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

[pcaro]

I am "in" again too. I have taken advantage of these months to hoard flippers.

[Hypersphere]

@lot_lizard: Thanks for the clarification. Glad to know that the currently planned GB will be the curved replacement drop-in assembly.

[Techno Trousers]

9/9/19 sounds like a great day to do a big run of cold rolled steel. Really looking forward to this, lot_lizard! End game keyboards are near.

I've been on the fence about ordering a couple of full keysets from that other F project; maybe a Mopar blue and dark gray set to mix and match. I'm thinking it's a good idea now.

[adamcobabe]

I'm new to this awesome project. Really excited about it coming back from limbo. There has been so much great work done so far.

Here are my two cents. Custom flippers and barrels are low on my wish list for a potential GB. Perhaps switches could be part of a phase 1.1 or phase 2? Seems like getting the plates, and pcbs into people's hands would help generate more interest in expanding the project. While XT/AT switches are annoying to obtain, they are available second-hand and from Ellipse. I guess I favor narrowing the scope of the project to help get at least a first phase "finished". That said, I really want both an SSK and full size version. Thanks, lot_lizard. All the best to you and your family.

[lot_lizard]

Custom flippers and barrels are low on my wish list for a potential GB. Perhaps switches could be part of a phase 1.1 or phase 2?

Welcome @adamcobabe... we would share your thoughts as well. We have something viable already that is ready to produce without any additional tweaks. Unfortunately though, we are left twiddling our diddles until 9/9 when the plates can be produced by our previous fabricator. I really would like to use them again given their talents and struggles I went through to get it right with them (it’s much more complicated than any traditional F plates).

So if we can potentially squeeze something in without impacting overall schedule, we MIGHT. Plus it’s something to yik-yak about while we watch paint dry

Speaking of prior desires... anyone want to take a stab at putting Bluetooth on one of wcass’s USB daughterboard designs? If you remember, we have 4 USB options on this thing (C, B, micro, and mini). There aren’t any concerns about cap sensing or anything. They are just a clever little board he put together to expose the USB out from the controller

[DMA]

Flat PCB works with no modifications - CSSK is the living example.

[adamcobabe]

So if we can potentially squeeze something in without impacting overall schedule, we MIGHT. Plus it’s something to yik-yak about while we watch paint dry

Sounds great. Thanks for the clarification.

[lot_lizard]

Here are some renderings of the modified barrel as previously discussed... assuming we ever make our own. The goal of this design would be to create a barrel that works with a larger number of plate form factors. Remember the original plate design really only did so because the curved plane allowed key caps to be moved to alternate rows without affecting the form factor (also made cap production cheap for IBM). While we would still work perfectly well as originally intended, we could now have a buckling spring on a flat plate plane as well. The "quarter circles" have been added to the corners to allow for 3mm pass-through bolts to help tune the board. We all know that buckling spring boards have dead spots. Even my VERY best buckling spring boards have at least some inconsistency (tuned Ms from Phosphorglow included). The properly tuned Ms are by far more consistent than the best Fs, but the M is obviously just "dead" feeling on its best day by comparison. Really only even bringing up the M because this concept is in direct alignment with bolt-modding an M... but with an F.

The bottom quarter circles can be in a consistent location for two reasons. One, it had to either be on the top or bottom , and two... it enables the flipper contact area to remain as large as possible. Because the bottom quarter-circles are in a consistent location, the top has to be variable because the align of rows vertically is offset (meaning, we aren't stacked in a perfect 2x2).

Welcome the thoughts. Again, this is ONLY if we decided to actually take the leap and create our own barrels and flippers. I left one render with an original flipper so you can see how we need to reduce the contact area of the flipper by ~10% if we wanted to make this work (I have it completed, but wanted to introduce this concept in stages). @DMA... this is the bit where I am not sure if everything falls flat on its face. It means the sensitivity of the controller needs to account for that potential ~10% reduction of circuit (assuming surfaces are clean, etc). Also note, these renders are on a flat plane. For the curved plane (stock), there would be a ~2mm gap between barrel rows

2x2 barrel stack

Screen Shot 2019-03-20 at 5.21.06 PM.png (132.95 KiB) Viewed 19147 times

2x2 barrel stack from above

Screen Shot 2019-03-20 at 6.12.55 PM.png (85.45 KiB) Viewed 19147 times

2x2 barrel stack from rear

Screen Shot 2019-03-20 at 6.05.14 PM.png (126.66 KiB) Viewed 19147 times

Bottom of a 2x2 barrel stack

Screen Shot 2019-03-20 at 5.19.08 PM.png (71.99 KiB) Viewed 19147 times

NOTE... the new flipper design requires reduction of ~10% of contact area

Screen Shot 2019-03-20 at 5.38.13 PM.png (58.01 KiB) Viewed 19147 times

[wcass]

if the rows are offset by a half unit (as they are for most of of a "standard" 60%) do the quarter-rounds form round holes? It kinda looks like this would work for non-stagger and quarter-stagger but half-stagger maybe not.

