(Model MF) Remodeling the Model M (aka.. the Mara)

[fohat]

it makes it truly special

It seems to me that this is the project that people have been yearning and lusting and dreaming for over the years, particularly if the metal case becomes real.

The F-62/F-77 revival project was great but rather esoteric, this one sails straight up the main stream.

[lot_lizard]

It is really great seeing all the people contributing to this project

Couldn't agree more!!! The talented help is fabulous, and adds to the motivation to knock it out.


On the laser cutter front, I would like to test with some neoprene foam using the layouts that WCass put together for the legacy F's. Neoprene is quite a bit cheaper than what we will be using, and want confidence in the machine before I cut into the expensive stuff. If someone is interested in new foam for one of your legacy F's... PM me.

On that same note, this machine will be available to people wanting new foam in the future for whatever. Wodan and I have been chatting about maybe doing that as some service, with him being a Euro (international) option for everyone to save on shipping. Would be cheap since we would just buy rolls in bulk. The supplier I am looking at buying everything else from would offer neoprene in 2.3622mm thickness (which would be snug but ideal if you can muscle it together).

[idollar]

it makes it truly special

It seems to me that this is the project that people have been yearning and lusting and dreaming for over the years, particularly if the metal case becomes real.

The F-62/F-77 revival project was great but rather esoteric, this one sails straight up the main stream.

For me the big difference is in the approach, I mean cultural approach.

The F-62/F-77 revival is kind of a commercial thing that could have happened in a business environment. ellipse has managed to reduce the risk by sharing it with the community.

This one is different. It is really a community effort. Lets make a little bit of history:

It all started with the xwhatsit -> f-o-r-s-a-l-e-f58/xwhatsit-s-grand-unif ... t7993.html

Followed with wcass' great XTANT idea and implementation -> post57749.html#p57749

I had the idea of applying the concept to the FSSK and then the FEXT -> post231963.html#p231963 and its manual post301466.html#p301466

This took us to the way that lot_lizard decided to use the FSSK PCB and the thread that we are all reading.

As I said, a real community effort !

[HAL]

Yes, me, me, me
I would love a foam for my Unsaver. And while you're at it I could also use some for my Model F AT.

[lot_lizard]

I would love a foam for my Unsaver. And while you're at it I could also use some for my Model F AT.

Sounds good. $5 a piece for material cost seem reasonable? Quick math says that is close. I will just ship with everything else later. For the Unsaver, it will be for the F122 and then you can just trim down. If we would prefer 1.5mm foam instead of 2.3mm, let me know. Thicker is a little more expensive, harder to assemble, "pings" more, but would hold the barrels more securely. Thinner is the inverse with a cost of around $4 a piece.


The following is the post where WCass added all the F templates that I would just use: keyboards-f2/replacement-foam-mat-for-i ... l?#p233627

This is the link to the Snufkin post that WCass mentions in the previous post for the XT:
keyboards-f2/replacement-foam-mat-for-i ... l?#p173100

[Spaceman1200]

It is really great seeing all the people contributing to this project

Couldn't agree more!!! The talented help is fabulous, and adds to the motivation to knock it out.


On the laser cutter front, I would like to test with some neoprene foam using the layouts that WCass put together for the legacy F's. Neoprene is quite a bit cheaper than what we will be using, and want confidence in the machine before I cut into the expensive stuff. If someone is interested in new foam for one of your legacy F's... PM me.

Color me interested! You have a pm!

[phosphorglow]

I think laser cutting neoprene might be dangerous?

I remember reading about it a while ago... here we go.:

Neoprene is (or contains)* a chlorinated compound and will release hydrogen chloride (hydrochloric acid as a gas) when broken down by heat in just the same way as PVC does.

Forum link.

Also, neoprene likes to break down over time with exposure to ozone. That's why I opted for EPDM o-rings for screws. I don't know how much of an issue it would be or how long it would take. Anyhoo - I *think* EPDM foam sheets are relatively safe for laser cutting, and it doesn't break down with ozone.

PS... Phosphorglow's soldering/paste skills are from another planet. $hit is legend if any of you have ever tried.

