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

User avatar
lot_lizard

06 Jun 2016, 18:49

elecplus wrote: So, are you guys possibly wanting Model M cases? I have some 122 keys that I have used the caps, so all that is left is the case and the "guts".
Actually if you have, or can find, 101-keys with undesirable connectors (terminals) for cheap. That is probably the target (the shell and key caps). It's a shame they never had terminal space savers, or I think you might be really busy :). There is no reason the M122 couldn't be done at some point, but I have a feeling it would be a while before we did this process for it

User avatar
Muirium
µ

06 Jun 2016, 18:54

Oh, you think you know IBM? Think again! Halvar has a Terminal SSK:

other-external-f66/great-interesting-fi ... K%20Halvar

The reason they used all those digits in their part numbers was because they needed them!

Anyway, I agree that fullsize Model Ms of any connector are a good donor for your work. IBM's caps are better than Unicomp's, not least. M122's are still less desirable, however. The F122 brigade isn't big enough, or starved enough, to be worth all that design and prototype work.

DMA

06 Jun 2016, 19:29

andrewjoy wrote: As far as i know the original F controllers ( and beamspring too i think) do have some limited auto calibration.

As for the rest of your points i do not know.

Support for Xwhatsit's is important ! A new controller may be a good idea in the long run , but why re-invent the wheel when we have something that works now ?
I actually come across the documentation for russian clone of model F. It's fairly faithful reproduction, the only difference is that because Ministerstvo Elektronnoy Promyshlennosti wasn't good at producing custom ICs they made that small aluminum cube out of standard components.
There calibration is, indeed, there. It consists of one potentiometer, set at the factory and painted over so vibration doesn't move it.

The new controller is not a part of the phase one. Enough new things there as it is.
The controller is for the next part, where changing layours and MX keycaps happen. It would lower requirements to the PCB layout because it will have per-key thresholds (and possibly general EMI awareness). This is possible to do with xwhatsit, btw - would just make scanning slower because you will need to adjust DAC every scan - and scan every row separately.

User avatar
lot_lizard

07 Jun 2016, 16:40

As promised... the comparison of individual parts. Apologize in advance for the length, but it is a lengthy topic.

The goal... understand what parts within an F had the greatest impact on the PERCEIVED difference between the M and F. I separated each of these sections based on the core part being evaluated, and then break down the different variables that could be introduced to prove/disprove the merit of that part. This was done very methodically, but there are things I am sure were overlooked (a ton of options). Most of this may seem obvious to many of you that are familiar with these boards... but hopefully there are a couple of points along the way that at least harden assumptions you have always made. There is quite a bit more that could be written here, but this seemed like the most revealing of the lot.

Springs
Spoiler:
This is actually where the most physics lives in all of this, but I will try to keep it interesting (things like Young's modulus and Hooke's law get boring quick :) ). So to surmise, the M and the F spring are VERY similar, with the exception of coil counts. As a free length spring undergoes uni-axial compression, the higher the coil count, the less force required. Here it is very slight, and barely perceivable (but still there). Where the real difference lies... as the helical spring continues to compress, deformation from the stress causes deflection, and eventually deflection stress turns into maximum shear stress (the buckle). When a spring with higher coils "buckles", the event is much more violent (a sharper snap), and is more consistent at the point which this occurs.

Rephrase... The F spring feels lighter because in fact it is (barely), but the shear stress at the point of buckling is a "livelier" event (more of a ping) due to coil counts. Others in the past have mentioned the M spring is thicker than an F. I found this to be untrue (tested from several boards). The free length (under no compression) is greater on the F, but that has little impact of the dynamics of how the spring behaves. Remember... these springs are never at their free length when a keystem is in place. Also, the the spring always strikes the front of the barrel sleeve (the part facing you while typing) which is a contributing factor to noise

