IBM 3101 Beamspring keyboard restoration.

chzel · 24 Jan 2016, 15:31

So, thanks to dzhoou, I got me a beamspring!
Advertised as a barn find, it turns out it was an accurate description!
After a couple of weeks in transit, it arrived in perfect packaging, in a huge box with styrofoam and stuff.

: D72_2073.jpg (681.76 KiB) Viewed 11087 times

I got it out and had a look...and immediately decided to start cleaning it...It was rather nasty, so some pics are going under spoiler tags for the sake of any squirmish reader among us!

Quite a bit of dirt on there, all caps are present, although 7,8 and 9 had bent stems.

: D72_2077.jpg (769.15 KiB) Viewed 11087 times

This one is for Halvar: The secret compartment!
Nothing too secret of course,just DIP switches!

: D72_2088.jpg (683.13 KiB) Viewed 11087 times

And instructions

: D72_2089.jpg (616.91 KiB) Viewed 11087 times

I pulled a cap, and the contamination shield has disintegrated completely.

: D72_2078.jpg (754.47 KiB) Viewed 11087 times

The back of the beast, unfortuntely the manual and cover are missing.

: D72_2079.jpg (783.54 KiB) Viewed 11087 times

1979 CANADA INTERNATIONAL BUSINESS MACHINES CORPORATION

: D72_2080.jpg (717.86 KiB) Viewed 11087 times

The label, born on the 18th week of 1982. The first keyboard I own that's older than me.

: D72_2082.jpg (902.42 KiB) Viewed 11087 times

So, I don nitrile gloves and dig in...
Blue screws with star lock washers!

: D72_2087.jpg (740.23 KiB) Viewed 11087 times

And I'm greeted by this little guy.

Spoiler:

As soon as I snap the pic he rushes somewhere inside, only to be found again later!

This is what I'm dealing with...

: D72_2095.jpg (873.46 KiB) Viewed 11087 times

Nastiness alert:

Spoiler:

To be continued...

chzel · 24 Jan 2016, 15:31

Nastiness continues:

Spoiler:

Only one rusted switch,

: D72_2104.jpg (944.1 KiB) Viewed 11066 times

So much contamination, so little shield...

: D72_2106.jpg (757.23 KiB) Viewed 11066 times

Shield pulled off, the plate has some corrosion, but nothing too bad.

: D72_2109.jpg (851.77 KiB) Viewed 11066 times

The controller board

: D72_2110.jpg (891.35 KiB) Viewed 11066 times

: D72_2111.jpg (866.48 KiB) Viewed 11066 times

: D72_2112.jpg (854.81 KiB) Viewed 11066 times

Lift the switch assembly and oh, the horror...

Spoiler:

The solenoid and cable entry

: D72_2117.jpg (766.49 KiB) Viewed 11066 times

: D72_2118.jpg (795.73 KiB) Viewed 11066 times

Spidey again...Getting a good macro shot of a spider running for it's life is damn hard...

Spoiler:

Switch sandwich removed and mounted upside down for safe opening

: D72_2121.jpg (845.56 KiB) Viewed 11066 times

: D72_2123.jpg (759.15 KiB) Viewed 11060 times

chzel · 24 Jan 2016, 15:31

The pad card had a ton of crud on it. IPA to the rescue.

: D72_2124.jpg (850.33 KiB) Viewed 11055 times

Switches exposed

: D72_2127.jpg (867.36 KiB) Viewed 11055 times

Nasty foam, as is normal for IBM...

: D72_2130.jpg (942.51 KiB) Viewed 11055 times

All switches removed and lined up!

: D72_2132.jpg (905.1 KiB) Viewed 11055 times

At this point everything got a bit messy, so I have no pics of the actual cleaning, but the case and the switch "cage" were washed and scrubed with a brush and hand soap.
I didn't go crazy on the case, as there is quite a bit of deep-embeded dirt, and it was already flaking in some spots.

Meanwhile the caps got a good soak and cleaning with a toothbrush.

: D72_2131.jpg (765.72 KiB) Viewed 11055 times

I removed the orings from the switches to clean and lubricate them.

