Repairing the IBM Model M2

[lowpoly]

The M2 has a bad reputation for not working. Clickykeyboards claims that 4 out of 5 M2s are broken. Bluemercury and xsphat had broken M2s too.

There were theories that PS/2 might use different signal levels than PS/1. I bought mine together with the original PS/1 computer so I thought it would be a working version. I threw that PS/1 away lately as it didn't come with the monitor which contains the power supply. As I couldn't use the 'board on a PS/1 as originally intended I decided it didn't matter anymore. I cleaned it and connected it to my Linux machine. Two LEDs came on immediately:



Otherwise it would do nothing. No characters, no LED switching. Another broken M2.

Here's the tag (german layout):



As I had cleaned it before I tried it, I had seen two SMD electrolytic capacitors on the controller pcb:



These can dry up over time and stop working. After 18  years it is quite possible that they're dry. The solution is to replace them. The small one is 2.2uF 50v, the big one is 47uF 16v. I used these for replacement:

2.2uF 63v electrolytic
47uF 16v tantal

Tantal is better because it doesn't dry but I only had a 2.2uF 35v tantal and didn't want to do it twice if that wouldn't work.

I didn't remove the pcb from the case.

The SMD capacitors are difficult to desolder. I use a very hot temperature, add a little solder for better heat distribution and push the legs to the side. If you push against the capacitor it will finally come loose with a snap and tear the other circuit path off in the process. So pushing the leg to the side seems to be the way to go. Here are the new parts, using correct polarity:



Immediately after soldering in the capacitors I connected the board to the PC. All LEDs stayed off. Tapping on the membrane showed that characters were generated as well!

During re-assembly there will be more holes than buckling springs. So it's good to take a pic before removing the springs:



Finished and working:



Unrelated, removing the numpad will be difficult with this version of the M2. There will be no room for the controller afterwards.

[itlnstln]

Sweet.  Congratulations!

[Therac-25]

I remember mine fondly.

It died, too :-(

[iMav]

I've got two M2's that work just fine.  Refurbished one of them with a full set of new buckling springs.

One great thing about the M2's are the easy accessibility of the BS's.  No frame assembly cracking necessary.  Quite easy to refurbish.

Thanks for the info on the capacitors.  Nice to know if/when mine go out.

[xsphat]

Oh well, I now have the old 1 year old who owns a Model M2, and to be quite frank, he likes the M2 better than I do. Plus, I have a Model M mini sitting right next to me, so I don't need it.

Andn for the record, I had 2 that were broke, but all three lights came on ...

[lowpoly]

Thanks for the comments.

Andn for the record, I had 2 that were broke, but all three lights came on ...

Bluemercury had that too (three lights). Might be the same cause or not. We will not know until someone who has that tries the fix.

[Mercen_505]

Oh, nice! I have two M2 boards: a perfectly working IBM from '93, and a non-functional Lexmark from '94. The bad one does just what you described: two LEDs fire, and nothing else happens. I'll have to try this next weekend when I have some spare time.

I *love* how the M2 sounds, as well as the ultra-strong action on the keys. At the same time, the lack of curvature on the board itself tends to mess me up, after years of typing on an M.

[lowpoly]

Let me know if the fix works for you.

[ozar]

The M2 has a bad reputation for not working. Clickykeyboards claims that 4 out of 5 M2s are broken. Bluemercury and xsphat had broken M2s too.

Ouch... my own M2 (1395300) has stopped working, too.

Thanks for posting this because it will come in handy when I get around to opening it up to have a look.

[jeffreytk]

I took my broken M2 (Caps Lock and Scroll Lock light and nothing works) and replaced the capacitors but it behaves the same when I plug it in *without being reassembled*.

When the microcontroller is working and the PCB is not installed to the sensing sheets with the key "things" in the right state (all "away" from the sheets???), do the LEDs typically go to all off when plugged into a working keyboard port?

