Model MF - GB CLOSED

[Techno Trousers]

This is all great news, lot_lizard. I really appreciate you keeping us up to date with developments. Best of all, no three week delay for lunar new year!

[just_add_coffee]

I just learned of this project. Looks like I'm just in time for Round 2.

[lot_lizard]

I just learned of this project. Looks like I'm just in time for Round 2.

I hate to be the bearer of bad news, but there won't be a round 2 (at least a remake of what has happened here). We have had a great following, and with that, our odd little mission statement has gotten lost in the posts (love them, and more the merrier btw ).

We are working towards making something new and novel from scratch with a refined F switch. That is what we have been referring to as Phase 2. This first phase that just wrapped up was to see if we could slam F parts in an M body. The irony, it is MUCH harder (at least in terms of tolerances) than phase 2 where we choose the full envelope. Phase 2 will focus on making an F assembly that is cheaper to produce, more modern in appearance, and refining the switch itself so that is ALWAYS consistent (less ping and twang). Not sure if that is actually at a lower dB level yet, but think more "crisp".

I am having some spares made of this Phase 1 round we just did. But they will not be at the same pricing as the previous buy. Not trying to be Debbie Downer, but my A.D.D. needs to keep this ball rolling to stay actively engaged, and additional funding helps us with the mold costs in Phase 2

What's the expected durability of that when kept under pressure in our keyboards?

Sorry that I missed this before. It's a great question. The truth is, we actually don't know the real world shelf life of silicone elastomers. There are classes that grade the expected "shelf life" and "life in service" of given polymers and elastic polymers, but until they are proven, they are just theory. These firms/groups (like BSI) get very flaky when it comes to declaring unknowns for obvious reasons (credibility). They choose more to say that something should be "inspected on a given interval of" X versus declaring actual life. TONS of factors go into real world life. Humidity, temperature, UV, exposure to chemicals, etc. The expectation is that silicone will outlive you before it degrades, but that is difficult to say real life.


Both the silicone rubber and silicone foam/sponge are made of the same materials. The only difference is the density, which in this case is controlled by the amount of air still trapped inside as closed cells. There is a process called degassing that polymers are put through. For the foam, the degassing is much less than it would be for the rubber. The amount of degassing before solidification controls the hardness (shore value).

[seebart]

Agreed, with your crafty skills a repeat of the same project would be a waste of time, note that I didn't even get a MF. With the combined lot_lizard / DMA skills I could see a various possible M/F even Beamspring projects.

[just_add_coffee]

Phase 2 will focus on making an F assembly that is cheaper to produce, more modern in appearance, and refining the switch itself so that is ALWAYS consistent (less ping and twang). Not sure if that is actually at a lower dB level yet, but think more "crisp".

Onwards and upwards isn't bad news really.

And it's exciting to see that there's an energetic community that is not only interested in keeping the Model F alive, but is also intent on improving it and making it modern and relevant.

Count me onboard for whatever comes next!

[seebart]

Phase 2 will focus on making an F assembly that is cheaper to produce, more modern in appearance, and refining the switch itself so that is ALWAYS consistent (less ping and twang). Not sure if that is actually at a lower dB level yet, but think more "crisp".

Onwards and upwards isn't bad news really.

And it's exciting to see that there's an energetic community that is not only interested in keeping the Model F alive, but is also intent on improving it and making it modern and relevant.

Count me onboard for whatever comes next!

Model F is alive anyway, almost indestructible if you ask me. That's why we need a modern F. There is one major F project already of course.

[just_add_coffee]

Model F is alive anyway, almost indestructible if you ask me. That's why we need a modern F. There is one major F project already of course.

I'm a brand new Model F convert and so it's awesome to have an opportunity to ask this ... What constitutes a "modern F?"

(and if anyone else has some thoughts on this, I'd be interested in reading them!)

[seebart]

Model F is alive anyway, almost indestructible if you ask me. That's why we need a modern F. There is one major F project already of course.

I'm a brand new Model F convert and so it's awesome to have an opportunity to ask this ... What constitutes a "modern F?"

(and if anyone else has some thoughts on this, I'd be interested in reading them!)

Since we don't have one yet: no idea! No seriously, it would have a modern layout, modern ergonomics and connectivity.