: D72_2138.jpg (809.2 KiB) Viewed 11055 times

Washed and soaked in just a bit of silicone oil.

: D72_2140.jpg (878.83 KiB) Viewed 11055 times

I cleaned the stems from the contamination shield remains with and eraser and a brass brush.

: D72_2136.jpg (940.17 KiB) Viewed 11055 times

A few springs were a bit rusty, so in the rust converter they go.

: D72_2142.jpg (919.56 KiB) Viewed 11055 times

You can easily remove them by "screwing" them off.

: D72_2141.jpg (646.88 KiB) Viewed 11055 times

The rusty stem also got treated and it was quite pitted.
It was a bitch to remove, I tried WD40, silicone oil, pulling, cursing, everything. In the end a tiny bit of heating with a lighter made it release.
After this shot it got another tour in the cup of rust converter.

: D72_2144.jpg (887.61 KiB) Viewed 11055 times

IBM's selection of foam was quite stupid. Even stupider than rivets on the M.
These are the glue backing remains of the foam layer.

: D72_2146.jpg (817.25 KiB) Viewed 11055 times

Plate remounted upsider down for switch placement.

: D72_2147.jpg (831.97 KiB) Viewed 11055 times

Instead of making new foam, I tried another solution. I used #15 o-rings on each switch.
I had plenty (widely used in paintball) and it was easy enough so I gave it a try. The black one is Buna-N 70A Durometer (used 84) and the clear is urethane (90A IIRC, used 4)

: D72_2150.jpg (855.1 KiB) Viewed 11055 times

Switches ready for action!

: D72_2151.jpg (785.68 KiB) Viewed 11055 times

chzel · 24 Jan 2016, 15:31

All caps mounted, and a bit of spacebar detail. A bunch of codes, and apparently it's made of PEI

: D72_2153.jpg (910.01 KiB) Viewed 11052 times

: D72_2155.jpg (615.46 KiB) Viewed 11052 times

Controller installed and cable secured.

: D72_2156.jpg (875.04 KiB) Viewed 11052 times

Caps are great, even the printed front legends, altough some are better than others.

: D72_2157.jpg (712.8 KiB) Viewed 11052 times

: D72_2159.jpg (820.93 KiB) Viewed 11052 times

: D72_2161.jpg (640.8 KiB) Viewed 11052 times

Overall a rather handsome keyboard, and so tall that it actually makes me want a shorter desk...

: D72_2164.jpg (753.98 KiB) Viewed 11052 times

Back is bland, just a block of metal...

: D72_2171.jpg (692.3 KiB) Viewed 11052 times

Bottom all cleaned up. All labels where removed and will be laminated as soon as I can get pouches.

: D72_2173.jpg (639.85 KiB) Viewed 11052 times

The IBM badge

: D72_2175.jpg (627.67 KiB) Viewed 11052 times

Artsy fartsy black and white!

: D72_2168.jpg (924.33 KiB) Viewed 11052 times

The orings work nicely, except for the middle row of switches that's somewhat loose. Perhaps softer orings would work better.
No ill effects, but the switches turn a bit in the hole, and caps end up wonky..
I'll open it up again at some point and make new foam for it.

: D72_2180.jpg (603.93 KiB) Viewed 11052 times

It feels great, there is a certain feel that's hard to describe, heavy but smooth and fluid. I think I barely bottom out, but it's a bit tiring.
I have to decide what to do with the case and switch plates. I'll check to see how much Cerakote costs, and if it's too much I'll go with powder coating.
I'd love to strip it and clearcoat it, but the hatch is a problem, as it is plastic, and I'd have to either paint it, or fabricate one out of aluminium.
I'm tempted to put Lock LEDs in the hatch, but xwhatsit doesn't support both LEDs and solenoid, and I'm definitely using the solenoid (driver is in the mail)!
Another thought is a switch to change the configuration from solenoid to LEDs, and with a simple change from the software have one or the other.