Thanks! ^_^
(No paste though - just my trusty spool of Kester solder and some good flux! It also helps to not drink a lot of coffee beforehand. )

[lot_lizard]

I think laser cutting neoprene might be dangerous?

I actually had a lengthy discussion with my supplier about this 2 days ago. There is an ozone safe variety of neoprene I will be using (about .30 cents more a square foot) that is burn safe. I had some of the same hesitations you mentioned, and appreciate you pointing out.

(No paste though - just my trusty spool of Kester solder and some good flux! It also helps to not drink a lot of coffee beforehand. )

You are way more gifted than I have given you credit. No narcotic (upper or downer) would enable me to do that by hand. I am getting shaky with age. I blame the booze

[Techno Trousers]

Does the "ozone safe" designation mean it won't break down over time, or just that it's non toxic when burned? Are you also going to try out the EPDM foam that Phosphorglow mentioned? Does that have similar material properties as the neoprene?

This is great stuff. I'm munching my popcorn over here.

[phosphorglow]

Oh goody! Yay for specialized materials! Those are going to look so freakin' good laser cut. Cutting those on my die-cuter "toy" is invariably a painful process apparently. Consistent results have not been so... consistent.

Major kudos for acquiring and making available these epic tools and your talents to the keyboard community. You deserve a huge hug.

(Booze helps with the shaking while soldering... it's just the attention to placing components in their proper location that gets impaired... )

OH! OH!

We were tossing ideas back and forth about the whole backlighting idea, and something came to mind that might just work:

The barrels might eventually be custom injection molded, right? Using a translucent plastic, the light would be easier to diffuse and scatter more evenly. My concern with backlighting has always been aesthetics. I don't think it's proper unless you can look at it from various angles and not see individual bright spots, which is why I haven't done it yet. Clear barrels *might* be the way to go from that standpoint.

[lot_lizard]

Does the "ozone safe" designation mean it won't break down over time, or just that it's non toxic when burned? Are you also going to try out the EPDM foam that Phosphorglow mentioned? Does that have similar material properties as the neoprene?

As you say, the ozone designation actually just applies to degradation over time, but the toxicity levels are also reduced substantially as a by-product. I spent upwards of a half hour asking "are you fucking positive?!?", and the consistent answer was "YES!!!". I will be venting outdoors regardless, but am a massive advocate for the reduced human footprint. Even though this is a VERY small number for testing, it is always a concern for me. That is our ONE responsibility as humans... don't shit our own bed

The barrels might eventually be custom injection molded, right? Using a translucent plastic, the light would be easier to diffuse and scatter more evenly. My concern with backlighting has always been aesthetics. I don't think it's proper unless you can look at it from various angles and not see individual bright spots, which is why I haven't done it yet. Clear barrels *might* be the way to go from that standpoint.

Brilliant... Love. Phase 2 when we make our own barrels from scratch for sure. There are multiple moldable transparent plastics that have similar hardness values to legacy barrels. I haven't ever sent the flipper and barrel back into the lab for FTIR testing after they "lost" my original sample... But will when we think phase 2 is warming up. For now LED backlighting rests unless overlays are interesting.


EDIT: memory is fading. I only ever sent the flipper paddle that was lost. We are supposed to get a free FTIR test since they lost the paddle. I'll see if I can talk them into the barrel as well to "gain face"

[Techno Trousers]

Hell yeah, clear barrels and gold plated springs for the ultimate pimp daddy MF'n keyboard on Earth!

[fohat]

If we would prefer 1.5mm foam instead of 2.3mm, let me know. Thicker is a little more expensive, harder to assemble, "pings" more, but would hold the barrels more securely. Thinner is the inverse with a cost of around $4 a piece.

It is all down to how hard it is to re-assemble. I have tried for the thickest firmest foam that I could reasonably force together, and I seem to be marginally better at it than a lot of people, considering 1.5mm "art foam" to be moderately easy.

The neoprene that I got from wcass was quite thin and compressed down to almost nothing, but has yielded a good result so far.

[lot_lizard]

Seeing mind_prepared's Bluetooth Model M in the Great Finds thread makes me wish we could slap BT in this.