[lot_lizard]

It kinda looks like this would work for non-stagger and quarter-stagger but half-stagger maybe not.

Exactly as you say. The other option is having two full half circles on the sides, but that is either impacting the flipper contact area a bit more (maybe up to 20% counting the ~2mm reduction for the row tightening) or potentially tightening the distance between the flipper pivots towards the top (probably more ideal). Let me throw that together (semi-circles towards the top) and see what we have. It’s certainly more ideal for tuning long-term if it can be every row as you say (given various form factors of today and future unknowns). I was attempting to avoid altering the flipper itself as much as possible, but if we are going to be shorter on the “lever” by a tad, then a tighter distance of the “fulcrum” legs in a similar ratio is probably ideal. Only testing will tell.

EDIT: I should also mention. If we were actually going to do this “in all corners” design (the ones posted), then we would likely get rid of the side wall around the bottom quarter circles. Then it has then upside of working with legacy flippers if the switches remain on a legacy curved plane. So these barrel housings could be intermixed with legacy housings and legacy/modified flippers could be intermixed as well with minimal perceivable difference. With the semi-circle approach, the flipper potentially gets enough of a tweak where it would no longer work with legacy boards . Long-term... no big deal, but short term it would be ideal to support intermixing (repairs to old boards, etc). Eventually the line needs to be drawn though that we should stop being a “Swiss Army knife” when making something new. The new switch, however we do it, would work with legacy boards. The debate point becomes whether the flipper and barrel be interchangeable with legacy barrels and flippers.

7CCDF763-8BAB-4E89-9405-71B71C50F1A8.png (58.01 KiB) Viewed 19085 times

[adamcobabe]

Nice.

It is a goal to have these compatible with old boards as well? Won't having both the XT and AT style "alignment notches" mean the barrels can't be used without modification? Lots more XT boards out there, but AT style seems cleaner. More importantly, AT would reduce the number of holes and the amount of cutting needed on potential plates.

EDITED

Maybe if the AT style notch was just a cylinder like the XT one, i.e. not attached to the barrel on the side, then the nub could be cut off more easily to adapt to old XT boards. Would reduce it's strength a bit, but makes it more universal.

[lot_lizard]

Maybe if the AT style notch was just a cylinder like the XT one, i.e. not attached to the barrel on the side, then the nub could be cut off more easily to adapt to old XT boards. Would reduce it's strength a bit, but makes it more universal.

Bravo... I at least like the concept. Someone might have actually mentioned this in the past, but I completely forgot if they did. The original plan was just to have us vote XT or AT, but this is an easy tweak (strength would be plenty in “AT mode”). Long run if we are creating new novel boards, it’s a bit of an eye sore to have both... but those are first world problems. Ultimately we vote if this comes to light anyway, but this is a nice third option to pick from. Well done

[wcass]

Half holes at the back is good, but might impact lateral stability of the pivot plate. How about immediately forward of the fulcrum?

Change in capacitance is generated by change in distance between the pivot plate and card pad. So it should be that the most impactful bit of the pivot plate is the bit furthest from the fulcrum and least impactful would be the closest to fulcrum.

[wcass]

As for the pin or buttress debate, i like to use neither.

The purpose of the pin or buttress is to prevent the barrel (and key cap) from rotating around the "chimney". What works just as well and i think is more aesthetically pleasing is to use something under the top plate that prevents the rectangular barrel base from rotating. The obvious thing to use is the rectangular barrel base directly left, right, above, and below. The only place you need to "design in" a block is around the perimeter of any key block and left-right of "not 1u" keys. I had the case for the CSSK milled out of a block of aluminum so i just left in place a few millimeters around the different switch clusters, left of Q, A, and Z; right of brace, quote, and slash; and between all the barrels on the bottom row. I now think this added unneeded complexity for the machinist, and could have been done with something like laser cut or printed plastic pieces.

[lot_lizard]

Now we are getting interesting!! Let me recap your Kool-Aid Mr. Jones to make sure I follow.