As a reference here is the 3101 manual from bitsavers

snuci · 24 Jan 2016, 16:10

Hi chzel,

That keyboard looked like a write-off on eBay. You did a fantastic job! Nice find.

chzel · 24 Jan 2016, 16:16

Thanks snuci, it was a gamble, but turned out to be quite nice under all the dirt and crap.
If you hadn't doubted I wouldn't be typing on this beast right now...so...Thank you!

idollar · 24 Jan 2016, 16:30

Nice work, very nice.

I would like to know how did you dismantled the switch.

chzel · 24 Jan 2016, 16:43

If it's not a particularly stubborn switch, you grab the stem with pliers, and pull STRAIGHT and steady. No rocking at all.
Just as the corners of the switch dig into your hand and it starts to hurt a bit, the stem starts moving. If it hurts a lot, stop pulling. That way it's easy not to go Hulk on the stem and break something. After you pull the stem out, the switch just pulls apart. To reassemble, put everything together, stand the switch on the table and grab the stem again with pliers and push it on the white slider until it seats correctly.
I didn't actually dismantle more than about 5 switches, no real need to. Also popping the fly-plates accidentally is quite hard if you mind your hands while handling the switch.

When I have it apart again for foam making, I'll make a guide, and maybe a video?

Halvar · 24 Jan 2016, 16:56

That's brilliant work, you really went the extra mile with those rusty switches. O-rings seem like a great idea.

chzel · 24 Jan 2016, 17:05

Thanks Halvar!
The orings work nicely, but ideally you either want softer orings (40-50A) or thicker ones along the centreline.

idollar · 24 Jan 2016, 17:21

chzel wrote: If it's not a particularly stubborn switch, you grab the stem with pliers, and pull STRAIGHT and steady. No rocking at all.
Just as the corners of the switch dig into your hand and it starts to hurt a bit, the stem starts moving. If it hurts a lot, stop pulling. That way it's easy not to go Hulk on the stem and break something. After you pull the stem out, the switch just pulls apart. To reassemble, put everything together, stand the switch on the table and grab the stem again with pliers and push it on the white slider until it seats correctly.
I didn't actually dismantle more than about 5 switches, no real need to. Also popping the fly-plates accidentally is quite hard if you mind your hands while handling the switch.

When I have it apart again for foam making, I'll make a guide, and maybe a video?

Thanks a lot

idollar · 24 Jan 2016, 17:21

chzel wrote: If it's not a particularly stubborn switch, you grab the stem with pliers, and pull STRAIGHT and steady. No rocking at all.
Just as the corners of the switch dig into your hand and it starts to hurt a bit, the stem starts moving. If it hurts a lot, stop pulling. That way it's easy not to go Hulk on the stem and break something. After you pull the stem out, the switch just pulls apart. To reassemble, put everything together, stand the switch on the table and grab the stem again with pliers and push it on the white slider until it seats correctly.
I didn't actually dismantle more than about 5 switches, no real need to. Also popping the fly-plates accidentally is quite hard if you mind your hands while handling the switch.

When I have it apart again for foam making, I'll make a guide, and maybe a video?

Thanks a lot

Great job

Crossfire · 24 Jan 2016, 17:29

Amazing job, great pictures too!

photekq · 24 Jan 2016, 19:10

Wow.. And I thought my beamspring was dirty!

Beautiful board though, and you've done a great job of cleaning it up.

seebart · 24 Jan 2016, 19:14

Nice chzel, now I can visually see how much work I have ahead of me.

scottc · 24 Jan 2016, 19:25

Holy shit, good job! The end-result is barely recognisable as the same keyboard in the first few pictures.

chzel · 24 Jan 2016, 19:28

Thanks guys!
At some point cleaning the switches was mind-numbing but it's worth it!
Seebart, I doubt your beamspring is as dirty, but it's likely more corroded. It's from the Dresden haul, correct?
And my restoration is not complete yet...Quite a long way to go!

seebart · 24 Jan 2016, 21:36

chzel wrote: Seebart, I doubt your beamspring is as dirty, but it's likely more corroded. It's from the Dresden haul, correct?