This is more of a reproduction:

group-buys-f50/brand-new-f62-kishsaver- ... 11046.html

[DMA]

group-buys-f50/brand-new-f62-kishsaver- ... -this-year[/b]-t11046.html

OH THE IRONY

[seebart]

group-buys-f50/brand-new-f62-kishsaver- ... -this-year[/b]-t11046.html

OH THE IRONY

An elliptical irony.

[just_add_coffee]

... modern layout, modern ergonomics and connectivity.

Hmmm ... The buckling spring Bluetooth Ergodox!

This is more of a reproduction:

group-buys-f50/brand-new-f62-kishsaver- ... 11046.html

I almost joined that group buy, but ended up getting a Model F XT and then an AT two weeks later.

[Techno Trousers]

To me, the single biggest test of "modernness" in a keyboard is using the well established ANSI and/or ISO layouts that have been used, with rare exceptions, since the mid-1980s. Having the muscle memory for those layouts was one of the biggest excitement factors of the MF for me: an SSK, but with the crisp switches of a model F is the perfect blend of both worlds.

I was tempted by the Ellipse project, but even the F77 is a bit of an odd layout for the nav cluster. If I were a lover of the 60% layout I would have considered the F62 further.

I think phase 2 is going to be really cool. We've already got people talking about 60% boards, ergonomic boards... The sky's the limit!

[XMIT]

After much soul searching I've decided not to buy one of Ellipse's F62 or F77 boards after all. It is a good project and I wish him well, but I'm more excited to see what lot_lizard can do and will be throwing my support there.

I guess my otherwise complete collection will simply have to do without a Kishsaver for now. Can't win them all. Though who knows, maybe I'll get lucky and find an original one some day.

[seebart]

After much soul searching I've decided not to buy one of Ellipse's F62 or F77 boards after all. It is a good project and I wish him well, but I'm more excited to see what lot_lizard can do and will be throwing my support there.

I guess my otherwise complete collection will simply have to do without a Kishsaver for now. Can't win them all. Though who knows, maybe I'll get lucky and find an original one some day.

This is one instance where I will not post my opinion since it may be taken the wrong way towards Ellipse's project. Being where you are it is not totally unrealistic for you to find a F62 / F77.

[just_add_coffee]

To me, the single biggest test of "modernness" in a keyboard is using the well established ANSI and/or ISO layouts that have been used, with rare exceptions, since the mid-1980s. Having the muscle memory for those layouts was one of the biggest excitement factors of the MF for me: an SSK, but with the crisp switches of a model F is the perfect blend of both worlds.

So what you're saying is that you're not interested in my 4-key, 167-layer RGB board group buy?

To me, a "modern" keyboard should be supremely customizable in both function and appearance, the guts should be hackable and repairable by anyone willing to take the time to learn and it shouldn't require many (or any) sophisticated tools, and the software/firmware should be open-source.

The only limitations should be imagination. At least, I can't imagine any other limitations.

[DMA]

Once it is confirmed, we'll order the PCBs and controllers as well (along with the plates).
--skip--
It is looking more like the middle of March given my work schedule is going to throw a one week delay in here coming up.

Since we're slipping a bit and controllers/PCBs aren't yet ordered - it may be worth reconsidering CommonSense as a controller.
Should not introduce much of additional delay.

Right now:
* Firmware can pretend to be a single-layer NKRO keyboard with firmware update capability.

* Hostside utility (FligtController) can tune activation thresholds - but only to "current floor +X counts" for now. Noise floor is stored in firmware per cell though - only interface is needed.
** FlightController can edit layout too - but no layers.

* __red__ was able to reproduce the system - he built firmware and even compiled host controller on linux (something I just never tried) in about 4 hours IIRC. So it's at least somewhat reproducible.

Remains to achieve bare minimum (which is "working standard keyboard" to me, don't know about you guys):
* Experiment with series resistors in drive/sense lines, select the best. I have 100R, 1K and 10K. About 1 day, will do ~Feb 25th.
(Who am I kidding, saving 32 components is OVERWHELMING argument for omitting resistors completely. But I don't want to screw up this one, this MUST be tested to confirm the effect is negligible. OR NOT.)

