i̶n̶ ̶5̶ ̶s̶l̶e̶e̶v̶e̶.̶

Edit: don't do dis

Red_October · 22 Jul 2018, 04:41

snacksthecat wrote: What turned out to be the issue with my voltage problem is that the din 5 connector I was using didn't have the sleeve connected to Gnd on the microcontroller. I presumed that you only needed to hook up the Vcc, Gnd, Data, Clock and didn't take into account the Case Gnd that comes through the din 5 sleeve.

I'm fairly certain that I hooked that up, by tying it to the USB shield... Is there a specific point on the Teensy for it? That was one of my theories, I'm going to take the whole converter downstairs tomorrow and go over it with a fine-toothed comb.

Oh, and both cables were fine. Neither had any conductance at all between conductors, (High-speed HP Multimeter reads "Overload" in Ohms scale), one had no conductance at all from any conductor to the earth, one had conductance but at several hundred thousand Ohms between any given conductor to earth, which really shouldn't matter at 5v data bus levels.

JBert · 23 Jul 2018, 09:54

snacksthecat wrote: What turned out to be the issue with my voltage problem is that the din 5 connector I was using didn't have the sleeve connected to Gnd on the microcontroller. I presumed that you only needed to hook up the Vcc, Gnd, Data, Clock and didn't take into account the Case Gnd that comes through the din 5 sleeve.

I have been taking apart my model F XTs after having similar issues as the OP.

I stuck the converter in the case though and noticed that if I move the USB cable snaking out of it then the keyboard goes haywire.

If I remember correctly the DIN shield is not connected to the controller but to the keyboard case's back plate, right?

snacksthecat · 23 Jul 2018, 14:10

Eeek, I may be spreading some misinformation. Edited post accordingly. In any case, I know my issue has to do (in some way) to this since it is working now. Curious about what the right way to hook this up is.

So I guess what I've learned is that circuit ground is different from earth ground. This probably seems absurdly obvious to most people. I guess I didn't pay good enough attention in science class

andrewjoy · 23 Jul 2018, 16:43

The main issue is improper grounding of the controller. If its not grounded properly it can go absolutely mental, that goes for originals AND the modern replacement.

Case ground and circut ground WILL be the same thing for a keyboard, you dont want your keyboard referanced to mains earth LOL

fohat · 23 Jul 2018, 17:38

andrewjoy wrote:
(1) Case ground and circut ground WILL be the same thing for a keyboard,

(2) you dont want your keyboard referanced to mains earth

Are you sure? Those small wires (usually yellow) with the round connectors have a dedicated screw and nut to ground the cable (sheathing and cylindrical surround) to the case. "Ground" as shown #4 on the DIN drawing is part of the interior pinout.

There is no main anywhere to be seen, unless it is traveling from the motherboard through the mesh sheath in the keyboard cable.

JBert · 23 Jul 2018, 17:44

fohat wrote:
andrewjoy wrote:
(1) Case ground and circut ground WILL be the same thing for a keyboard,

(2) you dont want your keyboard referanced to mains earth

Are you sure? Those small wires (usually yellow) with the round connectors have a dedicated screw and nut to ground the cable (sheathing and cylindrical surround) to the case. "Ground" as shown #4 on the DIN drawing is part of the interior pinout.

There is no main anywhere to be seen, unless it is traveling from the motherboard through the mesh sheath in the keyboard cable.

Quite frankly, I thought that USB specifies that the shielding can only be connected to circuit ground at the host side. On average they'll have the same potential but the host might check it for faults.

It's possible that the model F requires the same...

Red_October · 23 Jul 2018, 19:49

Mm, yes, this all gets confusing. By "ground" in my earlier posts, I meant the shield-ground, conveyed through the outer ring of the DIN, to one of the case screws. Not circuit-ground (neutral, return, etc.) nor Safety Earth Ground (Facility/Ferrous Ground). Anyway, it got Overtaken By Events yesterday and I hope to address it today. Might be a USB cable or even hub issue, I seem to recall my other Soarer's Converter not liking USB Hubs and behaving badly unless plugged directly into the bus...

Red_October · 24 Jul 2018, 08:45

andrewjoy wrote: The main issue is improper grounding of the controller. If its not grounded properly it can go absolutely mental, that goes for originals AND the modern replacement.

Case ground and circut ground WILL be the same thing for a keyboard, you dont want your keyboard referanced to mains earth LOL

So far this seems to be it! I checked the basics, that shield-ground from the DIN-5 was passing to the USB socket on the microcontroller. It was. I checked that the tiny, short cable that shipped with the microcontroller was passing shield-ground. It was not. I checked that the USB extensions that I was using were passing shield-ground. They were not. I checked that the shield-ground at the USB hub was straight-through to the shield-ground at the system. It was not. I replaced the cable that came with the Teensy with a longer one that respects the shield-ground AND has a nice magnetic noise trap RIGHT at the Mini-B in exactly the right place to pull double-duty as a strain-relief inside the case. Connecting this through the USB hub (resistance of 4.7 Ohms to the shield-ground of a chassis USB socket) and immediately began to experience errors. I connected it to the chassis socket and got no errors and a responsive keyboard. We shall see how this goes in practice, but I have learned something: the Model F must always have a continuous, positive connection to the system earth.

I had added an earth screw to the case with the microcontroller, in case no USB cables respected the shield-ground and it had to be established externally, but this turned out not to be needed.