Model M and the rubber sheet, better without?

[quantalume]

I recently performed a bolt mod on a 122-key Model M. After testing it for awhile, I discovered that a couple of keys would register slightly before the click. It was only noticeable if I slowly depressed the keys, not during normal typing. Upon further testing, I concluded that the problem was most likely the rubber sheet not being perfectly flat and uniform. I pulled everything apart and reassembled it without the sheet. Lo and behold, the problem is gone. Also, the tactility seems much crisper now. The feeling is somewhere between a beamspring and a Model F. I've never had a Model F apart, but I believe the hammers directly strike the PCB. This might explain why most people prefer the F over the M: the crisper feel due to the hammers hitting a solid surface. The rubber-less M does sound a bit tinnier than the F, most likely due to the smaller hammers and thinner metal backplane. I think the typing feel is just as good, however. The original IBM patent doesn't show a rubber pad. It's definitely loud now, perhaps not more so than the F, but not recommended for cubicle dwellers in either case. If I find my sound level meter, I'll do a comparison.

I wanted a 122-key "otaku" Model M in ANSI layout, and here's what it looks like. It still has 122 keys because I split the number pad zero. I've mapped comma to one side of 0 and period to the other.

M-122.jpg (59.98 KiB) Viewed 7269 times

I placed a Soarer's converter internally and put a usb socket in the back:

USB.jpg (12.74 KiB) Viewed 7269 times

Now excuse me while I spend the next couple of months trying to find the optimal key mapping for a 122-key keyboard...

[Muirium]

Nice look. And it sounds like hell! In a good way…

Someone did the same thing here with an SSK recently, if I remember. They said it sounded crazy. All the more reason to get back inside my own 122 key and try this out I suppose. Plus I want to expose some extra keys, too. Specifically split right shift and backspace.

Don't know if it's good for the hammers though, but light use is likely fine. By the way: Model F's don't use direct contact. They have a foam layer that I'm too shy to prise from my Kishsaver (as they always degrade) and I suspect the flippers are above that, while the capacitance sensors on the PCB are below.

[quantalume]

By the way: Model F's don't use direct contact. They have a foam layer that I'm too shy to prise from my Kishsaver (as they always degrade) and I suspect the flippers are above that, while the capacitance sensors on the PCB are below.

Judging from this image, it looks to me like the foam separates the top plate from the flipper housing but not the flipper from the PCB:

[7bit]

Why didn't you split up all the other keys?

Except for the Return key, each can turned into a regular key.

[quantalume]

Why didn't you split up all the other keys?

Except for the Return key, each can turned into a regular key.

I like a large backspace and large number pad enter, and I was happy to get rid of the small left shift. The only other modification I was thinking of making was a short right shift, but I wasn't sure I could find a blank key cap for that. It's not like there is a shortage of keys on this thing. On my 60%, yeah, definitely making use of every position.

[JBert]

By the way: Model F's don't use direct contact. They have a foam layer that I'm too shy to prise from my Kishsaver (as they always degrade) and I suspect the flippers are above that, while the capacitance sensors on the PCB are below.

Judging from this image, it looks to me like the foam separates the top plate from the flipper housing but not the flipper from the PCB:

Correct. It is only used to fill the space between the barrels and top plate while the hammers sit directly on the PCB. Any material between hammer and PCB might upset the capacitative sensing.

Maybe you could do without the foam, although it should slightly spread out the pressure on the plastic and hence prevent them from cracking.

[Muirium]

I stand corrected!

I assumed the point of cap sense was to reduce / remove physical contact. But instead it appears to be a way to get a cleaner signal than just slapping closed a direct metal switch.

[vivalarevolución]

Phosphorglow did the rubber mat-less SSK. I typed on it in person. That thing made some noise and had an interesting feel.

[Game Theory]

It seems the foam layer is primarily to push the barrels tight to the PCB. I used a silicone mat to replace a funking foam one on a F (but not the Kishsaver).

[7bit]

Is it really similar to Model F?

[Muirium]

Never! Well, it won't be NKRO, that I know for sure. But otherwise could be an interesting matchup. I'll check it out.

[Hypersphere]

By the way: Model F's don't use direct contact. They have a foam layer that I'm too shy to prise from my Kishsaver (as they always degrade) and I suspect the flippers are above that, while the capacitance sensors on the PCB are below.

Judging from this image, it looks to me like the foam separates the top plate from the flipper housing but not the flipper from the PCB:

Correct. It is only used to fill the space between the barrels and top plate while the hammers sit directly on the PCB. Any material between hammer and PCB might upset the capacitative sensing.

Maybe you could do without the foam, although it should slightly spread out the pressure on the plastic and hence prevent them from cracking.

Here is a post on GH with pics about taking an XT keyboard apart, cleaning, and reassembling. It appears that he cleaned out the degraded foam and did not replace it.
http://geekhack.org/index.php?topic=5492.0

In the following post on cleaning an XT keyboard, he re-used the foam, but he also did a bolt mod, using about 8 bolts.
http://www.brimstudio.org/2013/10/ibm-model-f-restore/

[quantalume]

Is it really similar to Model F?

