Custom Buckling Spring Membranes

[wcass]

date and time are march 17 4:44 pm

[hasu]

can you use git command? if so try 'git log'. if not get zip archive again.

EDIT: BTW, you should get debug output like this.

Code: [Select]

Device disconnected.
Waiting for new device:.........
Listening:

r/c 0123456789ABCDEF
00: 0001000000000000
01: 0000000000000000
02: 0000000000000000
03: 0000000000000000
04: 0000000000000000

---- action_exec: start -----
EVENT: 0003d(22163)
ACTION: ACT_LMODS[0:20]
keys: 20 00 00 00 00 00  mods: 00
processed: 0003d(22163):0

33333333333333333333333333333333333333333333333333
r/c 0123456789ABCDEF
00: 0000000000000000
01: 0000000000000000
02: 0000000000000000
03: 0000000000000000
04: 0000000000000000

---- action_exec: start -----
3EVENT: 0003u(24787)
ACTION: ACT_LMODS[0:20]
keys: 00 00 00 00 00 00  mods: 00
processed: 0003u(24787):0


[wcass]

i am not familiar with git tools. i re-downloaded and re-compiled. fc /b oldfile newfile found no difference. so i tried it on a second teensy and hid_listen still does not pick up the device.

it probably would help if you had a teensy, so please PM me with your address and i will send you one of mine.

[mkawa]

i'm happy to send a full suite of teensies to hasu. let me know before you ship out to him wcass

[hasu]

Thank you for offering. If you can donate a spare Teensy it is very helpful to me.
Actually I have one Teensy++ 2.0 as controller of my HHKB now, which is one of my daily driver boards. I must unscrew to open enclosure and get access to Teensy++, this is exactly what I'm doing now and it is PITA a bit.

OK. I forgot Teensy has LED on port PD6, we can't use this port as column line to sense matrix.
I updated repository to disable PD6, try it again. I believe you can see some on hid_listen now,
though you can't get 7,y,h,... column keys atm.

[mkawa]

wcass, can you pm me? i can forward your extra teensy on to hasu with the rest of the varieties and whatever else he wants from the states

[wcass]

success! it registered keystrokes!

the barrel frame is in the mail, so i am just tapping my fingers lightly on the top of the membrane. i am now very confidant that this will work. it does not seem to matter that the top traces are a bit lot crooked.

hasu, i am planning to etch a new PCB. will we be able to use PD6? the 17 pin FFC connects to all pins from PD0 to PC6 - i need 15 of these for columns. could we use PE0 or PE1? also, i flipped the teensy over so it is now on top of the matrix. it receives the row wires better that way and should be easier to mount the switch assembly to a case bottom. i know that changes the matrix row and column order - left-most column is now PC6 and right-most column is at PD0; top row is PB4 and bottom row is PB0.

[hasu]

PD6 might be able to used with some extra trick but I'm not sure. You'd better not to use PD6. Yes, you can use PE0 and  PE1, instead.
How do you want to use pins for column lines actually? PC6 for left most and PD1 for right most lines and PD0 is not used?

[wcass]

PD6 might be able to used with some extra trick but I'm not sure. You'd better not to use PD6. Yes, you can use PE0 and  PE1, instead.
How do you want to use pins for column lines actually? PC6 for left most and PD1 for right most lines and PD0 is not used?

you can pick the pins - any 15 of that 17 will do.

[hasu]

OK. I changed matrix code. Pin usage is indicated below. Column 15 doesn't exist, so D0 is not used.

Code: [Select]

* col: 0   1   2   3   4   5   6   7   8   9   10  11  12  13  14  15
* pin: C6  C5  C4  C3  C2  C1  C0  E1  E0  D7  D5  D4  D3  D2  D1  D0

* row: 0   1   2   3   4
* pin: B4  B3  B2  B1  B0

EDIT: Update code again. Try latest code if you have a problem.

[wcass]

so i am getting to the point where i need to hold the key assembly to actually test typing. what i have come up with seems to be very stable even without sides or the back edge attached.

to attach the back i will need a 2mm tap to thread the 22 gauge sheet metal. for a production run, the metal work would be done by an expert; the angles would be less curvy and screw heads countersunk at bottom front.

