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

User avatar
lot_lizard

09 Mar 2017, 23:07

Wingklip wrote: http://www.rccoilspring.com/rectangular-springs.html
Found this that could work for low profile bucklesprings
Of all of this mechanical stuff, this is the part where I am the weakest. Springs and buckling points in regards to outside lateral forces are just hard.


I have a feeling IBM made MANY attempts/prototypes to be honest to get the feel they really desired. For the quantities they knew they would be producing, it would have made proper business sense to have a spring CNC on hand and run T&E to prototype, and then capitalize the machine somehow when ready for retirement (might have the damn thing today for all we know). Unfortunately, that is almost impossible for us unless someone wants to the be the "buckling spring tycoon". We can control spring length easily by clipping coils, but it takes a pretty healthy dive to change coil spacing and wire ID/OD. Each run of a unique spring is several hundred dollars, even for one, if paying a supplier. We were lucky enough to have the Model F "end game" dimensions as a starting point (60-65g), and still missed the math when trying something "lighter", with help from consultants by around 3.6% on our 50-55g goal (ended up with 40-45g). ALL spring producers will be VERY good at telling when the spring would naturally buckle under load. They struggle when the spring is meant to be fired by outside shear tension at some distance (paddle being diverted breaking the perpendicular plane) along with load since that is an abnormal act in almost every situation of normal spring usage aside from our need.

All that said... I am very confident we got the math right after trying once and being off about getting a 50-50g in a typical compression spring of known length. If we keep the same ID, I think we can nail other attributes like changing length or OD... but if we change designs altogether (like rectangular), it is going to be random luck to get it right. Not trying to be Debbie Downer, but LOTS experimentation/deep pockets/hyper-brighter minds will be needed.

For our MF^2, where we want Cherry MX mounts, I have a shorter spring that should buckle at 50-55g spec'd since we need ~1.2mm shorter key stem length to be desirable. We have ran by 3 others that have better guesses than me as to whether it actually works, and they agree it does. Even with that, it's a leap when it comes time to produce it since machine tolerance is always questionable. I just want to make sure others are aware of the road you are going down in this space if you take it on. I hope you do, but be prepared to have hiccups anytime you change a proven design. Ellipse is finding that even mimicking it, let alone altering, was difficult. It's just is an odd switch, and the dimensions are certainly touchy

User avatar
micrex22

12 Mar 2017, 00:16

lot_lizard wrote: ...unless someone wants to the be the "buckling spring tycoon".
Hahaha, this is what came to mind:
Image

I'm looking forward to the 45g springs actually (if there turn out to be spares and they work well in a Model M, I may be interested in more).

User avatar
Wingklip

12 Mar 2017, 11:37

That drawing is absolute amazing lol. But nah if I were to design buckling springs I would be the buckling spring communist. (Draw me like one of your French girls xdddddd)

User avatar
Wingklip

13 Mar 2017, 13:51

What is Phase plan Beta btw, I'm curious

User avatar
lot_lizard

13 Mar 2017, 23:49

Wingklip wrote: What is Phase plan Beta btw, I'm curious
To get the alpha phase 1 in everyone's hands as quick as possible. I need my storage back...


Honestly though, it is still a flat (numerous) profile F using the MX adapters. While I have been twiddling my thumbs waiting on Phase 1 parts and testing, there was some nice additional progress. Not trying to be elusive, but I want Phase 1 in everyone's hands before we start documenting things that are already in the works for 2. It's important to me that people realize we are focused and very real. That said, the biggest hitters that are guaranteed for 2 are new barrels (backwards compatibility to both profiles to help save harvested legacy Fs), new flippers, and plans for a new spring (though it hasn't been proven yet completely). The new barrels and flippers allow passthrough tension screws in all 4 corners of EVERY barrel base (eliminating flat plate tension concerns). Unfortunately they will cost almost half as much as Phase 1 to produce ;), but will need some snazzy cases that aren't close to being done. This will be referred to as MF♭ (sheet music symbol for flat). MF2 will also be accepted. The fun in it will be deciding on the various layouts we want to offer (finally a F gets to be WHATEVER... Dox, etc)

Phase 3 is a partnership with several others that is very early in the works and a different switch altogether. Moral... There is a reason mfdrop.com was squatted, but we want Phase 1 to be in wild and loved a bit first before we even start showing real details of future designs. If you want to help design on a big scale though, PM. Things are moving a bit!!!

