Working out the theory behind a Model M

[berserkfan]

I was inspired by WCass’s awesome blue Model M ‘SSK’ that he created from a cut-down terminal M-122. But I don't seem to be able to find his old posts anymore on "mod a full-sized Model M into a "ghetto SSK", except as archived google stuff without pictures.

Since I also have a terminal M-122 that needs bolt modding and another long dead model M that has too many popped rivets, I am thinking of what could be done with the material I have.

I am thinking of how to make a buckling spring ergodox.

Let’s see if I understand anything. The below is purely theoretical now. I might do it eventually, but without the expectation of creating a polished, beautiful keyboard. As long as it is usable I should be incredibly happy.

1.   I should cut up my barrel plate and backplate.
2.   I have to cut the membranes
3.   I have to find some way of connecting the membranes to a teensy, so that electrical signals can be passed on
4.   I should use bolt modding to hold the newly halved barrel plate, backplate and membranes together.
5.   The keyboard casing will also have to be halved, unless I decide to discard it entirely and just use some other solution to hold together the exposed barrel frame so that the curve in the plate is supported. This should not be too hard; surely I can just cut up lots of LED spacers (they're super cheap on Taobao) and hot glue them to the bottom of the backplate? Or is there something electrically I need to solve first?
6.   After that it is a relatively simple thing to identify the matrix and assign the keys and connect this Model M ‘ergodox’ to a computer.

If I remember anything from reading so many things on geekhack, apart from general difficulties of proper execution (eg cutting the barrel plate without destroying it might be hard), the biggest issue should be the membrane. I remember reading that people have difficulties working with the membrane. Connecting a cut or modded membrane without losing electrical reliability seems pretty hard. Especially if I am going for an ergodox layout and need to do more than just cut the original membrane into two. If I want to recreate some thumb clusters presumably I need to cut more pieces of membrane and find some way of joining them.

Am just hoping that some people with real skills can take my ideas and run with them and produce a buckling spring Ergodox for people like me to follow…

[fohat.digs]

Yikes! That project would terrify me.

I might look into starting with 2 donors. For one thing, there would be some extra "margin" where you cut, so that the repair/closure might not be quite so hard. And you will need 4 "end cap" sorts of pieces instead of 2 anyway.

But, for another, if you could simply fold the membranes around the back and not cut anything it might make it all a lot easier electrically. Yes, there would be 2 controllers and some likelihood of confusion, but ultimately there is only one signal going down the cable at a time. I would guess that you might need the same model controllers, but I am not an engineer or electrician.
 

[0100010]

Probably easier to get new membranes printed I would think.  Wasn't it wcass who was working on that as well for his 60% M?

[wcass]

The Ergodox is a staggered column layout. If you want that layout, you will need to create a new barrel frame. If I were doing that, I would use XT barrels with a custom laser cut top plate. You would also need custom membranes for contact switches or PCB for capacitive switches. There is not much of a price difference between the two if you are making less than a lot. Finally, you will need a custom case - I find this to be the most problematic. You might look for someone in the forums that is interested in the project and good with CNC.

I would be happy to help with PCB or membrane design.

[0100010]

I wonder if you could cut the barrel plate from an M, consistently enough around each barrel to get ones usable with a custom plate for a project like this; versus harvesting the already individual barrels from an XT.

Consistent between thickness, height and width so they lend themselves to proper sandwiching in the layers and fitment in the layout.

[0100010]

Now that I think about it - isn't the reason one doesn't use M barrels like this is that there is nothing to index on to keep the barrel from twisting; unlike the F and XT barrels.

[berserkfan]

The Ergodox is a staggered column layout. If you want that layout, you will need to create a new barrel frame. If I were doing that, I would use XT barrels with a custom laser cut top plate. You would also need custom membranes for contact switches or PCB for capacitive switches. There is not much of a price difference between the two if you are making less than a lot. Finally, you will need a custom case - I find this to be the most problematic. You might look for someone in the forums that is interested in the project and good with CNC.

I would be happy to help with PCB or membrane design.

Ah, WCass, you’re the man!

I’m starting this thread to build and inspire a discussion. I’m technically far from able to embark on this ambitious project now (have yet to cut my first metal plate even).