[sleepy916]

Looks like things are moving along nicely, great work so far!

[mkawa]

nice! some neoprene might compensate for bolt thickness on the bottom of the plate when you get there

note: i don't think you're going to be able to countersink screws in there. it's really thin mild steel

[AKIMbO]

Sweet......looking good wcass!

[bazemk1979]

wcass your wizard men!!!!!!!!!!!!!!!! Making case for this would be very hard. Maeby its cheaper to make plastic case instead of aluminum?

[wcass]

wcass your wizard men!!!!!!!!!!!!!!!! Making case for this would be very hard. Maeby its cheaper to make plastic case instead of aluminum?

i'm hoping that i can design a case that is attractive, solid, and affordable; worthy of this keyboard's heritage. the cost for the raw material for my case is less than $10, but to make it pretty will likely cost 4 or 10 times that for labor depending on quantity. i think that cost per keyboard would be under $100 if quantity is > 500, but i don't predict that kind of interest. buckling spring is a niche market; 60% boards are a niche; people willing to spend $100 on a keyboard is a niche.

the best case scenario that i can think of is this:
i get it working and it is beautiful. i send it over to Unicomp for them to take a look at it. i talk them into agreeing to assemble and sell the keyboard IF i can guarantee some quantity will sell. Kickstart.

[bazemk1979]

that can be a start too, but don't  forget my brother, you might not find 500 that will go for the BS mini, you might be surprised if turns out they are around 300 interested in it..... Everybody here loves the small factor.

[wcass]

another weekend and more progress.

i re-did the top membrane trying for straighter lines. i also cut some grooves in the barrel frame for the wires to run up to the top.

much better, but won't work - look at the left side with the separator sheet in.


let's ignore that for a while and go to the other side where the wires come into the teensy. it is inconvenient having the wires soldered directly to the membrane and teensy so i thought i would use standard pins and sockets. i have strait pins already, so i soldered on 7; five for the rows and two for caps lock and num lock LEDs. but it looks like there is not enough clearance for the strait pins and socket. i'm accepting suggestions for another connection method. i also mounted a ground wire and USB socket on the rear of the case.


Going back to the OP, i have been researching other methods for creating custom membranes and i think these two look promising. please comment back which one you think i should try.

inkjetFlex.com is a contract manufacturing company out of the UK that will do prototypes pretty inexpensively. the minimum order is 300mm x 300mm (enough for two membranes) for £25. but the prototype service allows no choice for substrate - 4mil PET (Unicomp uses PE - 3mil for top and 5mil for bottom) and added price of shipping would put the cost per sheet at about $30.
http://inkjetflex.com/site/rapid-prototyping/

the other promising method is true DIY for now, but a commercial version might be right around the corner. early last year, researchers at the University of Illinois developed an ionic silver solution "ink" that evaporates leaving behind pure solid silver bonded to the substrate. the solution contains just 3 easily sourced chemicals - none of which react with plastics. i found an odd DIY write-up here: http://jordanbunker.com/archives/41  i would probably do the same thing but would also get some empty refillable marker pens for easy dispensing. that and some quality mask like FrogTape.

[mkawa]

the problem with the diy ink is two-fold i think. first of all, it doesn't come with a method of automated printing, and i imagine that shoving a syringe full into a 30$ inkjet printer will not end up happyfunland. second, i am not to be trusted with dangerous chemicals, period.

inkjetflex.com looks the most promising, but the pricing is not attractive at all. that said, the prototyping pricing isn't insane; it's actually pretty reasonable. the real problem is that there's no way of upping the volume once we feel good about a prototype... :/

basically, we can get like 4-10 copies printed for high but not impossible prototyping pricing with short lead time OR we can get 1000 made for true volume pricing. there's just nothing inbetween.

oh, AND to ***** even more, we still can't get the reall cool stuff we want with either of these options: capacitive sensors or on-fcb diodes. BLARGH

[wcass]

As I said before, the best case scenario is that Unicomp likes the idea and agrees to assemble and sell the keyboard. They would probably want a guarantee that some quantity will be purchased (this is where a Kickstart campaign would help). And this keyboard will be NKRO when in production. The diodes will go on the bottom PCB.