The following are the measured attributes of the F spring:
  • SPRING_WIRE_DIAMETER = .35; // .013 inch
  • SPRING_LENGTH = 18.0; // .709 inch
  • SPRING_OUTSIDE_DIAMETER = 2.67; // .105 inch
  • SPRING_COIL_SPACING = .25; // .01 inch
Flippers vs pivot plates
Spoiler:
This is the single biggest difference between the two models. As mentioned earlier, there are 3 different variants of the model F flipper, and each was tested independently. As far as I am aware, there is only one model M pivot plate design (though the colors and likely materials vary greatly). I tried each flipper and the pivot plate with:
  • Both styles of F barrel (peg and slot stabilizer)
  • Model M barrel frame
  • Model F spring
  • Model M spring
  • Striking the model F PCB (with and without blanket)
  • Striking the model M membrane (with and without blanket)
  • Thicker and thinner steel back plates
In each scenario, the F has a consistent crisper point of contact. As I mentioned earlier though, in every scenario, I could not differentiate between F flippers. I will discuss each of these other variables more in detail below, but the take-away is that the M and F flipper/plate actuated almost identically, but the audible feedback was dramatically different. If anything, the M's pivot plate is actually a crisper snap of the spring itself, but the lack of surface area of "the paddle" contacting the PCB/membrane makes for a much more muted experience.
Barrels vs barrel frame
Spoiler:
What seems trivial is actually the part that has the biggest impact. For those of us used to toggling back and forth between the M and the F, the scratchy resonance of the M is my biggest annoyance. At first I thought this was solely related to the barrel independence (which has impact), but the bigger variable is related to the quality of plastics used. The M barrel frame is a much softer material, but at the same time also has a lower tensile strength (in other words it scratches and chips much easier... aka "shittier"). Couple this with the connected barrels (all one unit), and you have a reverberation nightmare.

For the F barrels, I did not have the two varieties lab tested to confirm, but using a quick burn test (sorry XT/AT), I can confirm they are almost an identical material, and all physical testing (with moving parts and a basic gouge test) would confirm. Before we proceed with any type of barrel production, we will want to FTIR test to get the grade of ABS used (there are a ton), to make sure we are getting something of at least as high of quality.

The sound against an F barrel is dramatically different. The friction as the keystem slides down the sleeve, the sound of spring impacting the barrel, and as the flipper make contact back to the roof of the barrel plate on key release. All of these events against a higher grade of plastic make for a world of improvement. I tried each barrel and the barrel frame with:
  • Model F spring
  • Model M spring
  • Model F flipper (all three)
  • Model M pivot plate
  • Striking the model F PCB (with and without blanket)
  • Striking the model M membrane (with and without blanket)
Top plate isolation (foam vs o-rings)
Spoiler:
This really only applies to the independent F barrels nested within the metal top plate. But the goal of this exercise was to test sound deadening, sharpness, and feel. The metal top plate alone is a substantial improvement over the M barrel frame (so not calling that out separately), but some sort of material is needed between the barrel and top plate to eliminate the tolerance for the cutouts. I have tried without, and each key is literally more wobbly than a beamspring, but with a cheap feel (it's a terrible experience).

I have tried both the foam and o-rings in previous rebuild projects in the past few years, and the feel is almost identical. Both do a very good job of eliminating any vibration in the barrel, and eliminate any tolerance between the barrel and top plate when assembled. The foam (regardless of thickness or material used... within reason) seems to consistently behave similar. Only when using VERY dense foam (think memory foam), can I notice any additional improvement. There is an obvious trade-off here though with difficultly of reassembling the board (especially with those F plate tabs). The o-rings have a dramatic variance based on the thickness of the material used (thicker the better), but suffers the same assembly trade-off.

Regardless of thicknesses of each material, there is a significant advantage to foam's ability to lower the audible levels. It's really not even close. I appreciate loudness of a keyboard as much as anyone, but muffling resonance with foam provides a much cleaner feedback to me. I like hearing the click/ping, but have do desire to hear the additional rattle... foam does a fantastic job of making the experience "crisp"

Foam has a massive downside though... rate of decay. The polymers used now are exponentially better than they were in the 80's, but regardless, decay will happen. This is where a denser foam is actually more ideal. The rate of decay for foam is directly tied to the material used and the size of the "bubbles". The larger the bubbles, the faster the decay. Because of the decay, I have always wanted to try a THICK rubber (latex or neoprene) blanket, but never had made time. Before we order anything for this project, it would probably be wise to try that approach out. The shelf life of rubber is much longer, but does crystallize and become brittle eventually (even newer grades). This is a better problem than foam's eventual state, but still far from ideal
Back plate thickness
Spoiler:
This is a bit of a red herring in our world. I tested new model M back plates as compared to older model (thicker) F plates, and the difference was minute (when tested with the F top plate). Regardless of the back plate thickness used, the strength of the assembly remains similar (and I type hard), and the audible feedback was negligible. Any combination of PCB, flipper, spring, etc... produced very similar results.