But I believe that if we’re scaling a huge mountain there are many lower base camps where we must stop and acclimatize. So let’s start on a much lower mountain for now, making use of extant (pun intended) material. Long before I get my ideal keyboard I will build less than perfect but usable and good keyboards.

Let’s start at the Model F XT and Xtant project levels then. I have heaps of XT barrels and materials and I am sure many geekhackers have the same.

Let’s say I wanted to turn my miserable F XT layout into a matrix keyboard. There is a 2x5 space on the left side that I’ll leave alone, and a 20x5 space on the right side that could be turned into a big matrix. That’s enough space for a matrix-type ergodox and maybe I can put a numpad in the center between the hands where there are unused keys.

So what I’ll need would be a custom metal top plate, and custom capacitative membrane, with bolt holes for bolt modding. The bottom plate and casing will be extant, of course.

Can this be handled within the confines of your Xtant project?

Also, is the metal top plate relevant towards the capacitative sensing? Can it be replaced by another material eg a nonconductive one? I ask this because I’ve not been able to find people who can do custom metal work here in Singapore so I’m thinking of looking for people who can do acrylic instead.

[jacobolus]

What you need:

(1) Harvested barrels and flippies from a Model F
(2) Metal plate with holes cut where you want the barrels to fit, then curved on a slip roll
(3) Metal plate without holes cut in it, also curved on a slip roll
(2-3a) Probably some bolts or something to hold the two plates together; the plates would need holes for these
(4) Custom PCB with pads (which you’ll sense capacitance with) corresponding to your switch locations (I think this part takes the most work)
(5) Cap-sense controller, either lifted from an existing Model F board, or – probably easier – one of the ones recently designed by community members
(6) Keycaps harvested from whatever buckling spring keyboards match your layout or bought from Unicomp, or if those don’t cover it, maybe some custom keycap tops?
(7) Some type of case?
[8] Lots of patience and elbow grease

I think you should start by building Cherry MX or Alps (or similar) keyboards in various layouts, until you figure out exactly what layout you want, since those take a lot less effort to iterate on. The Ergodox layout is very far from optimal. [Also, I think it’ll work best if all your switches (on each hand at least) are oriented the same way. Column-stagger is fine, but keys at an angle like the Ergodox thumb keys are going to not work as well with the curved plates.

[berserkfan]

What you need:

(1) Harvested barrels and flippies from a Model F
(2) Metal plate with holes cut where you want the barrels to fit, then curved on a slip roll
(3) Metal plate without holes cut in it, also curved on a slip roll
(2-3a) Probably some bolts or something to hold the two plates together; the plates would need holes for these
(4) Custom PCB with pads (which you’ll sense capacitance with) corresponding to your switch locations (I think this part takes the most work)
(5) Cap-sense controller, either lifted from an existing Model F board, or – probably easier – one of the ones recently designed by community members
(6) Keycaps harvested from whatever buckling spring keyboards match your layout or bought from Unicomp, or if those don’t cover it, maybe some custom keycap tops?
(7) Some type of case?
[8] Lots of patience and elbow grease

I think you should start by building Cherry MX or Alps (or similar) keyboards in various layouts, until you figure out exactly what layout you want, since those take a lot less effort to iterate on. The Ergodox layout is very far from optimal. [Also, I think it’ll work best if all your switches (on each hand at least) are oriented the same way. Column-stagger is fine, but keys at an angle like the Ergodox thumb keys are going to not work as well with the curved plates.

that's a great reply.

yes, I think I'm still far from building an Xtant matrix keyboard.

but I think I've misled readers by talking about ergodox. I'm currently using a non staggered matrix keyboard and quite happy with it. There are no weirdly oriented ergodox thumb keys - I just use 1x1 for space since I always hit the same places on my space bar. Other spaces on the space bar line Row 1 have been changed to things like up and down arrows which I have found very useful. It is nice to be able to move your mouse cursor without lifting your hand from the keyboard, and the arrow keys help a lot.

So as far as I am concerned, wcass' Xtant project is already nearly where I want. I just need Wcass to design a matrix layout PCB and to find a way to get a metal top plate that's matrix.