I think I have created enough prototypes to have worked out most of the early design flaws. I need to get quotes for factory produced PCBs and sheet metal work. If there is interest I will organize a small group buy.


My DIY membrane is only semi-functional. There is still interference where the copper tape is deformed or rippled. I did create a CAD file for a professionally made membrane that will curve around and slides into another FFC connector on the top of the Teensy. One of the pins that I have been using for rows (PB4) is blocked by the Teensy's USB connector, but it looks like PF0-PF4 might work.

[wcass]

i just bought a bunch of Pyralux, so i will be making my top sheet the same way as i make the bottom sheet; DIY etching.
http://www.ebay.com/itm/190815165931

[wcass]

I have asked Hasu to make a few changes to the code - hopefully the last.

• enable Caps Lock on PB6 and Num Lock on PB5
• move rows to PF0 - PF4
• other small changes to column pin and key assignments (detailed below)

I had to add a dedicated Delete key because this prototype does not have diodes (the next prototype will!), so Ctrl+Alt+(Fn+Backspace) would generate a "ghost S". Adding the delete meant that some key had to go and I chose the tilde/grave. I would like to still be able to produce these characters and the obvious choice is Fn+Esc - but how would i do a tilde without getting a caps lock? So maybe we need a macro (Shift+grave) and assign that to Fn+backslash.

We also need a Num Lock key (Fn+N) and a few changes to the Num Lock layer to match the key caps. The num lock layer is primarily for typing escape characters which requires using the keys on the num pad. 

[hasu]

OK. I need to know some.

1. Which pin of LED do you connect with those ports? Anode or Cathode?
2. No problem.
3. You flip column pins again?

Note that
A. 'NumLock' LED is controlled by host, it doesn't indicate state of Num Lock layer.
B. 'NumLock' key(NLCK) just register numlock scancode to host, not activate 'Num Lock Layer'. You need Fn key on 'Function Layer' to switch that layer.
You probably need two separate keys; normal 'NumLock' key to control numpad state of host and Fn key to activate 'Num Lock Layer' respectively.

I think this is close to what you want. You can toggle Numpad layer with Fn+N(FN2+FN1) and 'N' key works as Numlock key on Numpad layer(Layer1).

Code: [Select]

static const uint8_t PROGMEM keymaps[][MATRIX_ROWS][MATRIX_COLS] = {
    /* Layer 0: Default Layer
     * ,-----------------------------------------------------------.
     * |Esc|  1|  2|  3|  4|  5|  6|  7|  8|  9|  0|  -|  =|  \|Del|
     * |-----------------------------------------------------------|
     * |Tab  |  Q|  W|  E|  R|  T|  Y|  U|  I|  O|  P|  [|  ]|Backs|
     * |-----------------------------------------------------------|
     * |Fn2   |  A|  S|  D|  F|  G|  H|  J|  K|  L|  ;|  '|Return  |
     * |-----------------------------------------------------------|
     * |Shift   |  Z|  X|  C|  V|  B|  N|  M|  ,|  .|  /|Up |Shift |
     * |-----------------------------------------------------------|
     * |Ctrl|Gui |Alt |      Space             |Alt |Left|Down|Righ|
     * `-----------------------------------------------------------'
     */
    KEYMAP_ANSI(
        ESC, 1,   2,   3,   4,   5,   6,   7,   8,   9,   0,   MINS,EQL, BSLS,DEL, \
        TAB, Q,   W,   E,   R,   T,   Y,   U,   I,   O,   P,   LBRC,RBRC,BSPC, \
        FN2, A,   S,   D,   F,   G,   H,   J,   K,   L,   SCLN,QUOT,     ENT,  \
        LSFT,Z,   X,   C,   V,   B,   N,   M,   COMM,DOT, SLSH,     UP,  RSFT, \
        LCTL,LGUI,LALT,          SPC,                     MENU,LEFT,DOWN,RGHT),