Let's assume we have zero ties to legacy anything. That said, everything you are proposing would be capable of being a replacement switch on ANY legacy board IF it was surrounded by another pattern of switch(es) that provided stability (say adjacent switches that were "righted" in some fashion on at least n minimum side(s)). The only switches not replaceable would be the Esc key and similar brethren IF additional parts weren't produced that "trued" the "unlocked" barrels since they have no buttress or peg lock.

We would create a switch that is tightly coupled between the barrel and the flipper (if including the semi-circle tensions previously mentioned). This means that the new barrel only works with the new flipper, but the new flipper could work with any F barrel. The new switch is held true on the XY plane by some external feature if needed. In the case of the CSSK, you have the luxury of Aluminum... so plumb is scribed into the plates themselves. In the case of anything harder than what is typically milled (say steel), we manufacture simple parts to act as the stabilizers within the "foam layer" at well-thought locations to act as guides to plumb.

This comes back to the "line in the sand" of legacy vs. novel manufactured parts. Do we abandon this concept we have up until now that parts are interoperable and backwards compatible? Why are legacy replacement parts important? Should new parts live on their own without a responsibility to the past? Being a Swiss Army knife for past and future is intriguing, but is it needed? Responses VERY welcome. We are debating producing a several thousand dollar mold. It should be of sound purpose for current and MORE importantly... future needs.

EDIT: There is zero shame in saying "I need renders before I have an opinion". This is all very abstract by two people that are in the trenches. I have a strong feeling we need a vote (we love voting :roll eyes:) on this topic if everyone really chimed in

[Techno Trousers]

I've really missed this interchange of ideas. Again, a heartfelt welcome back LL.

I really like the idea of a modernized version of the F switch mechanism that can work in a curved or flat orientation. The thing I'm wondering, though, is if shortening the pivot plate might change the amount of spring force required to pivot it, leading to the need for a different type of spring, and fundamentally changing the feel?

[Darkshado]

I don't think it is essential that *every* part be interchangeable between legacy and new for our purposes, as per the original post:

• Phase 1 is the drop-in replacement for the Model M inner assembly that turns a Model M into a functional F. The only parts remaining from the original M are the outer shell (case), and the keycaps.
• Phase 2 is to be determined, but we would be making a new and unique board from scratch using a new buckling spring switch (hopefully improved) that would support modern layouts, keycaps, case designs, etc.

I'll add a third, "bonus" one: secure a new supply of scarce replacement parts for legacy Model F boards. As seen with wcass's CSSK, just about everything outside of the assemblies can be bought (e.g. caps from Unicomp) or fabricated reasonably by enthusiasts.

So long as the barrel + spring + flipper combo is usable in legacy boards, that fulfills the objective of maintaining them. Legacy boards will in the vast majority of cases have enough leftover pin or buttress barrels to cover edges so they're not essential on new barrels.

With that said I like adamcobabe's universal "dual pin" suggestion: new plates can have both the indentation and hole, supporting all types of barrels, while using the barrels on legacy boards requires only simple flush cutters. It also avoids potential concerns new parts between the plate and PCB may have on tensioning and overall feel.

[lot_lizard]

The thing I'm wondering, though, is if shortening the pivot plate might change the amount of spring force required to pivot it, leading to the need for a different type of spring, and fundamentally changing the feel?

Bringing the tension screws forward from the pivot point like wcass mentions makes a ton of sense. When I was deciding where to put them originally, it was actually one of the first spots since it is so logical. I dismissed because I had become obsessed with making legacy flippers work in our barrel. But really... how many flippers are in the wild without corresponding number of barrels (some, but not many)? It would just mean at assembly time you needed to use the appropriate flipper for our barrel if intermixing with legacy ones.

What he is talking about doesn’t effect the pivot at all. It is the portion of the flipper that is dead space... between the pivot point directly under the barrel and the PCB contact point at the other end. It will eat into a LITTLE bit of the contact surface, but not much.

Upside with everything we are talking about, we could take a legacy barrel/flipper and machine them down to try it. Shorten the barrel and flipper length to account for a flat board with simple grinding or laser cutting. Add the semi-circles (even walled with printing and adhering) to the base of the barrel. And giving the new flipper a more “hourglass” figure to account for the tension sem-circles of the barrel. I don’t expect there to be any real perceivable changes to the tactile feel or audible feedback. Testing will make certain though. I’ll make a few of them when I do and we can pass around. Maybe “orihalcon” a piece of acrylic on the bottom so the movement is visible. Maybe send a loose one as well that you can plop in a board if you wanted a real world comparison

So long as the barrel + spring + flipper combo is usable in legacy boards, that fulfills the objective of maintaining them.

[Techno Trousers]

Upside with everything we are talking about, we could take a legacy barrel/flipper and machine them down to try it.