If someone wants a MASSIVE gold star, the daughter board that WCass put together is begging for BT (as soon as I saw it, it's immediately where my mind went). Its the perfect spot for it, but trickier than normal because it is already in USB protocol by the time it hits the board. So you would need to firmware it differently than a conventional HID key mapping. I did reach out to mind_prepared about it because I was impressed by his approach, but he doesn't have the "tools necessary to write a custom firmware for the BC417". He's bright btw based on his response if anyone ever needs help in that space.

To back up beyond this project, if someone did make the daughter board BT enabled, you would actually be creating something that could BT enable any xWhatsit currently around (or any USB keyboard for that matter) assuming there was room in the case for the board and battery. It would only potentially require minor tweaks in the firmware and connection considerations. That is the genius of this modular approach WCass came up with (software folks should appreciate the abstraction of controller vs comm). If we ever do dream up a better connection (BT, XBee, whatever), it only takes swapping the daughter board out. The trickiness of the xWhatsit remains untouched as the actual controller, and can receive comm upgrades over time without impact.

440723_orig.jpg (67.77 KiB) Viewed 6417 times

[phosphorglow]

Welllllllll...

Basic functionality would likely be possible. The 32U2 only has two unused pins (B3 and C2) which could probably be swapped to free up D3 and D2 for UART. But then no pins are left to check for battery level, flow control, and other useful related control (depending on the BT module used).

The expansion header could be removed to free up another four pins, but no room left for LEDs or solenoid. Or redesign it again with the '1286. Also, everything will have to be stepped up to 5V running from a battery. Battery life will likely be awful. Throw a big 'ol battery in there, though...

Then it's a matter of adding a serial protocol to the firmware.

Certainly some very clever people around here can do this. Bluetooth Model Fs and Beamsprings really need to happen. Just because.

[lot_lizard]

Hahaha... I think that answers why it hasn't been done. Didn't even think of lost extra features. Is there a better method we could enter the daughter board with? Instead of USB, we relay everything to the daughter boards that had value via headers (expansions, other controls, etc). Let the xWhatsit be just a controller, and passthrough anything that could ever be updated to the daughter board?

If it is not obvious that I only know 60% of what I should in this space... I am announcing it now. But from an abstraction point, make it so that the xWhatsit is responsible for deciding keystrokes only. Everything else lives on the daughter. Someday if we laugh at having used USB, it would be irrelevant to the xWhatsit since it had been upstreamed to the point that it didn't care the protocol. Again, this could be crazy talk, but how I would write software

EDIT:

Also, everything will have to be stepped up to 5V running from a battery. Battery life will likely be awful. Throw a big 'ol battery in there, though...

The 5.5v would be because of solenoid charging of the driver and the LED? The solenoid we could say is a no-go with BT (deal with it). The LEDs... how do other boards handle that now? I could certainly live without it (I rely on the OS or PC side firmware to yell at me if the capslock is on (which it RARELY is))

[phosphorglow]

Ooooh, there's some heavy forward thinking! Definitely beyond my current understanding of things, but I LOVE modular designs and that sounds keen.

For the sake of relative simplicity, using the AT90USB1286 in a future revision would leave ample amounts of expansion at an increased cost of about $6 and a larger footprint. Modifying the firmware *should* be simple enough for someone who knows what they're doing.

[lot_lizard]

I'm going to rephrase for my own recount... "That is a Phase 2 idea. Onward ho with Phase 1 as-is!!!"

I am making a mental note for later (I shake soldering because of booze, but recall most anything)

[Ratfink]

My gut feeling is that if we make a new controller with new capabilities, we shouldn't keep using 8-bit MCUs.

I'd love to redesign xWhatsit's controller with a shiny new ARM microcontroller, USB, and Bluetooth. I'm honestly not sure I understand the desire to make a modular keyboard controller, though. All the modules would need to communicate somehow, and even venerable interfaces like SPI or UART could become old and silly-looking. The whole thing is open-source anyway, so as long as one of us is alive when Bluetooth and USB look silly, it could be redesigned to talk directly to our brains or something. Using a board-on-board Bluetooth module would probably make sense though: it would save me from designing any RF circuitry, and could be omitted if some cheapskate doesn't want an extremely heavy wireless keyboard.