When using the M barrel frame as the top plate, the differences in bottom plate thicknesses became quickly noticeable. With the extra resonance of the M's barrel frame, the thinner back plate becomes much more "pingy"... and not in a good way (think "clank").

Moral... if using a traditional M, go for the thicker back plate of the earlier models. For the F, the only perceivable upside is the strength during assembly and the added weight of the board.
PCB vs membrane (including and excluding the blanket)
Spoiler:
This had less impact than I was expecting. For the bottom plate blanket, I used the traditional M blanket (black rubber impregnated fabric) and white latex from Unicomp.

Regardless of striking the PCB (with or without a blanket) or membrane (including blanket) with the F flipper, the sound given off was eerily similar. The membrane sheets are very thin, and I expected a more muted "thud" than "clack", but it was almost an identical experience. As expected, there was ZERO impact when using the PCB with or without a backing blanket.

When you remove the backing blanket from the membrane though, and the flipper strikes the membrane sitting directly on top of the back plate, it is LOUD. Think "bullets hitting the side of a tank you are inside of" loud. Obviously we would never do this since the flippers and membranes are completely incompatible, but a fun test. This same setup using the M's pivot plate instead of the flipper produces much more muted results.

Moral... for the F, there is no need for a blanket with our PCB unless someone is experiencing conductivity issues (I have not)
2-piece vs 1-piece keys
Spoiler:
In most of these scenarios, I tried both the 1-piece and 2-piece caps. Again, I type hard, so factor in based on your typing style, but the 1-piece caps provided a SLIGHTLY more solid audible feedback (I cant notice any difference in feel). The 2-piece caps are more hollow (especially on key release). This was most noticeable to me when using in conjunction with the F spring. The differences are so minute though that I would never advocate for one over the other.
Conclusion
The independent barrels, quality of barrel plastic, foam, and metal top plate have the largest impact on eliminating the cheaper feeling resonance of the M. The F's quality of the barrel plastic, higher coil count of the spring, and larger surface area of the flipper make for a much lighter, crisper, and audibly pleasing keystroke.

Each one of these things independently make small impact. When all these improvements are merged... well... that's why we are doing this project :)

User avatar
Techno Trousers
100,000,000 actuations

07 Jun 2016, 17:41

Wow, that is probably the single most interesting keyboard related post I've read all year! The first question that jumps to my mind, now, are the springs. Would it be possible to test springs with even more loops the the stock F springs? Could using those possibly even improve upon the crispness of the feel? To me, the Model F has the best typing feel of anything I've ever tried (and I've been at this since the 80s). I had never even considered until now that it might be possible to improve upon it.

User avatar
lot_lizard

07 Jun 2016, 18:00

Techno Trousers wrote: Would it be possible to test springs with even more loops the the stock F springs? Could using those possibly even improve upon the crispness of the feel?
I agree the F is about as perfect as I could imagine, but the world of springs has tremendous choices. It is a fascinating device. It would take substantial testing, but it is certainly possible it could be improved upon. Even keeping the same actuation force, you could have varying shapes, variable pitches (coil counts) to force actuation at certain points, etc. Not limitless options, but close. This is a tiny subset of helical options (in this... only constant pitch, variable pitch, and hourglass would buckle properly, but there are many more options).
Image

It appears that variable pitch has even been tried in the past with M clones
photos-f62/buckling-springs-can-5001cw-t2043.html

DMA

07 Jun 2016, 19:11

"We need total coverage!" (from Fear and Loathing in Las Vegas, of course)
That's quite a complete comparison.

It would be a sacrilege, but I will still ask.
Is it possible to make silent model F while keeping the tactile feedback?
I understand that complete silence is not possible (though I know a guy experimenting with magnets, teflon and Hall effect sensors in a quest for the silent tactile keyboard).
I also understand that it will probably lose that feeling of the live spring under the keycap when you're releasing the key slowly.

But still. F is quieter than M (I dental-floss-modded mine, so they're about equal methinks), but I recently noticed some pitchforks casually laying around at the office, and there are some unusually large chopsticks near the kitchen. I do not like the direction it's going.