Is there a replacement part for the microcontroller (and its firmware) on this thing? I would easily spend $25-ish dollars on this to make it work since the replacement cost can be much larger than that.

The coolest thing about these old keyboards and Win2000/XP/etc is that you can remap keys.

I made Caps Lock -> Ctrl, Left Ctrl -> Alt (swapped key tops with Left Alt too), Left Alt -> Windows Key, Right Alt -> Mute, `~ -> Esc, Esc -> `~.

Now Escape is where it should be, Ctrl and Alt are where they should be, and I have a Windows Key to do things like WinKey-D to toggle windows, WinKey-E for explorer, not to mention just bringing up the start menu to launch something.

[lowpoly]

with the key "things" in the right state (all "away" from the sheets???)

Not sure what you mean by that.

I didn't remove the pcb from the lower case for soldering. After capacitor replacement I plugged the 'board in and the LEDs stayed off (unlike before).

Did you preserve polarity for the capacitors?

[lowpoly]

Oh, and welcome to geekhack.

[jeffreytk]

(thanks for the welcome)

Let me clarify:
With the keyboard disassembled, one cannot know what the state of the capacitive switch is supposed to be. Does key up correspond to the spring+paddle being flat against the membrane, or does key up correspond to the spring+paddle being tilted upward?

I.e., without any spring+paddles installed, will the microcontroller still go into a working state or will it go into some error state ("your keyboard is utterly destroyed, man")?

Since you replace the caps while the PCB was installed, did you test with no spring+paddle parts on the membrane??

I removed my PCB in order to inspect it better and to see if/what components there are on the other side.

My 2.2uF is bipolar. My 47uF is installed right--negative toward silk label.

(If I had a cap tester, I could see if my removed parts were even bad. If they test bad, I either destroyed them desoldering ((small chance)) or they failed in circuit. If they test good, then I can rule out bad caps.)

So--anything else I can do? Any source for the microcontroller?

[lowpoly]

Let me clarify:
With the keyboard disassembled, one cannot know what the state of the capacitive switch is supposed to be. Does key up correspond to the spring+paddle being flat against the membrane, or does key up correspond to the spring+paddle being tilted upward?

Upward. If no key is pressed no switch is closed. It should work without the membrane connected. The switches are not "capacitive" but buckling spring over membrane. The membrane can be actuated without the buckling spring assembly by pressing the contact area with your finger.

I.e., without any spring+paddles installed, will the microcontroller still go into a working state or will it go into some error state ("your keyboard is utterly destroyed, man")?

It should work.

Since you replace the caps while the PCB was installed, did you test with no spring+paddle parts on the membrane??

Yes.

I removed my PCB in order to inspect it better and to see if/what components there are on the other side.

My 2.2uF is bipolar. My 47uF is installed right--negative toward silk label.

Maybe it's the 2.2uF. Try a tantal or electrolyte one. I did this repair with little electronics knowledge though, so maybe it doesn't matter. However, polarized capacitors won't work if loaded from the wrong side. What if the circuitry depends on that?

So--anything else I can do? Any source for the microcontroller?

I don't think this is an option.

[keyb_gr]

Currently going through the keyboard pics thread, where I found this mentioned. Yeah, those pesky early surface mount electrolytics. Good for all kinds of fun in all kinds of devices - DATs, CD players, shortwave portables (Sony in particular), video cameras, whatever.

As for what to do if a swap doesn't work, hmm. The new caps shouldn't be all that much larger than the old ones I imagine (parasitic inductance and all), and it probably doesn't hurt to use tantalums for both (careful soldering them though, they're not that fond of heat). It may be helpful to clean away residue from leaked electrolyte (isopropyl alcohol, sometimes also mixed with H2O dest, should do the trick), as this tends to be conductive and may even cause corrosion. Also check for lifted traces or similar fun stuff.

[lowpoly]

shortwave portables (Sony in particular), video cameras, whatever.