Or, alternatively, wcass could design a slightly different pcb from his current Xtant. One with a full 4x5 matrix where the numpad is, and with the spacebar area converted into 6 barrels. A layout like this will probably not require too much redesigning/ work on his part, not to mention it won't take up too much effort on my part to drill holes in my top plate. This seems like a practical solution. It's pretty close to the 7bit PCB solution that some geekhackers are already familiar with.

If PCBs are really so cheap at 10 for $60, I don't mind paying for all 10 if no one else wants that layout. Then I only need to drill a few more holes in my metal top plate, and I have my keyboard with 1x spacebars and lots of spare keys on Row 1 for arrow and function keys. That will also make me very happy.

obviously this kind of layout is not the Ultimate End. It'll be improved on in future. For instance, other geekhackers may be able to discard the arrow keys, keep the small spacebars, and use the extra space to install a little trackball with some cutting of the XT case. The most important thing should be to open the possibility for further mods.

[0100010]

I think your 10 for $60 figure is actually, $60 for each PCB, if 10 are ordered.

[berserkfan]

I think your 10 for $60 figure is actually, $60 for each PCB, if 10 are ordered.

Ha, that totally upends my dreams.

But I also want to know whether the Model F backplate really does anything electrically, apart from physically supporting the hammers and pcb. I am thinking that if a backplate only needs to be physically supportive, then just about any material - wood, acrylic, stiff plastic can be employed.  This will be a game changer, because it again substantially reduces cost and difficulty of a Model F revival. I might even go for a wooden backplate since at least wood is easy to cut and inexpensive to buy.

[fohat.digs]

I always assumed that it was the sandwich of plates in compression that gave the Model F its exquisite feel.

Mucking around with the fundamental structure might be like building a piano and significantly cheapening the sounding board.

[berserkfan]

I always assumed that it was the sandwich of plates in compression that gave the Model F its exquisite feel.

Mucking around with the fundamental structure might be like building a piano and significantly cheapening the sounding board.

Actually if Model F could be made quieter, it would be more possible to use it in today's social work environments. That's important.

I am a trained pianist, but even I welcome the modern proliferation of electrical pianos. A pianist must be a musician at heart, not a luddite. If something opens the door to more musical possibilities, by all means.

It's not as if the piano (aka pianoforte) hasn't gone through changes in its existence. Even the name (Italian for soft and loud) comes from the fact that the pianoforte was an innovation over the earlier instruments like harpsichord which didn't have volume control. And then the foot pedals were invented which allowed dampening and sustaining effects.

The music of the Baroque period is significantly different from the music of the Romantic era precisely because the keyboard composers of the Baroque period were composing for the harpsichord, clavichord and organ. The keyboard composers of the Romantic era were composing for the pianoforte. The musicians of the earlier period such as Scarlatti had to use harmony, speed and rhythm to differentiate their music. The musicians of the Romantic era, such as Beethoven, could use volume and tone color.

As far as I am concerned if the Model F can be deliciously tactile without being so noisy, it will be more possible to see the Model F in all kinds of work environments.

[fohat.digs]

I really love the feel of the Model F. I have never typed on any other keyboard that felt so alive - nothing even comes close. Somehow, I believe that it is the 3 plates, all pressed into an intricate array of tension and compression, that reacts to each changing pressure variant from the keystrokes.

That is part of what is behind my adoption of the thick firm foam mats that cause so much grief when sliding the plates back together. While they might dampen some of the vibrations, they also distribute and equalize the internal inter-/intra-plate pressures.

The sound(s) of the Model F, on the other hand, is/are a negative for me. That is why I use pliable materials to buffer every attachment area, carefully pack and pad the cases, floss-mod, and place all the keyboards on my nice firm mats. I dare say that my Model Fs are only marginally louder than most other tactile clicky keyboards, and possibly even quieter than some.

[wcass]

I really love the feel of the Model F. I have never typed on any other keyboard that felt so alive - nothing even comes close. Somehow, I believe that it is the 3 plates, all pressed into an intricate array of tension and compression, that reacts to each changing pressure variant from the keystrokes.

My experiments would seem to agree that the feel comes from the plate sandwich.