So, how much quieter it can be designed while still offering much the same characteristics? Thin dampers between PCB and the flippers? shock-absorbers at the side of the barrel the spring hits? More coils on the spring? Less coils on the spring? Variable pitch (has to be symmetrical, I think - so that it buckles in the middle instead of god knows where)

PS: sense card on a membrane with a latex underneath? Parasitic capacitance will be crazy, of course, should you do it on two sides of the same mylar sheet. But can be dialed higher or lower by number of mylar layers between sheets with conductors and cutouts in those layers (not sure if capacitance will be more in air or plastic for a given gap - I'm not a capacitor designer - but extra layers increase the gap, and that brings capacitance down.)

User avatar
Muirium
µ

07 Jun 2016, 19:47

You've got me stroking my whiskers here, Lizard. That's some fine experimenting, and does agree with my understanding quite well. F's are better built, using harder materials, in larger quantity where it matters. Mind, I've always assumed the actual membranes are a component of M's… meh. That's my mental image, anyway, when I press the keys.

As for Dima's sacrilicious question: the way I'd damp buckling spring beyond the floss mod is by putting something in the barrel at the point of contact, right where the spring wallops when it buckles. But what?

User avatar
lot_lizard

07 Jun 2016, 21:10

Muirium wrote: As for Dima's sacrilicious question: the way I'd damp buckling spring beyond the floss mod is by putting something in the barrel at the point of contact, right where the spring wallops when it buckles. But what?
That's an intriguing idea...

I have always previously thought about using rubber compound sound deadeners (like Dynamat) as the "foam", and then using o-rings to secure the barrels. If you played with sizes and thicknesses of each, it might hold water. Dynamat is ridiculously expensive though (for the Extreme version at least), plus you have to account for 168+ holes (XT barrel). It could also go on the top of the barrel plate, but I would prefer not to see it.

The other alternative that does work on vehicles is a spray in deadener. Not promoting this product btw... but the concept would be the same. Their online demo is a little ridiculous for the frequency they have the mic tuned in at, but I can say this stuff does work (friend has in his old muscle car). Also, I apologize in advance for their shitty background music :):
http://www.lizardskin.com/sound-control-insulation.html

We could spray either side of the top plate (or both), and use o-rings... Bottom plate too for that matter. We would just need to clean up the areas where it makes contact with the shell (or tape off ahead of time). This stuff can be painted over the top of as well if a color other than black is desirable, and would prevent rust.

Something to consider if we are really trying to quiet it down. But it would probably take several different approaches combined to be REALLY effective

User avatar
Muirium
µ

07 Jun 2016, 21:20

I'm talking specifically that one place: where the spring buckles and hits the inside of the barrel. I'd rather leave the flipper feel intact, see. And I'm lazy, so…

The beauty of barrels is that you could install something like this much more easily than Topre silencing rings, which are accessed from below. Speaking of which, I have thought aloud before about silencing buckling spring on the way up, too. But it's probably a red herring. Unlike Topre, Alps and MX, I don't think there's much of a clack to silence there. With IBM, it's all spring.

User avatar
lot_lizard

07 Jun 2016, 21:39

I just reread my last post... I didn't make it clear AT ALL. I find the barrel sleeve silencer idea very intriguing, but the sound deadener was completely unrelated to the barrels. Just another technique we might try in addition to. Totally follow your point though... I like it.

Myself, I am not turned off by the sound of the originals, but I certainly understand where people would be (I don't share space with others when I am working). Again though, like the advantages of the F over M, I think it takes several small improvements to have real impact (no one silver bullet).

I will check how much space we have inside the barrel sleeve before we impact the performance of the spring. Even an outside wrap would reduce sound (like the Cherry/Topre silencers, but a longer sleeve). Doing both inside (something in the barrel slot where the spring strikes) and out (a full/partial barrel sleeve) is something worth researching. There are sound deadening ABS plastics as well (though I don't know much of their durability as a moving part).

There definitely is a sound as the paddle returns to the roof of the barrel when the key is released. We could always put something thin there on the roof too (like your mentioned "on the way up"). It would leave the key maybe feeling slightly "quarter cocked", but maybe not... Not sure til we try.