That's where I learned about the issue, Sony ICF-SW1.

[keyb_gr]

That's where I learned about the issue, Sony ICF-SW1.

For me it was the ICF-SW7600. Apart from the dead-o-lytics[tm] (and the annoying lack of the external antenna jack in the German version), I actually rather like this model. Mine was entirely recapped fairly recently, and I hope to get the antenna jack issue sorted eventually.

[lowpoly]

Nice. Here is mine (well, not actually mine):

http://www.dr-boesch.ch/radio/sony-icf-sw1-test1.htm (site in german)

Looks like it uses the same keys/buttons.

[keyb_gr]

Looks like it uses the same keys/buttons.

Yeah, same NEC microprocessor / PLL IC in there. After I had a look at the datasheet I knew where the (somewhat strange) frequency entering concept came from...

And before it gets off-topic too far, these sets actually have pretty good key feel, tactile with short throw. Older sets (e.g. 7600D, 2001D) still had rubber mat over microswitches, soft but not that bad either. That said, I can't complain about key action on DE1102 and E100 either, only size is an issue with the E100. The Chinese don't always get it right though, the RP2000 (Roadstar TRA-2350P) suffers from slow key scanning.

[Rajagra]

I should be getting an M2 soon (or maybe not, the seller underestimated postage and has to sort it out.)

I was thinking of buying some replacement capacitors in advance. I have a couple of questions:

If the old caps have failed, do they go open circuit / low capacitance? If so you could just solder the new ones in in parallel.

Is it obvious from the board which way round to connect the caps? Or do you have to take notes from the existing caps before you remove them?

[lowpoly]

The + symbol is on the pcb:

[Rajagra]

Ah, thanks I see them now! :embarassed:

[timw4mail]

Ah, the good old original Model M.  My electrolytics know they can be easily replaced so they BETTER not act up.

Show Image

My controller has two of these commonly available guys.  Still surprised how long they last.

After that era of exploding computer caps...the only capacitors I'm comfortable with are the solid ones.

[keyb_gr]

You might enjoy my page on capacitor nonlinearities and linked literature.

It always pays to be somewhat skeptical with boutique stuff - oil paper caps seem to have quite some dielectric absorption, or so Bob Pease mentions. Huge film caps obviously have a good bit of series inductance as well.

Regular electrolytics with a nice bias voltage across them that are not subjected to high ripple currents or temperatures can easily last 25..30 years. The smaller ones which are more prone to drying out may still do 15..20 years with no major problems, so it's not surprising that our Ms are doing so well. A lot of these early surface mount 'lytics seem to have been junk though (as observed with the M2).

The ExplosiCaps[tm] found in PCs were a result of both faulty electrolytics with wrongly copied electrolyte and wrong derating at excessive ripple currents, i.e. the caps were deliberately run at ripple currents over spec (so one could use fewer of them) but the formulas for life expectancy calculation were off. PC mainboards are manufactured under extreme price pressure, don't forget.
When replacing dead electrolytics, some people have installed small film or ceramic bypass capacitors in parallel in order to keep the high-frequency ripple away from the 'lytics. Even though one may have to consider parallel resonance ('lytic series L || bypass C) when using ultra low ESR 'lytics, this usually seems to work well in practice.

Given that the 'lytics in the regular Ms are axial types mounted a good bit away from the microcontroller, I don't think they need anything fancy.

[timw4mail]

The ExplosiCaps[tm] found in PCs were a result of both faulty electrolytics with wrongly copied electrolyte and wrong derating at excessive ripple currents, i.e. the caps were deliberately run at ripple currents over spec (so one could use fewer of them) but the formulas for life expectancy calculation were off. PC mainboards are manufactured under extreme price pressure, don't forget.

Planned obsolescence makes me sick... especially in regards to computer parts... What's wrong with the board working in five years still, when I bought a new computer?