User avatar
seebart
Offtopicthority Instigator

13 Mar 2017, 23:59

lot_lizard wrote: Phase 3 is a partnership with several others that is very early in the works and a different switch altogether.
:shock:
anti-beam-beam_o_749056.jpg
anti-beam-beam_o_749056.jpg (24.66 KiB) Viewed 1876 times

User avatar
lot_lizard

14 Mar 2017, 00:29

seebart wrote:
lot_lizard wrote: Phase 3 is a partnership with several others that is very early in the works and a different switch altogether.
:shock:
anti-beam-beam_o_749056.jpg
hahaha... exactly. The *word "wannabe" should be in there someone as a footnote. Excited about that one for a variety of reasons, mainly could be a strong partnership of our "hall" of fame brethren to pool together and "effect" the "beam"ing down of something new altogether one of these "spring"s #cheesymovietrailer. Don't want to say too much about it yet though. Smoke and mirrors until it's not.

User avatar
Ratfink

14 Mar 2017, 00:32

Hall effect beam spring, eh? I understand how, but not immediately why. Why's that supposed to be better than capacitive beam spring?

User avatar
lot_lizard

14 Mar 2017, 00:38

Ratfink wrote: I understand how, but not immediately why. Why's that supposed to be better than capacitive beam spring?
UNDERWATER BEAMSPRINNNNGGG!!!! (monster truck voice)


Completely playing. I follow your point. Mostly !!science!!

EDIT: in the spirit of...

User avatar
XMIT
[ XMIT ]

14 Mar 2017, 03:13

Damnit, that was supposed to be our secret. Ah well.

User avatar
Techno Trousers
100,000,000 actuations

14 Mar 2017, 04:17

I won't tell anyone! :twisted:

User avatar
just_add_coffee

14 Mar 2017, 06:49

Techno Trousers wrote: I won't tell anyone! :twisted:
Won't tell anyone what?

User avatar
DMA

14 Mar 2017, 18:57

lot_lizard wrote:
Wingklip wrote: The new barrels and flippers allow passthrough tension screws in all 4 corners of EVERY barrel base (eliminating flat plate tension concerns).
Careful with those. You're putting huge grounding rods around unshielded cap you want to measure. Not necessary a bad thing, just to remind that there are electrical aspects to the design.

Though my recent experiment with forgetting to connect PCB grounds show that the system can take a lot of abuse.

PS: there are also PCB routing considerations - one can't put a trace thru that hole (though there are some possibilities for creative via abuse)

User avatar
Wingklip

14 Mar 2017, 21:57

DMA wrote:
lot_lizard wrote:
Wingklip wrote: The new barrels and flippers allow passthrough tension screws in all 4 corners of EVERY barrel base (eliminating flat plate tension concerns).
Careful with those. You're putting huge grounding rods around unshielded cap you want to measure. Not necessary a bad thing, just to remind that there are electrical aspects to the design.

Though my recent experiment with forgetting to connect PCB grounds show that the system can take a lot of abuse.

PS: there are also PCB routing considerations - one can't put a trace thru that hole (though there are some possibilities for creative via abuse)
Wait a second, I didn't write that lol

User avatar
DMA

15 Mar 2017, 00:53

Wingklip wrote: Wait a second, I didn't write that lol
Sorry about that, quoting system is a bit funky.