I do think if we had the desire though, we could cut the noise down considerably. Let me know if you think that has a larger appeal

EDIT: If we do decide to produce our own barrels (likely) there isn't much preventing us from making at least portions of the sleeve thicker as well. Then the "But what?" from your previous post becomes wrapped up in a single step better. We have mimicked the F up until now, but nothing prevents us from improving it. We are making this from scratch after all. The imagination becomes our playground :)

User avatar
lot_lizard

07 Jun 2016, 22:07

You do realize if we "Type-S" this thing... We will need a new badge yes?!? I leave in your capable hands
Attachments
image.jpeg
image.jpeg (30.08 KiB) Viewed 1317 times

User avatar
Muirium
µ

07 Jun 2016, 22:10

Exactly! IBM didn't have the modern materials, manufacturing techniques, and the same set of priorities that we do. They were mass manufacturing a peripheral for their systems, with the need to keep costs low. Naturally, that was what pulled them ever cheaper, first to the Model M and then down away from buckling spring entirely. We can be pretty wild if we want to be. Ours is a very different position!

Typing on my AT just now, I'm trying to make out the topping-out sounds I'm on about. Actually, here's a video I made of several different boards, precisely on this topic:
Muirium wrote: Here's a five minute video of me nattering about, and typing on, my four Topre switch keyboards:
It's a bit blurry at times — just an iPhone 4S! — and the sound levels won't be a perfect perspective on these boards either, but I managed to keep the story straight at any rate.

The damped NovaTouch is something of a star. I gathered the others together to put it into perspective. What a difference a set of these makes…

Image

Any questions — especially those caused by my quick talking without a script — don't be afraid to ask.
Spoiler: the mystery guest is an SSK, as I remember. You can hear how much louder it is than those Topres. That sheer volume difference is what I'm questioning myself about here. On quieter keyboards, topping out becomes the dominant sound. But is it really a thing with buckling spring's… well… buckling springs!

There's definitely a return sound on my AT. I'd like to see what effect there is on the typing experience by muting it. Even if just to confirm: "nah!"

codemonkeymike

07 Jun 2016, 22:19

What about isolating the PCB/Plate with rubber clips, and lining the case with mass loaded vinyl. Just a thought.

User avatar
lot_lizard

08 Jun 2016, 03:11

codemonkeymike wrote: What about isolating the PCB/Plate with rubber clips, and lining the case with mass loaded vinyl. Just a thought.
The "lining the case" bit would need to wait to phase 2 (new board), since we are wanting to make a true drop-in replacement without M impact (non-evasive), but the clips have merit right away. NOTE: almost all of this sound reduction fluff is phase 2 regardless... we are dreaming at the moment. I like it though. Keep it coming. Any small change to reduce reverberation is worth noting now (even if for later)

User avatar
Techno Trousers
100,000,000 actuations

08 Jun 2016, 05:21

I read through this kind of quickly, but would wrapping something like dynamat around the outside of each barrel help? Or making the barrel thinner inside at the point where the spring impacts (and down to the bottom) then line that inside lower back portion with dynamat? I assume inside-the-barrel dynamat would interfere with the stems if it's up too high...

I am fortunate enough to have my own office and work with folks mostly old enough to have used Model Ms, so the noise has never been a problem for me. But I'm also not opposed to any sound attenuation measures as long as they don't negatively affect the feel.

Interestingly, I used to assume that the audible feedback of key switches was an integral part of the overall enjoyment I got out of typing on a great mechanical switch board, but lately I've noticed that even with my earbuds in so I don't hear the keyboard at all, I'll occasionally just close my eyes and revel in that perfect amount of fingertip up and down feedback I get from my F-122. So I think that even a mostly-silent model F would still be better than anything else. And if it's quieter, that might raise demand for this project, which I think would be a worthwhile goal. I'd love it if everyone could try capacitive bucking spring.