I'm actually liking the feedback at the fingertips better than an F at this point. It seems to have a cleaner, sharper break (the point at which the spring buckles). The F is still more musical though.

[fohat]

For $10 you can order a new "blanket" (Unicomp parlance) from Unicomp that is thin white latex instead of thick black rubber.

They are very nice, and I suspect that you might get "the best of both worlds" that way.

[Hypersphere]

For $10 you can order a new "blanket" (Unicomp parlance) from Unicomp that is thin white latex instead of thick black rubber.

They are very nice, and I suspect that you might get "the best of both worlds" that way.

Interesting. I somehow missed seeing the "blanket" on their site. Too busy choosing "buttons".

[quantalume]

For $10 you can order a new "blanket" (Unicomp parlance) from Unicomp that is thin white latex instead of thick black rubber.

They are very nice, and I suspect that you might get "the best of both worlds" that way.

Interesting. I somehow missed seeing the "blanket" on their site. Too busy choosing "buttons".

I'm not seeing it on their site. I guess this is something you have to ask for?

[Hypersphere]

For $10 you can order a new "blanket" (Unicomp parlance) from Unicomp that is thin white latex instead of thick black rubber.

They are very nice, and I suspect that you might get "the best of both worlds" that way.

Interesting. I somehow missed seeing the "blanket" on their site. Too busy choosing "buttons".

I'm not seeing it on their site. I guess this is something you have to ask for?

I also could not find it on the Unicomp site. After sending a couple of questions, I got this reply:

<< Yes we do sell the rubber blanket for the 101/104 keyboards. You can purchase using the link below:
pckeyboard.com/page/Misc/MEM
In the part number field enter "Blanket part number 1403033" >>

[xwhatsit]

I stand corrected!

I assumed the point of cap sense was to reduce / remove physical contact. But instead it appears to be a way to get a cleaner signal than just slapping closed a direct metal switch.

Not so much, you're right, it does remove/reduce physical contact. The foot doesn't have to touch the actual pads; there is solder-mask between the foot and the copper. The big advantage is the foot just has to get close enough to the copper to make it trip. With actual metal-on-metal contact, you need a really good connection and a certain amount of force to make sure you've actually got a proper electrical connection, never mind corrosion etc. Hence why Cherry MX/ALPS et al are so friction-y and scrapy, and why the Model M needs a bit more force to bend the membrane to get the contacts to touch.

The extra force and physical touching means eventually the metal (well, coating of it, leading to corrosion) or membrane will wear out. With the Model F, it's just a plastic paddle tapping a very robust bit of plastic soldermask; the actual copper underneath stays sealed and dry and protected and will last practically forever.

Stuff like keybounce etc. is only a side-effect really; it still happens (my controllers have to do a small bit of debouncing just to be sure), just much less, as when bounce happens the foot is still normally close enough to the sense pads to keep it above the sense threshold.

[Muirium]

Right, gotcha. The Model M insides led me astray with that mat in the way. But it does boil down to noncontact, in the end.

I'm pretty sure that cap sense is the superior switching technology, anywhere it can be used. Metal contacts are bad, bad news.

[quantalume]

The extra force and physical touching means eventually the metal (well, coating of it, leading to corrosion) or membrane will wear out. With the Model F, it's just a plastic paddle tapping a very robust bit of plastic soldermask; the actual copper underneath stays sealed and dry and protected and will last practically forever.

Any guess as to whether the function of the rubber sheet is primarily noise damping or protection of the membrane?

[Muirium]

It is to strike a sense of fear into fiddlers like me who consider taking it apart, decades later. All that crumbling! What about my desk!!

[JBert]

Does the rubber degrade that badly compared to the foam in model F keyboards? I would think it wouldn't crumble as much, and especially not when your model M keyboards is relatively new.

[Muirium]

Whoops. Crossed wires!

The foam in Model Fs crumbles and is what I was inexplicably on about.

The rubbery mat in Model Ms stays in fairly good shape (if you can wash the spilled drinks out of it…) and generally remains intact as a single piece.

Now that I've sorted myself out, to answer Quantal's question: probably a fair old bit of both. I'm sure the membrane's lifespan is reduced without that cushioning mat. And the noise without it is also terrific!

[keokitsune]

I’m also interested in how not having the rubber mat affects the keyboard in the long run. I’ve been looking for ways to make my model m louder but not sure if I want to risk damaging the membrane down the line because of it.

I’ve also heard bolt moding and getting new springs would help so trying to figuring out if I want to take out the mat while I’m in there already.

[quantalume]

Personally, I haven't seen any real difference between a bolt-modded Model M and one with all the rivets intact. As a consequence, I now install self-tapping screws selectively where the rivets have popped when restoring Model Ms, rather than performing a full bolt mod. I do tap each rivet individually to make sure it isn't just hanging on by a thread.

Regarding the durability without the mat, I haven't had any problems with my daily driver after several months. Besides, new membranes are readily available from Unicomp. I do enjoy the extra clickiness. It's like the sound and feel of glass rods breaking beneath your fingers.