User avatar
lot_lizard

16 Mar 2017, 10:28

DMA wrote:
lot_lizard wrote: The new barrels and flippers allow passthrough tension screws in all 4 corners of EVERY barrel base (eliminating flat plate tension concerns).
Careful with those. You're putting huge grounding rods around unshielded cap you want to measure. Not necessary a bad thing, just to remind that there are electrical aspects to the design.

Though my recent experiment with forgetting to connect PCB grounds show that the system can take a lot of abuse.

PS: there are also PCB routing considerations - one can't put a trace thru that hole (though there are some possibilities for creative via abuse)
Still on the road. Have some updates to make soon in the other thread though that are goodness.


For the MF2 (MF♭), even though the barrel would have a quarter-circle in each corner, you are very right that we wouldn't need/want to use them all (for the sake of screwing that many in if nothing else)... though it would actually become strong enough to be a legit murder weapon at that point.

We are going to have the opportunity to step up in thickness of the both plates since we aren't trying to squeeze into any existing case. This part excites me probably the most to be honest. I "think" we would always stick with steel for things we offer directly because it is just a better metal for the job. Aluminum would finally become interesting though if you had thicker plates. I am hoping people are more apt to create their own plates from the finished plans on their own too though. Send something to BigBlueSaw and have max thickness aluminum anodized (maybe etched), have a wood/copper/brass (at least inlay) made, etc. The curve of the plates of the legacy Model M and F just really kills the options.

User avatar
micrex22

16 Mar 2017, 14:48

Regarding springs, I have some interesting information.

It appears the M2 is very sensitive to tolerances, in particular the amount of windings on the springs. I took original Model F springs and put Model M flippies on them. After installing all of the springs in an M2, 80% of the springs would not buckle, 10% would buckle only when completely bottoming out, and the remaining 10% would 'sort of' work. When the Model M springs were put back in there were no issues.

What was interesting is that on the ones that would 'sort of' work, the SOUND of the springs sounded very similar to a Model F and not like a Model M at all.

SO... as lot_lizard was saying in regards to the how very sensitive the tolerances are (re: 45g springs), it appears this is an exact science and not something with loose tolerances--and is exaggerated on a more delicate "low profile" buckling spring assembly.

ALSO the Model M2 *is* lighter to press despite using the EXACT same springs as a regular Model M. This means that even more variables are involved with how stiff a spring feels in a buckling spring keyboard and not just the windings. In the case of the M2 my theory is that because the barrels are much shorter and more spring is exposed, this causes a more extreme "buckle". Of course since regular Model Ms and Fs use the same barrels, their 'lightness' is exclusively due to windings and rubber mat vs. no rubber mat.

And yes, an M2 with Model F springs does become even lighter... at least for the very few switches that work.

User avatar
E TwentyNine

16 Mar 2017, 15:49

lot_lizard wrote: MF2 (MF♭)
That's a perfect name. Someone should start working on a logo for that if they haven't already.

User avatar
lot_lizard

17 Mar 2017, 01:38

E TwentyNine wrote: Someone should start working on a logo for that if they haven't already.
sounds like an excellent idea... The SVG for the logo is here: https://github.com/lot-lizard/Model-MF/ ... ter/mf.svg

Anyone is welcome to run with it. Please keep as SVG if you would

User avatar
Wingklip

18 Mar 2017, 05:17

You should try running it with the arcon font and also striping it like IBM's logo xd

User avatar
DMA

18 Mar 2017, 21:37

A package arrived to my door today.
package.jpg
package.jpg (178.84 KiB) Viewed 1616 times
Looks like someone run out of plain packing tape :)

Now I'll let it sit for a couple of hours - it's not like condensate formed on it, but I'm not taking chances.
And then I'll bring it to the scope.

User avatar
Techno Trousers
100,000,000 actuations

18 Mar 2017, 23:50

Ooh, exciting!