Duh, of course! That would be a simple way to make a proof of concept. I like the idea of having a few to make the rounds. I have an Orihalcon model F keychain, so I could send that around with the prototype, so those without one could make a direct comparison.

[E TwentyNine]

The debate point becomes whether the flipper and barrel be interchangeable with legacy barrels and flippers.

I'd think with the shortening/removal of the circular support on the bottom corners (as was suggested somewhere up there) old flippers would work fine in the new barrels.

[lot_lizard]

Busy couple of days... but I think I have a start at what @wcass mentioned previously. This is the modified barrel and flipper. Personally, this is an ideal state for legacy and future designs. It works with curved (legacy XT/AT plates) and would allow flat plate designs.

I just want to reiterate though that if we went down this path, our switch (barrel/flipper combination) WOULD work on any existing F. I would need to alter the "buttress" lock (name per @fohat) to be easily removable to work with both XT and AT legacy boards per @adamcobabe's suggestion... but that is one last easy fix. What the new barrel WON'T do is work with a legacy flipper because of the tension screw tolerance, but the new flipper would work with a legacy barrel. I am extremely content with this sacrifice, but it needs to be very clear to everyone. Reread the above if it didn't make sense (feels like a double negative, though it isn't )

Barrels overhead on a CURVED plane (ie... our MF, and other F boards)

Screen Shot 2019-03-25 at 9.05.04 PM.png (131.81 KiB) Viewed 18825 times

Barrels overhead on a FLAT plane where the rows have to be tightened by ~2mm

Screen Shot 2019-03-25 at 9.06.17 PM.png (130.97 KiB) Viewed 18825 times

Barrels on a FLAT plane (~2mm row reduction)

Screen Shot 2019-03-25 at 9.06.32 PM.png (158.12 KiB) Viewed 18825 times

Barrel by itself with pride :)

Screen Shot 2019-03-25 at 9.07.39 PM.png (98.11 KiB) Viewed 18825 times

Switch from the bottom with new flipper

Screen Shot 2019-03-25 at 9.08.27 PM.png (77.61 KiB) Viewed 18825 times

New "hourglass flipper" to allow for tension screws

Screen Shot 2019-03-25 at 9.09.57 PM.png (62.17 KiB) Viewed 18825 times

New flipper profile as seen directly from beneath

Screen Shot 2019-03-25 at 9.10.28 PM.png (50.4 KiB) Viewed 18825 times

[Techno Trousers]

To recap:

1. The new barrel and flipper combo will work together in any F, new or original.

2. The new flipper will work in a new or original barrel.

3. An original flipper will NOT work in a new barrel.

Moral of the story: to repair or replace the switch parts in an original F, you MUST buy at least a new flipper, with a new barrel being optional if you already have a legacy barrel for the location being repaired.

I think that this is a really clever design, and the single compatibility restriction of original flippers is quite reasonable. Having both the forward (flat) and backward (curved) compatibility with the new barrels is a huge bonus. I would actually like to stock up on some of these just in case of future repairs.

[E TwentyNine]

Why the notched corners at the bottom instead of the original diagonal cut? Minus the cutout, is the dimension the same, looks a little shorter.

I think with a little dremelling of the barrel an original flipper would fit, so long as you don't need the side screws (e.g. in an original board).

[lot_lizard]

@Techno... well worded. Much better than my drunken typing last evening . Exactly as you say.

Why the notched corners at the bottom instead of the original diagonal cut? Minus the cutout, is the dimension the same, looks a little shorter.

It is ~1.5mm shorter. The notched corners are just to increase the contact surface area as much as possible. There are actually a small percentage of legacy F flippers that leveraged a notched corner design as well (seems like more of them came out of “blue switch” boards iirc). We could get rid of the notch altogether, but would need to remove the walls at the bottom of the barrel as well. They do a nice job of shoring up the “aircraft hanger door” at the front of the flipper (helping to keep it plumb under pressure). You can definitely take an existing legacy flipper and bring it down to this footprint. It just requires some effort as you mentioned.

The one bit I’m unsure of is it’s ability to reliably complete the circuit. Testing will tell, but we do have less contact area now on the PCB. This was the question for @DMA, @__red__, @wcass, and company earlier about if this even works. I am quite confident in it mechanically

[lot_lizard]

Checked around with several vendors on the molding for the barrel and flipper plastics. The prices are very consistent in the states and consistent within Asia. I didn’t seek out Europe or SA, but if someone has some inside track to either locale (or anywhere for that matter) please chime in. China nets a ~20% discount compared to the states it seems. That’s a smaller gap than in years past. You obviously deal with QA issues, but it’s pretty difficult to Fraggle Rock this one up. We’ll see.