    /* Layer 1: Numpad
     * ,-----------------------------------------------------------.
     * |   |   |   |   |   |   |   |  7|  8|  9|   |   |   |   |   |
     * |-----------------------------------------------------------|
     * |     |   |   |   |   |   |   |  4|  5|  6|   |   |   |     |
     * |-----------------------------------------------------------|
     * |      |   |   |   |   |   |   |  1|  2|  3|  *|   |        |
     * |-----------------------------------------------------------|
     * |        |   |   |   |   |   |   |Nlk|  0|  .|  /|   |      |
     * |-----------------------------------------------------------|
     * |    |    |    |                        |    |    |    |    |
     * `-----------------------------------------------------------'
     */
    KEYMAP_ANSI(
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,P7,  P8,  P9,  TRNS,PMNS,PPLS,TRNS,TRNS, \
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,P4,  P5,  P6,  TRNS,TRNS,TRNS,TRNS, \
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,P1,  P2,  P3,  PAST,TRNS,     PENT, \
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,NLCK,P0,  PDOT,PSLS,     TRNS,TRNS, \
        TRNS,TRNS,TRNS,          TRNS,                    TRNS,TRNS,TRNS,TRNS),

    /* Layer 2: Function
     * ,-----------------------------------------------------------.
     * |  `| F1| F2| F3| F4| F5| F6| F7| F8| F9|F10|F11|F12|  ~|Ins|
     * |-----------------------------------------------------------|
     * |     |   |   |   |   |   |   |   |   |   |Psc|Pus|   |Delet|
     * |-----------------------------------------------------------|
     * |      |   |   |   |   |   |   |   |   |   |   |   |        |
     * |-----------------------------------------------------------|
     * |Caps Loc|   |   |   |   |   |   |Fn1|   |   |   |PgU|Caps L|
     * |-----------------------------------------------------------|
     * |    |    |    |                        |    |Home|PgDn|End |
     * `-----------------------------------------------------------'
     */
    KEYMAP_ANSI(
        GRV, F1,  F2,  F3,  F4,  F5,  F6,  F7,  F8,  F9,  F10, F11, F12, FN0, INS, \
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,PSCR,PAUS,TRNS,DEL,  \
        TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,     TRNS, \
        CAPS,TRNS,TRNS,TRNS,TRNS,TRNS,TRNS,FN1, TRNS,TRNS,TRNS,     PGUP,CAPS, \
        TRNS,TRNS,TRNS,          TRNS,                    TRNS,HOME,PGDN,END),
};

static const uint16_t PROGMEM fn_actions[] = {
    [0] = ACTION_LMOD_KEY(KC_LSFT, KC_GRV),
    [1] = ACTION_LAYER_BIT_TOGGLE(1),
    [2] = ACTION_LAYER_BIT_MOMENTARY(2),
};


[wcass]

1. Which pin of LED do you connect with those ports? Anode or Cathode?
2. No problem.
3. You flip column pins again?

1) the backplate is ground and the pin next to these two is +5, so eather will be easy for me, just let me know. pulling the pins low might be the most elegant wiring.
3) yep, re-design of top membrane.

Thanks

[hasu]

Updated the code. Try it on your new keyboard.
LEDs should be connected to Teensy with cathode and to 5V with anode. Also try the keymap and let me know if it differ from what you want.
You can see the change here.
https://github.com/tmk/tmk_keyboard/commit/ec1c321fb144f7991ab0e0add48b841cc3cccf85

BTW, I didn't know Paralux at all and it looks like an interesting material. It is nice to know.

[wcass]

copper clad polyimide (Paralux is a brand name) should work better than any of the other DIY solutions. i have considered it in the past, but it always seemed too expensive for prototypes. but after getting quotes for conductive silkscreening and printed copper i decided to check out eBay again and found this vendor. $45 (with shipping) but enough material for 12 boards - $3.75 a board is very nice! You do have to etch it yourself, but that only adds about $2 material and an hour of time.
 
Mouser is delivering the FFC connectors on monday and the Paralux should arrive about the same time.