I digress....

Are the aluminum caps susceptible to anything similar to the electrolytic cap CAPastrophe?

[keyb_gr]

Are the aluminum caps susceptible to anything similar to the electrolytic cap CAPastrophe?

You mean solid electrolytic / specialty polymer? Don't think so, they can't dry out.

Well, that's what's surprising me.  The first Model M's shipped in 1985 (Source: Clickeykeyboards) so we are talking 24 years.

I'm not worrying about it, just surprised.

I would guess that the electrolytics present are bulk capacitance and not chosen very tightly. They could show considerable worsening of capacitance and ESR and you may not notice right away. The board might be some more finicky with long cables and stuff like that, but do you notice that unless you have another one for comparison?

[timw4mail]

When these guys start to dry out I think we'll start seeing a lot of posts like that one.  Caps don't last forever.

Electrolytic ones don't, anyway. I bet the solid ones last a whole lot longer.

[Rajagra]

Electrolytic ones don't, anyway. I bet the solid ones last a whole lot longer.

Brawndo! It's got electrolytes!

[o2dazone]

Would you like some more BIG ASS FRIES

[itlnstln]

Wow, the "Thirst Mutilator."  That's quite a claim.  I could have used that for the disc golf tournament I was in this past weekend.  It was HOT (over 100 deg. F).  I was sick afterwards (headache, nausea).  I drank 3+ gallons of water between the night before and the day of the tournament to get/stay hydrated, but it didn't work.

[lowpoly]

Wow, the "Thirst Mutilator."  That's quite a claim.  I could have used that for the disc golf tournament I was in this past weekend.  It was HOT (over 100 deg. F).  I was sick afterwards (headache, nausea).  I drank 3+ gallons of water between the night before and the day of the tournament to get/stay hydrated, but it didn't work.

Too much water can also be bad. I read that it happens to Marathon runners sometimes. They drink all the time to avoid dehydration but when running, the body can't remove excess water from the blood very well. So with too much water in the blood they feel nausea etc., go to the first aid tent, get misdiagnosed, get an infusion with more water and that's it. Not saying this happened to you and I'm writing this from memory.

Edit: http://www.nytimes.com/2005/10/20/sports/othersports/20marathon.html

[itlnstln]

Too much water can also be bad. I read that it happens to Marathon runners sometimes. They drink all the time to avoid dehydration but when running, the body can't remove excess water from the blood very well. So with too much water in the blood they feel nausea etc., go to the first aid tent, get misdiagnosed, get an infusion with more water and that's it. Not saying this happened to you and I'm writing this from memory.
 
Edit: http://www.nytimes.com/2005/10/20/sports/othersports/20marathon.html

Hmm.  Maybe that was it.  Perhaps I drank too much water.

[keyb_gr]

Also remember that drinking a lot of pure water is a good way of getting rid of minerals and stuff. Seems like even tea is better already.

[keyb_gr]

That's strange. One of my broken M2s just came back to life.

All by itself (or with some assistance)? Might be time to look for cable breaks and bad solder joints, some cleaning of the board with alcohol couldn't hurt either.

[lowpoly]

Did anyone ever confirm this fix works (other than the OP's keyboard)?  JeffreyTK's obviously never did.

He used a bipolar cap instead of a polar one. The question whether this would work was never answered. Unless his still non-functional pcb was the answer.

Still, with my board it could have been a case of disassemble-assemble-works. Not very likely though as I didn't remove the pcb.

[keyb_gr]

A bipolar electrolytic should generally do just as well. Not so sure about ESR though.

[huha]

I repaired my M2 today. Used an electrolytic and a tantalum cap just as the OP did. Maybe two electrolytics would suffice, but I had to buy some electronics anyway, so I went for the tantalum.
The small capacitor is quite nice to desolder. Heat everything up (really cheap and nasty soldering iron here I'm afraid) and gently push it to the side. Done. The large one was a pain and I did manage to remove about 2mm of the trace on the PCB. Not as bad as it sounds, as this was still a contact pad, so no real harm done there. I probably should have heated it up even more.
I also tried letting it sit here and unplugging/replugging it after some time webwit-style, which didn't work.