[berserkfan]

I really love the feel of the Model F. I have never typed on any other keyboard that felt so alive - nothing even comes close. Somehow, I believe that it is the 3 plates, all pressed into an intricate array of tension and compression, that reacts to each changing pressure variant from the keystrokes.

My experiments would seem to agree that the feel comes from the plate sandwich.

Hello Wcass and Fohat

do you think something like this is possible? The barrels fit inside the spaces of any standard Cherry MX plate. If we bolt together two standard ANSI plates and one of WCass' Xtant PCBs designed with dimensions similar to standard US-ANSI, we will have the basic internals for a Model F keyboard with modern layout. Then we can combine these with extant casings eg Filco or Rosewill. The top part of the casing won't snap all the way down of course due to the thicker internals, but we could probably devise something like longer screws to hold the casing together. So everything will be held together by the tensions of bolts and screws. It will be done using existing or standard parts that OEM factories should be able to supply, and it will be in a format we all are used to.

The same could apply to TKL format keyboards.

The only thing we need to do is to have plenty of bolts to ensure that the tension is evenly spread out. But these could be pretty small much like the Model M bolt modding bolts so I don't think it will result in any major PCB redesign.

(I'm interested in standardization because of the probability that we can attain economies of scale for plate making or case buying and make these significantly cheaper)

btw there were two metal plates in my picture actually. The other one is a black Filco plate underneath the red Rosewill plate and these two sandwich the Model F barrels in between.

[0100010]

Take a pic of the red plate with the F barrels from the back.  Need to see the spacing.

[berserkfan]

Take a pic of the red plate with the F barrels from the back.  Need to see the spacing.

They are pressed together with no spaces. The key spacings are actually the same; that's why you don't have problems adapting to typing between buckling/capacitive and Cherry MX/Topre.

If there's isolated keys, they would be the modifiers which have nothing near them to prevent barrel movement. Since my plates being used in this example are Cherry MX the square holes do not prevent barrels from moving a bit. (The barrels will not be moving on a Model F plate because of the round perfectly fitted holes plus the little locking tab.) I suspect the best solution is a bolt modding one - to have more 'rivets' / bolts near the modifiers to secure them more tightly. We can't get away with just tabs at the top and bottom of the plate to fit.

[wcass]

A custom model F plate should cost about the same as a custom <fill_in_the_blank> plate - $20 to $30. There WILL be issues fitting all keys other than 1 standard unit tall and wide and you will need notches/pins to hold these barrels in place. Within the 60% cluster, there is not really room to put M style rivet holes between the barrels. Most buckling spring caps are slanted back 8 degrees and are therefore not the best caps to use on a flat keyboard (M2 and M15 caps are the exceptions).

[berserkfan]

A custom model F plate should cost about the same as a custom <fill_in_the_blank> plate - $20 to $30. There WILL be issues fitting all keys other than 1 standard unit tall and wide and you will need notches/pins to hold these barrels in place. Within the 60% cluster, there is not really room to put M style rivet holes between the barrels. Most buckling spring caps are slanted back 8 degrees and are therefore not the best caps to use on a flat keyboard (M2 and M15 caps are the exceptions).

Thanks for the detailed reply Wcass. The only issue I have is your estimate that a custom plate will only cost $20-30. I have heard more than once that a custom plate is cheap or easy to get, but I haven't been able to find any real person or business who will cut me a switch moddable plate at that kind of prices. (Or even cut me a switch moddable plate to begin with.) Once you tell them you only want two plates, they don't even bother to answer emails anymore.

And in my case I can't expect many people to want the same layouts as me. You won't believe how (few) standard layout keyboards I have.

OK geekhackers, those in the know, please tell me how I can get a genuinely custom (ie not group buy US ANSI) switch moddable plate cut for under $45 USD in USA or under $60 in Singapore with maximum purchase of only 2 plates. Please give real names, real links, real email addresses that are actually answered, not any vague 'I know it's cheap to get a custom plate cut' comment. 

[wcass]

I ran into the same issue; there are a bunch of shops in south Florida that would quote me in the hundreds of dollars for a single plate. Then I heard that The_Beast has a guy near him was cutting custom plates for reasonable; that is where I got my plate. It took a little more time than I had expected, but that might be because of the design was so different from anything else they had done. Also they do not have a slip roll to put a curve in the plate, but I found a HVAC shop local that did that for free.