But all that will take time and effort. I'm busy with other things in life, so I can't promise that I could redesign the controller all by myself at any particular time. If someone else ends up working on it before I do, I'd love to help out!

[lot_lizard]

One thing that will "never" change is how capsense is detected (at least until someone like DMA comes along and shatters that theory). Beyond that, things (how stroke detection is communicated to the OS) evolve immensely. When you develop scalable solutions, there is seldom a need to combine functionality. Let each "layer" in an abstract model do what it does best. If capsense itself is isolated, the only updates that need to occur happen with the OS interchange. That is the idea. As WCass experienced shrinking the controller, there are a substantial number of things that can "hiccup" with Capsense. That is the reason for modularity of comm itself.

I for one would love to see what anyone could come up with downstream of Capsense. Goodness awaits.

EDIT: it's why Soarer's solutions were so brilliant. Instead of trying to fight your way through a new controller for key detection, leverage the existing controller, and only worry about turning the detection into something usable by modern standards (in his case XT/PS2/etc... to USB). The same concept could be applied to xWhatsit. Why slam everything on the single controller if the hard part of it was close to "golden" (note that I agree it has short comings and could be improved)? Capacitive sensing is just "touchy". Isolate the two (detection vs relay of detection) if it's feasible. Again... I muddle my way through this stuff when required (you guys are much more gifted on the EE sides of things), but from a generic software design standpoint, it would be the preferred approach.

[drevyek]

I have some experience with low-level development with Atmel (most familiar at the firmware level), ARM, or MSP430 (my most familiar at the transistor level).
As noted many times above, I think that the optimal method to keep the assembly current, regardless of interface standards of the time, is for it to do one thing well. If the controller can do the job of reading the keystrokes/matrix, and translate those to scancodes, then the task of converting those to whatever protocol we want becomes nearly trivial, and easy to update as time goes on. Emulating Soarer's work is effectively what we'd want to do, with the xWhatsit being the XT, and the daughter boards being the classic "Teensy". We'd be wanting something really simple on the "converter" end, like a Cortex M0, which are cheap, dead easy to program for, and run comically low power. They don't need to be too smart- just translating protocols- so things like Atmel's "heavy" processing power is really unnecessary (unless I'm missing something crucial).

However, I'm not familiar with how Bluetooth really works on an electronic/component standpoint, having never worked with it before. I'm also not too familiar with the method that USB/HID actually encodes the scancodes- in packets, or serial, or ___? Certainly should be exciting to find out.

[lot_lizard]

I think it is very safe to say we aren't doing any controller changes in Phase 1 (I love our new approach already)... but a fun idea to kick around for later.

On a more productive note, we have a first pass at a "foam" layer that I would like to try that mixes two properties of very different materials for two unique reasons. To stabilize the barrel, I would like to use silicon rubber. It is much more dense than foam, yet still compresses well. It is on the expensive side, so we will leverage only where it provides value. I have a few isolated keys that will need a slightly different cutout than what is picture (say Escape, the entire bottom row, etc), but this is the general idea. The overhang of silicon above the barrel secures the base of the next barrel(s). We will start testing somewhere in the neighborhood of 1-1.25mm thickness.

I originally would have done these as "rows" when cutting out (still might), but Wodan's single barrel foam is a very intriguing idea here since we are mixing materials in a single layer. Something to chew on...

The actual foam will be butyl rubber, and will fill all the voids where the silicon is excluded. Butyl rubber actually deadens sound better than silicon (both are excellent) if we can dial the compression of it in just right. For sure it would be cut as a single sheet, and I will start testing somewhere around 2.5mm thickness before compression. We want it substantially thicker than the silicon because it needs additional compression to be optimal.

These were two of the "layers", If you notice, they would sit entirely inside the assembly. To separate the top and bottom plate, I am going to experiment with latex film under the PCB. It can be as thin as .15mm (going up in small increments). I would like something under our thin PCB anyway to further deaden ambient vibration, and it also provides the perfect means of separating the plate assemblies (sits directly above the bottom plate).

Another "layer" I have considered is a thin membrane above the silicon and butyl to prevent any LONG term butyl decay from bonding to the top plate (our pervasive F issue with foam). Butyl breaks down at different rates based on compression (less gases can penetrate the more compressed)... so I will need to research that more.