So now, I can confirm two things:
1) The seller was full of ****
2) It works now, so the modification/repair seems to work. Big thanks!

-huha

[huha]

Great news!  Glad that worked.  Maybe the other guy just used the wrong cap and they can't be bipolar.

So the seller said it was working?

Yes. Oh, it was quite a long story. I only kept it because they're somewhat hard to find. I still feel slightly ripped off although he did give me back half of what I paid for it. I just opted for that option because communication was painfully slow--waiting a week for a reply to even the easiest questions wasn't exactly trust-inducing either, so I didn't want to go through the hassle of returning the board to him.
I'm just pissed because he said it worked, it didn't and the caps were at fault. That's just nothing that can happen during shipping.

-huha

[keyb_gr]

I once got a tip for desoldering these surface mount electrolytics: Squash them with some pliers or so, then you should be able to remove most of the cap with only the legs being left. These can then be desoldered comfortably.

[lowpoly]

I repaired my M2 today. Used an electrolytic and a tantalum cap just as the OP did. Maybe two electrolytics would suffice, but I had to buy some electronics anyway, so I went for the tantalum.
The small capacitor is quite nice to desolder. Heat everything up (really cheap and nasty soldering iron here I'm afraid) and gently push it to the side. Done. The large one was a pain and I did manage to remove about 2mm of the trace on the PCB. Not as bad as it sounds, as this was still a contact pad, so no real harm done there. I probably should have heated it up even more.
I also tried letting it sit here and unplugging/replugging it after some time webwit-style, which didn't work.

So now, I can confirm two things:
1) The seller was full of ****
2) It works now, so the modification/repair seems to work. Big thanks!

-huha

Good that it worked for you. :smile:

I would have taken two tantalums for mine but went with what I had then. New electrolytics should last too.

[keyb_gr]

New electrolytics should last too.

I'd prefer some brand-name 105° types to some cheapo 85° Taiwanese caps of doubtful quality then, i.e. better a Panasonic FC than some Jamicon or so.

In general, with at least half-decent electrolytics I wouldn't worry too much about them failing in the next 20 or 25 years.

Some people advise against tantalum beads, citing reliability problems (generally intermittent shorts, not a nice failure mode). Tantalums may not be that fond of rapid voltage changes and moisture either.

BTW, the 2µ2 should be easy to replace with a ceramic X5R surface mount part (better don't use Y5V or even the super crummy Z5U).

[talis]

Some people advise against tantalum beads, citing reliability problems (generally intermittent shorts, not a nice failure mode). Tantalums may not be that fond of rapid voltage changes and moisture either.

Its actually the low ESR* and the ability to more quickly adapt to changes in voltage that make tant. caps so desirable for power supplies.  The thing that tends to kill them is the initial inrush current (going from 0v to their steady state voltage), usually they design in a small ferrite to help limit this current.  Once at steady state they can more quickly react to fluctuations in voltage and produce less noisy voltage rails.  I guess not taking into account the inrush current is a fairly common design mistake.

The moisture thing is a problem with most larger high precision components.  You'll find a lot of chips are shipped with desi-packs to help reduce the moisture they absorb.  During reflow the moisture vaporizes and can cause cracks in the component body (or structure in the case of capacitors).  Once through the reflow process, moisture shouldn't really be a huge issue (unless the components get really hot).


*ESR - Equivalent series resistant.  Think of it as an imaginary resistor in series with the capacitor created by the technology used to make the capacitor.

[Rajagra]

I should be getting an M2 soon (or maybe not, the seller underestimated postage and has to sort it out.)

Mine finally arrived, and it has this problem. Time to buy some caps.