User avatar
lot_lizard

08 Jun 2016, 07:40

Techno Trousers wrote: I read through this kind of quickly, but would wrapping something like dynamat around the outside of each barrel help? Or making the barrel thinner inside at the point where the spring impacts (and down to the bottom) then line that inside lower back portion with dynamat? I assume inside-the-barrel dynamat would interfere with the stems if it''s up too high..
....
Interestingly, I used to assume that the audible feedback of key switches was an integral part of the overall enjoyment I got out of typing on a great mechanical switch board, but lately I've noticed that even with my earbuds in so I don't hear the keyboard at all, I'll occasionally just close my eyes and revel in that perfect amount of fingertip up and down feedback I get from my F-122. So I think that even a mostly-silent model F would still be better than anything else. And if it's quieter, that might raise demand for this project, which I think would be a worthwhile goal. I'd love it if everyone could try capacitive bucking spring.
I should have called out better earlier, and might eventually edit the post, but the sound dampener (Dynamat, spray on, whatever) idea was in reference to only the top plate itself, and not the barrels that pass through it. In combination with a "silent metal" top plate, Muirium was suggesting that we figure out a way to lower the audible dB level of the barrel itself at the point when the spring strikes (the original idea). Two completely separate ideas. If you re-read with that in mind, it will make more sense.

And I am with you on the sound levels. The headphone/earbud idea is a great way to test if the "clack" is actually part of the experience that makes buckling springs magnificent. I'm on the fence, but if it is a hurdle we can overcome and increase adoption of the Cherry/Alps/Topre ("CAT") lovers, it is a worthwhile goal.

Also.. An update. I received the 101-key top plate water-jet cutout from Big Blue Saw. I have already deburred, rolled the radial bend in the four plates, primed (for handling), and dropped off the two top plates with my machine shop friends to break the ~4.1mm offsets with their press. It might take them a couple of weeks to process (requires die changes on the press, and they are busy), but we should have something in the coming weeks. All that would remain is painting and tapping the M2 screws, and I'll handle that myself. We are very close to having a metal prototype

User avatar
infodroid

08 Jun 2016, 17:43

Muirium wrote: I'm talking specifically that one place: where the spring buckles and hits the inside of the barrel. I'd rather leave the flipper feel intact, see. And I'm lazy, so…
As best I can make out, the spring wallops both against the side of the barrel but also against the underside of the keycap. See for example:

Image

User avatar
lot_lizard

08 Jun 2016, 18:43

infodroid wrote: As best I can make out, the spring wallops both against the side of the barrel but also against the underside of the keycap
I do love that little gif. It always makes me wonder how many takes it took to get it to strike square without the left keystem guide. I can totally envision someone pressing it with a winced face and goggles on.

It would be tough to do much about the front keystem guide without adding substantial cost of mixed materials (applying yourself would be a horrible experience). There are three total guides (two sides and the front), but you are right... only the front takes on spring impact during buckling.

For the barrel however (where the highest force of the impact would occur), we do have an empty slot there for basically zero reason (maybe the IBM barrel-master left us an easter egg).
Image

It would be pretty easy to eliminate the slot and test. It would certainly have impact on the spring travel, but there are adjustments to compensate that could also be made (if even necessary). If the test proved valid, we could discuss a mixed material (say silicon rubber) that could be adhered to the slot instead of just eliminating the slot altogether with ABS. Note though, the mixed material price (either via dollars having someone else do it, or your time doing a self application). I think a thicker ABS presence there would have REAL impact on reduction though. Good place to start.

Side note, the machine shop just called, and might have a sample as early as Friday for me to come in and approve (much earlier than I anticipated). This would NOT be the machining of my actual plates, but to show me on some stock sheet. Then maybe even as early as next week sometime we could see life in the metal prototype.

User avatar
Muirium
µ

08 Jun 2016, 18:51

Good stuff.

My simplistic reckoning is that if we bung something into the barrel, a damper target right in the gap you spotted, so that it takes the direct wallop, then most of the energy is exchanged right there. Yes, the inside of the keycap is also slapped by the buckling spring, but I picture a damper taking most of the sting out of it. Guesswork, mind.

The neat thing is that this really shouldn't affect the feel of the switch in use. Because the damping happens after the buckle, when the spring collapses and the cap slides with much less resistance. Competent damping ought to be quite the silver bullet!

User avatar
pyrelink

08 Jun 2016, 19:30

Just crudely tested that Mu, It does not effect key feel all that much, and you right that it silences the initial buckle of the spring, but the flipper hitting the pcb is still audible, and you still get a nice ping when the spring returns to form on the upstroke. I just used a piece of thinly cut foam where the spring should buckle, and I imagine a more elegant solution built into the barrel would work better and more consistently. I will try to take a video of it, but not sure how well the iPhone mic will pick things up.