Still needs cleanup, but a reasonable first start

EDIT: I should mention that I will be trying out several materials (Nitrile, EPDM, etc). So this is not set in stone as to the substances or thicknesses used by any means.

[Spaceman1200]

Nice work as usual, i enjoy reading up on your latest work

[Wodan]

Damn this looks impressive! I just wish I had a laser at hand large enough to cut full keyboard sheets. Love the two component philosophy. Getting more and more excited about this project. It's so crazy what can happen when the right nerds work together.

Really excited about your experiments with new materials, too. Hobby foam has worked great for me so far but it was just the easiest and cheapest material I could find

[wcass]

what he is using is this plus a USB battery (for charging a USB charge device) with an adapter between the two. As this is a kit that needs assembly, it would be easy to connect everything direct to pin instead of the actual USB connectors.

[Techno Trousers]

Ha ha, that's a heck of a daisy chain! Good thing there's plenty of room inside the phase 1 cases for those who want to try this one.

[lot_lizard]

what he is using is this plus a USB battery (for charging a USB charge device) with an adapter between the two. As this is a kit that needs assembly, it would be easy to connect everything direct to pin instead of the actual USB connectors.

It is certainly cleaner than an Ada + separate arduino. The only downside is it looks like the actual BT receiver is not included... Or am I reading that wrong (from the phone)? Certainly slick though, even if pricey.


If we can go direct to pin easy enough, does it make sense to just do that with mind_prepared's approach, and have everything (BT/USB) live on the daughter?

image.png (1.38 MiB) Viewed 6242 times

[lot_lizard]

Status update... a few exciting events this coming week:

• Ultimaker 3D printer delivers tomorrow, so we should have USB port retainer clips, stabilizer clips for the wire stabilizers (vertical and horizontal), and updated spacers for the assembly plates (haven't ever talked about, but I have better designs now that we have "hardened" the foam a bit that will prevent debris much better than what has been previously released)
• I get a demo of the LaserPro Explorer II laser cutter/engraver sometime mid-week, and planning to purchase if it performs well. This will produce the foam (all layers) and we will use for stainless steel etching (possibly more etching if it performs well on a curved surface). Again, I consider this (like the 3D printer) to be thought of as community tools going forward if ever anyone wants to experiment
• Cardboard box (packaging) prototypes are ready to view. This has been ready for a week now, but my holiday has been spoiling the fun
• After tensioning experiments with o-rings and the like, I am absolutely in love with our feel and audible levels now. It is unbelievably consistent (more than I expected). Several of the tension screws in the pictures so far provide ZERO value, and will be removed to reduce cost. Can't wait for others to try it out. It's not a beamspring switch (never will be), but might be the best F I've used at this point (and we haven't even started with the "Nike Shox" foam yet)
• Conceptually thinking about the "foam", I have decided to make the silicon portions "1 piece per barrel" like Wodan did in his thread. He did because of the size of his cutter, but it makes remarkable sense given these are the points of compression and stabilization for the board. When compressed, by being independent with tolerance on all four sides, they compress much better than being a single combined row. This provides much more consistency than if they were one strip. If one strip, the keys in the middle of the row have less room to displace the silicon, and would provide uneven tension on the top plate (the opposite of what is desirable). Wodan is either a genius, or "Flemming"-ed his way into something very novel (accidentally discovered penicillin on the reference)

I believe we are still tracking good for a "sometime in September" GB commencement (October at the latest). I am ready to wrap this fucker up (bow on it for Christmas at the latest is the hope.. expect before)

[DMA]

I am going to experiment with latex film under the PCB. It can be as thin as .15mm (going up in small increments). I would like something under our thin PCB anyway to further deaden ambient vibration, and it also provides the perfect means of separating the plate assemblies (sits directly above the bottom plate).

While you're at that - can you please check how thickness of that affects key signal (detection stability, pressed/released levels..)?

[Wodan]

.... Wodan is either a genius, or "Flemming"-ed his way into something very novel (accidentally discovered penicillin on the reference)

More like Pfizer when they discovered Viagra
I didn't know I was working on something that - with some adjustments and actual brains put into it - could give everyone a hardon.