User avatar
lot_lizard

19 Mar 2017, 02:04

This is all my fault with traveling so much lately. I rely on my wife quite a bit to facilitate shipments. I gave her DMA's a LONG time ago. She unfortunately got confused and thought it was a "received" parcel. When I finally pressed about it, she admitted it was a mistake. In all fairness, I understand. I rarely SEND large packages, so she unpacked it for me, got rid of the box, and it's been sitting. After pointing out the mistake, she had a box big enough for one beamer plus accessories laying around... So it got sent out. Thanks again DMA for being patient, and I'm jealous because you are the first to see the tape

User avatar
Wingklip

19 Mar 2017, 04:14

Wait you have beam springs? Please send me one I'll pay xd

User avatar
DMA

19 Mar 2017, 09:00

Wingklip wrote: Wait you have beam springs? Please send me one I'll pay xd
Got one on loan from lot_lizard.

Looks this way now:
IMG_20170318_231725.jpg
IMG_20170318_231725.jpg (115.4 KiB) Viewed 1487 times
I'm afraid it's not for sale though.

In other news - beamspring gives ~2 times stronger signal than F122, doesn't mind being driven by rows, scanrate 7.75kHz, and (surprise!) is bouncy on the way up: if you lift your finger too quickly, you have a chance to get this:
DS1Z_QuickPrint32.png
DS1Z_QuickPrint32.png (69.7 KiB) Viewed 1487 times
I.e. hurray, unified capsense hardware is possible and lies at my table.
But that's for another topic.

User avatar
lot_lizard

19 Mar 2017, 13:07

DMA wrote: I.e. hurray, unified capsense hardware is possible and lies at my table
Glad it arrived safely, and glad the experiments are turning out in our favor. It looks like bluetooth in a beamspring is off the table though ;), unless we are going to stick a car battery inside (good thing there is room).

EDIT: This is the first time I have seen Capsense against something I can really register the scale on in my brain. Its VERY small. It should be with such few components, but your pic really helped to see it. So many possibilities...

Vizir

19 Mar 2017, 15:52

I'm sorry DMA, I had an extremely busy work week. I'll take some pics of the space inside the case this week if you still need.

Sent from my Nexus 6P using Tapatalk

User avatar
DMA

19 Mar 2017, 19:34

lot_lizard wrote: It looks like bluetooth in a beamspring is off the table though ;), unless we are going to stick a car battery inside (good thing there is room).
Scanner itself has surprisingly modest power requirements. Probably the fact that it's almost completely in hardware helps. Whole system draws around 25mA - and 18 of those are supposed to be eaten by CPU. Note that there's not a single sleep instruction in the code yet, and USB transceiver itself is pretty hungry - so this can definitely be brought down significantly. I'll try a bit later, when I get to suspend-resume.
Has to be two-chip system though - no psoc5 with BT, and psoc4 BLE only has 8 fixed ADC input pins, which is not enough.

But RGB LEDs on the same PCB are probably out of consideration still. I keep reading those are quite noisy neighbors.
lot_lizard wrote: This is the first time I have seen Capsense against something I can really register the scale on in my brain. Its VERY small. It should be with such few components, but your pic really helped to see it. So many possibilities...
Should I rename it to "capsense"? A bit too close to "capsize" though :)
Can be smaller even. The chip is 8x8mm. There's WLCSP version which is 6x6mm - but I don't feel comfortable with solder balls and being BGA it will probably require four layers anyway. But, as you can see, connector occupies good quarter of the real estate - so if integrated into sense card it will require about 10x18mm or something. Also caps and resistors can easily be 0402 which should get it under 10x15mm comfortably.

Vizir np, I have enough things to play with for now :)

User avatar
fohat
Elder Messenger

19 Mar 2017, 20:55

Great work, but fretting to shave off each millimeter of something already very small seems to have diminishing returns.

My big desire is long-term durability and compatibility - I am typing this on a keyboard over 3 decades old and all the original electronics are still going strong.

User avatar
Wingklip

19 Mar 2017, 23:58

Is this becoming a beam spring conversion project on the side too?

Post Reply

Return to “Workshop”