User avatar
Muirium
µ

08 Jun 2016, 19:41

Colour me intrigued. Remember to shoot a comparison with two different barrels, so we have a control sample. That ought to make it clear what's going on, even with a phone mic.

User avatar
pyrelink

08 Jun 2016, 19:43

I really like the idea of Model F Type S barrels, if we can find a way to silence that upstroke.

User avatar
Muirium
µ

08 Jun 2016, 19:45

Yup. Very noticeable when you hold a key down, but completely overshadowed by the spring unbuckling on the way up. Needs some means of controlling that to make an effective difference.

User avatar
lot_lizard

08 Jun 2016, 20:13

Nice video Pyrelink...

The primary way to get the unbuckle of the actual spring to quite down would be to coat the keystem unfortunately, or produce with a deadening plastic (expensive).

The more sensible alternative would be to break out some contact mechanics theory, and produce a keystem (same over-all shape since this is a moving part... and they are tricky), but either remove material or add material to make the contact surface have less mass. To rephrase, you either add material (like a rib running the length of the keystem on the point where the spring returns... so it contacts only the summit of the rib vs the curved surface), or you remove material (less effective).

For sure something to think about. Fixing the flapper coming back up to the roof of the barrel is a pretty easy fix with similar approaches. I will post something in a little while that explains contact mechanics better

User avatar
tentator

08 Jun 2016, 20:25

Guys now that's interesting discussion here! Model F flippers in silenced barrel.. uhm.. cover the wall of the barrel with a thin layer of gumsheet??

User avatar
lot_lizard

08 Jun 2016, 20:36

tentator wrote: Guys now that's interesting discussion here! Model F flippers in silenced barrel.. uhm.. cover the wall of the barrel with a thin layer of gumsheet??
It's a great idea, but we wouldn't be able to put it all the way around the inside of the barrel (or outside of the keystem) without refactoring sizes to compensate for the added material, and then there is the issue of the barrel interior not actually being circular.. so you would need to "heat expand" (opposite of heat shrink) the tubing. I really would like to avoid altering anything where we have lateral movement of two surfaces. Moving parts are just hard to create something ideal. You wouldn't think so, but getting the right tolerances just proves difficult. I still like the idea of adding a dampening sleeve like you mention, but doing it to the outside of the barrel (similar to the o-rings in Cherrys and Topres)

Actually wikipedia explains contact mechanics pretty well. They get too bogged down in the Hertz and JKR formulas, but they have some nice diagrams. This is a animated one from them that shows the idea. If we used a rib at the back of the keystem, it would be the equivalent of two colliding circles (think two index fingers clapping). We currently have the last frame of the animation (two hands clapping).
Image

User avatar
lot_lizard

08 Jun 2016, 22:25

Here's another idea I might try (keep in mind all this testing will be on the Cherry MX adapter). Never let the spring come back to vertical... At least just slightly offset. So on the return, we slap the key stem with the spring being slightly convexed. It will certainly still make a sound (less), but there would be almost zero reverberation (crisp), and actuation force would be reduced slightly (on the top of the keystroke)

I am going to work up several prints to test the various approaches (including the barrel mass increase at the front "superfluous slot"). Within the week I'll send off for the 4/5 best approaches to test. If something pops in your mind... POST IT!!!

No bad ideas
Last edited by lot_lizard on 09 Jun 2016, 01:12, edited 1 time in total.

User avatar
tentator

09 Jun 2016, 00:36

yes true I did some tests here and indeed half of a millimeter of anything (I tried with some shrink gum for cables) is defeating the whole spring system and feel if at all being tactile at most times.. :(
probably the material of the keycap (or better the under cap) should be something porous and gummy.. if we could have a material like that produced.... uhm..
or you where thinking about those great mx stems indeed.. if you plan to produce them keep me in your list of interested ones! :)
dental floss idea is another one for instance that tries to make the spring reverberate and thus ring much less.. maybe comingin both ideas.. floss + gummy + porous material for the inside of the stem?

DMA

09 Jun 2016, 20:28

..darn it it's all so hard!
I want silent buckling spring.
But I love the feel of the vibrating spring after the key returned home even more! And silencing the upstroke - if even possible - will dissipate all energy so this will have to go.

Arrgh!
Well, clicking it is, then. Need to whine me an office.

Post Reply

Return to “Workshop”