* PCB layout for QFN-68 package. About 2 days +2 weeks turnaround from oshpark (proto PCB turned out great from the first try, and I'm optimistic! )
Want to do it myself, matters of honor.

So, not that much is left.

Expected PCB size is "connector width x 0.5 inch" - because QFN is 8x8mm and there's virtually no other components - though I'll try to make it as small as possibble.
Bonus points to make PCB universal for all IBM capacitive keyboards (achievable with a breakaway PCB portion) - which will leave our beamspring friends happier as they won't be held hostage by their small numbers.
Hardware itself should be universal - I expect it to be, just need to confirm with those beamsprings lot_lizard had graciously loaned me.

BoM is expected to be ~$5 at 100 units, ~$4@1kU for the cheap version, ~$10/$9 for the expensive one (PCBs are the same, 1 extra 1uF cap and dual-ADC chip).
I would've wanted to get my scope and soldering iron money back running this via separate GB, but that's the poor guy in me talking, I just need to shut him up. That $1k won't change anything for me, and the joy good tools bring when you're using them is a good investment by itself

One other thing: solenoid/LED interface cards interface pinout must be changed. Right now if you turn the connector 180 degrees - you're bound to fry something, because you're swapping VBUS and ground. And the connector is not keyed. There must either be a key which will ensure correct positioning or the pinout which will at least connect VBUS to some relatively safe pin if not connected correctly.

[Techno Trousers]

Since we're slipping a bit and controllers/PCBs aren't yet ordered - it may be worth reconsidering CommonSense as a controller.
Should not introduce much of additional delay.

Hmm. Now that's an interesting idea. I think one issue might be the form factor that lot_lizard had settled on for the slip-on controller plus outboard USB interface. Would there be a lot of rework needed for that?

If CommonSense can accommodate the layout, and works as well or better than xwhatsit in its core capacitive sensing function, then I'd be willing to give it a shot even if the customization software needs further refinement. If CommonSense is going to be the controller of choice for the MF project going forward, this could be an opportunity to field test it.

I suppose in a worst-case scenario, we could do a second order of xwhatsit controllers, right? They'd be super easy to swap in and out with the slip-on connector.

[wcass]

I'm 100% ready to help make CommonSense a drop in alternative to the xwhatsit. I also agree that a different pin-out would have been wiser, but there is also something to be said for going with a accepted standard. If resistors improve performance, i have no problem dropping a few down. Resistor arrays would keep chip count down.

For your prototypes, consider trying "hot plate" soldering. You can get a stencil cheap from OshStencils.

[lot_lizard]

My personal vote would be to have the best controller option we can, but timing is less of a liability to me that it might be to some. A couple of weeks in the grand scheme of this type of buy is nothing given we are very close to achieving our original objectives. That said... it's a "group buy", and I would like to make it a vote so that everyone gets a voice in the matter. I would feel different if we were planning on running this buy again, but Phase 1 is a one-timer.

I have had to extend my work trip through the weekend. The next couple of weeks are dicey for me on a few fronts anyway. Maybe we give this a 1-2 week burn-in to help DMA, red, WCass, and anyone else that can contribute to get their heads around it fully?

Feedback is welcomed. Once we see what the net of that really is... then we vote? In the meantime, I will make sure to send DMA a couple of the prototype PCBs as well.

[Techno Trousers]

I agree with you 100%, lot_lizard. I'm a big believer in doing it right, over doing it fast. Doubly so when it's a small delay. Still, putting it up to a vote seems prudent since we've all got money invested in this project.

[fohat]

Now I am starting to feel stupid. Can someone post a 1-paragraph explanation of the differences?

Soldering makes me feel more secure sometimes, if it is with the "normal" sized connections, and I doubt that I will change anything much once I build it. But long-term durability and maximum compatibility are most important to me.

[Techno Trousers]

I'm not an EE, and honestly it's kind of hard for me to follow DMA's posts (but I love that he's working on this). As I understand the big picture, xwhatsit is fairly complex as far as number of chips required, and only lets you set a single activation threshold across the entire board. I believe CommonSense was started to take advantage of increased processor power to simplify the controller, making it less expensive, and allow for per-key activation threshold levels.