For those that haven’t ever dabbled with parts vs molds, typically manufacturers like to boil in the mold pricing in on the parts production run. Since our volumes will obviously be low, they attempt to get too much per part to justify the mold production. The mold costs themselves are relatively low if done independently. The problem is molds are designed to specific volume/pressure/heat of a given press. So if we produce molds independently, we have to be picky about the presses we might rent to make our run on.

[lot_lizard]

One of the first things I think we order is @DMA’s CommonSense controller. We will likely get a reasonable amount of them (more than we will need for our MF run) so we have on hand for future projects. Then we can design PCBs, cases, and the like around it’s tiny footprint and our connector. Would be nice to send to him and others ahead of time to make sure the producer met his desires.

On that front, it probably makes sense to have a default layout pre-loaded so people can plug and play, and then remap as needed with his utility.

Do you guys want to hash out what that key mapping should be? The first layer is pretty much a given, but the extra layers I would need some help. It isn’t anything I typically mess with on a TKL or 101-key (full size). Media keys... etc. Guidance appreciated

[DMA]

One of the first things I think we order is @DMA’s CommonSense controller. We will likely get a reasonable amount of them (more than we will need for our MF run) so we have on hand for future projects. Then we can design PCBs, cases, and the like around it’s tiny footprint and our connector.

Not much sense producing more than needed*, tbh. It's not xwhatsit, you can buy hardware in any quantity for years to come.
*) Some spares must definitely be built, but how many is debatable.

CY8CKIT-059 is significantly cheaper whatever we'll do (Weird, I know), and the only benefit is ours will be preprogrammed.
Which is a non-issue - for GBs organizer/factory will program everything, for one-offs you'll need changes (matrix size, obviously. Can't have unconnected columns on beamspring, and gaps in matrix just look weird. But not only that. There are dev modes in firmware, you'll likely need them while debugging you custom one-off PCB.) - so will end up setting up dev environment anyway.

Oh, and we can _try_ to design it to withstand those killer USB-C ports macbook pros tend to have. Whether or not that be successful - frankly, I don't know (though I have "a test stand" - lost 2 kits to it ).

As for designing around the connector.. if you're not going for this - kit's fine, and if you do - no existing form-factor is fine

So no point stockpiling the stuff. COTS hardware uber alles.

Edits: spelling.

[DMA]

The one bit I’m unsure of is it’s ability to reliably complete the circuit. Testing will tell, but we do have less contact area now on the PCB.

Nope. Contact area stays the same - get some ink on the flipper bottom, drop it onto a paper sheet. So, spring/legs side being "top", everything "above" the inked area doesn't participate in charge transfer, so you can do whatever you want with it.

And even if you halve contact area - no big deal, a via between top and bottom or the pad gets you about 4x signal boost.

Not saying "metallize model M flippers and you're golden" - that _will_ require testing. No, haven't tried that.

[Techno Trousers]

Since no one has weighed in on the "stock" CommonSense default bonus function layer configuration...

I would personally like to have the right side extra "Windows" key location between Alt and Ctrl mapped to a Fn type key, and the default layer for that could have media keys in the following locations. This would work for people used to some of the "standard" spots for these keys, and I don't see any reason why having both of these variations mapped by default would be an issue, unless someone else objects.

Function key "standard"
Fn+F5 = Play/Pause
Fn+F6 = Stop
Fn+F7 = Skip-
Fn+F8 = Skip+
Fn+F10 = Mute
Fn+F11 = Vol-
Fn+F12 = Vol+

Editing block "standard"
Fn+Insert = Play/Pause
Fn+Delete = Stop
Fn+Home = Skip+
Fn+End = Skip-
Fn+PageUp = Vol+
Fn+PageDn = Vol-

On Cooler Master keyboards, both of these variations use Fn+F9 as the (left) Windows key lockout function. That's for gaming, so I'd just as soon not map it by default to prevent accidental activation, again unless someone else feels it's really important.

[DMA]

Since no one has weighed in on the "stock" CommonSense default bonus function layer configuration...

I would personally like to have the right side extra "Windows" key location between Alt and Ctrl mapped to a Fn type key, and the default layer for that could have media keys in the following locations. This would work for people used to some of the "standard" spots for these keys, and I don't see any reason why having both of these variations mapped by default would be an issue, unless someone else objects.

You no longer have to reflash to change layout. And for things you do - FC does that, no other tools required.

Windows key lockout.. I have some spare keycodes which can be taught to do that. But currently not supported.