You will need to get your design into DXF format for it to be cut.

[berserkfan]

I ran into the same issue; there are a bunch of shops in south Florida that would quote me in the hundreds of dollars for a single plate. Then I heard that The_Beast has a guy near him was cutting custom plates for reasonable; that is where I got my plate. It took a little more time than I had expected, but that might be because of the design was so different from anything else they had done. Also they do not have a slip roll to put a curve in the plate, but I found a HVAC shop local that did that for free.

You will need to get your design into DXF format for it to be cut.

WCass, I heard it costs $90 at The_beast's shop if you are getting just one plate.

[wcass]

It was less for my order, but that was over a year ago when demand was low and the service was just starting.

I ran into a similar thing recently with PCB production. To get one XTant PCB was $100 (+ $30 for shipping); to get 10 XTant PCB was $139 (+ $35 for shipping). Economics of scale is a real thing.

[berserkfan]

It was less for my order, but that was over a year ago when demand was low and the service was just starting.

I ran into a similar thing recently with PCB production. To get one XTant PCB was $100 (+ $30 for shipping); to get 10 XTant PCB was $139 (+ $35 for shipping). Economics of scale is a real thing.

OK, so you mean a PCB, even for Model F, is really not that expensive but it is the metal plate that is so costly? Because I think there will be demand for an Xtant matrix PCB. At the above prices I'm even willing to buy 4 PCBs to help meet MOQ of 10. But at $90 or something per plate, I will only be building one matrix xtant. And since I suspect at most 5 or 6 people will be willing to buy an Xtant matrix PCB I don't think plate prices will drop much. And the cost again will discourage people from trying an xtant matrix layout.

I am getting quite tired of desoldering and plate modding. If I can find my ideal layout I would dump everything and stick with one keyboard. But I think one precondition for an ideal layout for me is that I have to design my own PCB and be mentally prepared to be the only person who buys it.

The biggest single reason why I use matrix layout now is because I need the flexibility to change my layout.

[jacobolus]

I’ll repeat myself:

I think you should start by building Cherry MX or Alps (or similar) keyboards in various layouts (perhaps using laser-cut acrylic plate/case and direct wiring?), until you figure out exactly what layout you want, since those take a lot less effort to iterate on.

An additional benefit will be that if you can explain your layout and empirically demonstrate its advantages, other people are much more likely to join in on a group buy with you for a custom Model F of that layout.

[berserkfan]

I’ll repeat myself:

I think you should start by building Cherry MX or Alps (or similar) keyboards in various layouts (perhaps using laser-cut acrylic plate/case and direct wiring?), until you figure out exactly what layout you want, since those take a lot less effort to iterate on.

An additional benefit will be that if you can explain your layout and empirically demonstrate its advantages, other people are much more likely to join in on a group buy with you for a custom Model F of that layout.

direct wiring and less effort do not go in the same sentence 

[jacobolus]

If you’re making a single keyboard to test the layout, direct wiring is absolutely faster (also cheaper) than designing a PCB, waiting for someone to send it to you, and then soldering everything into it.

Direct wiring really isn’t so bad, if you make sure to batch operations, and do it in organized lowpoly style. I’d guess it takes about twice as long as hand-soldering SMD diodes and switches to a PCB.

If your goal is to experiment with layouts, you can iterate really fast when you’re direct wiring.

[berserkfan]

If you’re making a single keyboard to test the layout, direct wiring is absolutely faster (also cheaper) than designing a PCB, waiting for someone to send it to you, and then soldering everything into it.

Direct wiring really isn’t so bad, if you make sure to batch operations, and do it in organized lowpoly style. I’d guess it takes about twice as long as hand-soldering SMD diodes and switches to a PCB.

If your goal is to experiment with layouts, you can iterate really fast when you’re direct wiring.

I had a good laugh, thank you!

Our frames of reference were not the same, no wonder! I haven't hand soldered SMDs before.

But can you give me some idea about an acrylic plate cost? Now I'm thinking of it because apparently it might be cheaper and I am tired of dremeling.