Any hints on disassembly? I assume the 2 "security" bolts have to come out.

[lowpoly]

Don't remember these bolts and I don't have access to the 'board right now.

IIRC, there are a few screws but the strength comes from many many clips. You have to remove the keys to see them.

[huha]

Now that's interesting!

My M2 featured two slotted screws. I guess the guy who sold it to me was an *******, after all.

-huha

[lowpoly]

Mine have the common Phillips cross slot.

[Rajagra]

Ugh, that wasn't fun. The SMT cap took forever to remove. I couldn't find my solder sucker, and braid doesn't seem to work well with the new lead-free solder.

I eventually replaced the dodgy looking 2.2 µF 50V electrolytic with a 2.2 µF 35V tantalum bead (all the shop had.)

Before:


After first assembly, noticed the Esc key foot was out of position. Managed to put that right by prying the case open just a bit.

Fully assembled again, then found some keys didn't work. (Space, B, N, Left, Down, Right, Numeric -).

Stripped it down and tried to remove the PCB to clean the membrane edge conectors. Couldn't do it, too many clips hold it down, and pushing back 2 clips at a time allowed zero movement of the PCB.

Resorted to wiggling the membrane around. It moved freely, and I think this wiping action cleaned the contacts.

Assembled again, and so far it works!

My advice is:

• Don't buy a faulty one of these expecting a quick fix.
• When fixing one, test it in every way you can at every opportunity. Use a PS/2»USB convertor so you can hot-plug it as often as you want.
• Don't do the labour-intensive steps (inserting the feet or keycaps) until you've done all the tests you can.
• I did the final key cap installation while running Aqua'S Key Test, so I knew the keys worked as well as clicking properly.

EDIT> The faulty cap did seem to be open circuit. Simply soldering the new cap in parallel might well provide the fix.

[Rajagra]

Did you try the crushing/removing the cap trick?

Soldering looks fine to me - I'm doing some Blue Cube connector mods and the desoldering part is by far the hardest.  On through hole that last bit of solder just doesn't want to come out.

No, I thought crushing might rip the tracks up.

It looked like the small cap had leaked to me. I couldn't get any response from it using a resistance meter (I reversed the probes, and hence voltage, but there was still no flicker of current.)

The square stabilising base under the SMT cap appears to be the cause of difficult removal.

[JBert]

Soldering looks fine to me - I'm doing some Blue Cube connector mods and the desoldering part is by far the hardest.  On through hole that last bit of solder just doesn't want to come out.

I'm by no means an expert, but in such cases I sometimes drop a litte bit of recycled solder onto the lead so I can put my desolder pump into a larger pool of solder. This also helps to transfer the heat of your solder iron to the lead whereas it can be hard to melt that last bit of solder.

Of course, this may not work if the PCB is not a mere one-sided one. In such cases, the copper in the hole goes from the back- to the front-side and may attract the last bits of solder.

[talis]

If its RoHS the issue tends to be getting the solder all the way through up to a temperature hot enough to melt it (and hot enough that it doesn't drop below the melting point the second you take the iron away).  The other issue that tends to come up more often is a small amount of solder keeps the pin connected to the through hole plating.  This is most dangerous as it tends to result in the plating being pulled out with the pin.

If you can be destructive in your through hole desoldering, its usually best to just plan to suck up the remainder of the pin into the solder sucker (unless its a vacuum type one that tend to plug easily).

SMT electrolytic aren't bad if solder wick works well for you :

Start by adding a bit more solder to one side.


Then use the wick to remove as much solder as you can from both pads :


Apply heat to one side while gently rocking the cap away from the pad.  This will lift one contact (you don't need to go far, since most of the solder was removed by the wick).


Then heat the other side and remove the cap.  It doesn't take much to remove them with this method, and its very unlikely that you'll pull up a pad or damage the board.  After you're done, clean up the pads with a bit of solder wick.