Did I miss anything, guys?

[Ir0n]

Hmm would the commonsense board be slip on as well?

One thing I like about the xwhatsit is the capsense software.. It's more useful and more clean than 99% of the "gamer" boards software and I use it a lot. (I like my macros..)

Does commonsense use something similar?

Forgive me for all the questions I haven't seen any of the commonsense stuff until now.

[DMA]

Hmm would the commonsense board be slip on as well?

Supposed to be drop-in replacement. I'd stive for a smaller PCB though - PCBs should not be larger than they absolutely need to be.

One thing I like about the xwhatsit is the capsense software.. It's more useful and more clean than 99% of the "gamer" boards software and I use it a lot. (I like my macros..)

That would temporary be a problem - firmware currently only supports a single-layer, no-macros layout.
This is a software problem though - you can upgrade software in the field, but not hardware (well, you can buy another controller and swap). And there's more people here who can do software.

to take advantage of increased processor power to simplify the controller

Not "increased processor power" - rather "advantages of contemporary hardware". All those small ICs are now available as a software-configurable building blocks on the MCU chip. So you can, for example, make a displaywriter controller from 3178 one just by flashing a different firmware.

[__red__]

All those small ICs are now available as a software-configurable building blocks on the MCU chip. So you can, for example, make a displaywriter controller from 3178 one just by flashing a different firmware.

This is the bit that everyone should read and understand and this is why PSOC is our future imho.

You know all those complex analogue hardware schematics you see on xwhatsit controllers et al?

Doesn't exist in DMA's PSOC design.

All the hardware is on-chip and configurable in firmware.

No PCB changes. No hardware futzing, just code changes.

For those familiar with what an FPGA is... a PSOC is basically an analogue FPGA with embedded microcontroller.

[Techno Trousers]

Thanks, guys. It definitely sounds to me like CommonSense is worth waiting for, but of course we'll all abide by the democratic vote when lot_lizard runs it.

Just to probe a little more on the software side: I myself don't plan to do extra nonstandard layers, but for those who do, what sort of rough time frame do we anticipate before an xwhatsit-level configuration program is ready?

[__red__]

Just to probe a little more on the software side: I myself don't plan to do extra nonstandard layers, but for those who do, what sort of rough time frame do we anticipate before an xwhatsit-level configuration program is ready?

That is a question for DMA.

I have the currently published firmware running here but I'm in "matrix calibration mode" as opposed to keyboard mode since I'm experimenting with PCB design / routing etc...

[DMA]

but of course we'll all abide by the democratic vote

Yeah, "group" in "group buy" means that group decides, that's why I've asked

Just to probe a little more on the software side: I myself don't plan to do extra nonstandard layers, but for those who do, what sort of rough time frame do we anticipate before an xwhatsit-level configuration program is ready?

Like "Initial (buggy, but usable by others) version in a month after I get the actual keyboard in my hands". It's quite hard to find motivation to write code for the thing that doesn't exist. Gratification is too un-instant!

Sure, there's F122 at my desk - but turns out I can't use the keyboard which doesn't have Esc key and proper arrows, so it's more of an LCR meter than an input device. As a result I have sensing part pretty polished by now - because watching pretty lines on the scope is fun - but the actual keyboard part is not there like at all (even though the firmware was initially TMK-based).

[mana]

Is there any update on those MX to buckling spring adapter portion of the project? While I didn't exactly buy one of these and went for one of Ellipse's F62's, I think the endgame for me would be a buckling spring F with sculpted spherical keycaps...

[lot_lizard]

Is there any update on those MX to buckling spring adapter portion of the project?

Not too much to report on that front other than I'm using the 1u as the Esc for the wingnut. Kinda fun.


The issue really boiled down to the Capslock. I could make everything else work, but since that one key was going to defy physics... questioned whether even made sense to proceed. We would need 6 unique adapter molds, and you would need to get the Dremel out for that Capslock.

Going forward where we make something new and novel from scratch, we would just target making use of the adapters (potentially just 3), BUT the new novel board would NOT be backwards compatible to IBM caps.

All that said, we would make the 1u backwards compatible, but that would probably be it. Fun for an Esc key, but that's about it