That is correct. I have not looked at it closely, I'm giving it a 50/50 chance so not to be disappointed later.

vivalarevolución · 27 Jan 2016, 01:37

Cleans up well, great job. There are few things more mind numbing than cleaning certain keyboards parts.

Also, Dresden haul? Sounds tasty...

Redmaus · 27 Jan 2016, 03:40

Fantastic restoration. I notice most beamsprings don't have an IBM logo.

Any custom paintjobs in mind?

Engicoder · 28 Jan 2016, 22:22

Anyone else think beam spring switches look like spark plugs? First time I saw one that was my first thought.

Madhias · 01 Feb 2016, 20:21

I really enjoyed reading and looking through this pictures! Well done, and the result was it definitely worth.

Muirium · 01 Feb 2016, 20:34

Beautiful pics! This board's in great hands.

The corroded beamspring stems reminded me of the German one now in Seebart's possession. You've got this board typing reliably now? If so, that should be greatly reassuring to Seebart, because it's further than I ever got with his 3276. I could write just about long enough to get a post out of it on DT, but it was like a stallion that hadn't quite been broken, and would throw me to the dirt and charge off with furious ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

Aaargh!

Any engineers among us who reckon Xwhatsit's controller could be programmed with a specific signal threshold *per key* instead of just one value for the whole keyboard? Calibrating beamsprings is a wild art with just the one figure. Model Fs can handle autocalibration well for the long haul, but even my American 3276 drifts and throws me if it's been a while. Which it will be, again, by the time I get it home!

chzel · 01 Feb 2016, 20:42

Thank you Madhias and Mu! It's definitely worth it, and it's a work in progress! I need to figure out what to do with paint, make a foam sheet, add a contamination shield...
Also the solenoid driver is in the mail, and I have a couple of ideas for the hatch!

Mu, I've been using it every day since, totally reliably and even played a bit of World of Tanks with it!
It calibrated beautifully, and auto-calibrates nicely too. Threshold fluctuates between 120 and 124 when auto-calibrating, but works reliably up to 140.
To be honest, I don't see a reason for the calibration to drift too much, unless the pad card is too gunky...

Muirium · 01 Feb 2016, 21:06

Give it time. I'm not sure what's up with my American layout 3276, but when I pulled it out for the first time since April, it went apeshit and I had a hard time getting it to calibrate again. I've heard others asking for the same thing: multiple thresholds so you can find a sweet spot for each switch instead of a compromise for the whole keyboard. I don't think it was even dust in this case (I store the board boxed, of course, because of those shiny white ABS caps) but beamsprings are notoriously sensitive. As you've seen while cleaning them up, the mechanism barely lifts the fly plate (or whatever its name is) from the PCB, as compared to Model F which flips the thing around with a lot more energy, and signal!

chzel · 01 Feb 2016, 21:15

I hope it doesn't drift too far...The right side is more "sensitive", starts to register at about 113. But the useful range for the board as is now is 117-145 with only one key acting up to 155.

Muirium · 01 Feb 2016, 21:19

I'd ask Xwhatsit and Mr_A500 about what it's like in a daily driver over the long haul. But neither of them's around at the moment. And A500 would find some way to make a battle out of the question, as he usually does!

A level per key would make the task of zapping ghosts and reviving dead keys so much easier. Yet I suspect there's not the memory available in Xwhatsit's controller to store so many numbers. Damn ATmega! I know he was pleased he could squeeze macro support into it.

chzel · 01 Feb 2016, 21:26

I don't know if it matters any, but the calibration stays the same whether the top is on or not.
I actually like the bare metal industrial/steampunk look of the board without the cover!

Muirium · 01 Feb 2016, 21:31

They're so boss without the case, and without the contamination shield. Especially the wedge shaped DisplayWriter style ones like yours. The 3276 and 3278 aren't quite as needlessly tall at the back!

Could absolutely see a naked beamspring being the perfect prop in a steampunk alternate future movie. The case is what really dates it back to our humdrum reality's past.

Nivlom_Drol- · 07 Dec 2018, 02:22

This really is amazing. Worth a bump.

There is a insect inside that keycap!!! WTF