[mkawa]

i just bought a bunch of Pyralux, so i will be making my top sheet the same way as i make the bottom sheet; DIY etching.
http://www.ebay.com/itm/190815165931

not bad actually. cheaper than inkjetflex.com.. does the seller have a steady supply of sheets?

[wcass]

it looks like the seller is a liquidator - but he has a lot of this on hand. he is selling 6 more batches of (4)12"x18" and 8 lots of (50)12"x18"

[mkawa]

is it 30 per 4 sheets or 30 per sheet?

pyralux would allow both capacitive sensing and potentially even diodes.. it's also technically printable (but i'm not sure how scalable the printing process is..)

quite interested to see how this goes for you wcass. it may not be a magic bullet but it's definitely an option worth exploring

[wcass]

is it 30 per 4 sheets or 30 per sheet?

pyralux would allow both capacitive sensing and potentially even diodes.. it's also technically printable (but i'm not sure how scalable the printing process is..)

quite interested to see how this goes for you wcass. it may not be a magic bullet but it's definitely an option worth exploring

"$30 for 4 sheets" is the way i read the ad. i'll let you know when i get it. these sheets are single-sided, double sided or thin PCB (like the 8mil stuff i got for my bottom membrane) would probably be better for capacitive; the problem there is finding controller.

diodes is possible, but you probably want to put those on the bottom membrane (because it moves less) and keep them away from the actual switch area. this is exactly what i have planned.

[mkawa]

no, there's space on top of the pcb under the frame sills. there actually isn't space under the pcb, because that's flush with the backplate. also you'd need vias to route traces on the bottom of the pcb. yuck

[wcass]

no, you misunderstand.

the "membrane set" is made up of 3 sheets - the sheet closest to the backplate is the "bottom sheet". this sheet is least effected by vibrations from hammer strikes. i have my "bottom sheet" made from 8mil PCB. diodes will go on the top side of this sheet near the back of the keyboard at the end of each key column. that way the slight bump from the diode does not create a ripple for the "middle sheet" (just holes) or the top sheet.

[mkawa]

oh, i was thinking that if we have pyralux, ie, a material strong enough to actually survive repeated light contact with the hammer, we might as well do a capacitive switch, in which case we only need one fcb layer..

[wcass]

oh, i tried that first. back during the "great GH outage of 2012" i posted here:
http://deskthority.net/workshop-f7/bringing-the-ibm-pc-xt-into-the-21st-century-t3047.html
 
i managed to completely duplicate the capacitive pads of a model F and cut a new barrel frame that put the keys into a modern tenkeyless configuration - all this fitting back inside an XT case, but was not able to get a controller to work. the last post of that thread i discuss the idea that i am working on here.

[mkawa]

ah. either one's a possibility, and you are right in that the membrane style will be the easiest to try first (although i worry about the pyralux's resilience..)

in the meantime, we're still waiting to see how many sheets you get for your 30$

[wcass]

 I’m not sure I mentioned this before, but I suck at soldering. I was using a “helping hands” and did this to my Teensy. It no longer works, so I have sent for a new one.

But I am not about to stop just because I suck at something. I ordered 30 diode arrays – enough to do two boards with a couple left over. I figure that since the PCB is so thin and the diodes are so far apart that it might not be too difficult to mount these. I will try tinning the board contacts, placing the part, and then heating the footprint with an iron from the bottom of the board.
 I am happy to confirm that you get 4 sheets of Pyralux for $30. It is thicker and stiffer than I expected – perhaps because it has 2 oz/ft² copper. I will be etching away most of the copper, so expect that flexibility will improve.
It should be a busy weekend.

 

[mkawa]

not a bad deal at all. as soon as i'm a bit less poor money and time-wise i'll follow your lead and grab some

[wcass]

not a bad deal at all. as soon as i'm a bit less poor money and time-wise i'll follow your lead and grab some

perhaps i'll just send you a sheet if i have any left.

back to the bottom PCB. i just etched a new board and will attempt to mount diodes. i used a new (to me) masking technique where i laser print on a vinyl sheet instead of paper. i also purchased a too-cheap laminator. some advice for anyone reading this, a $20 iron works better than a $20 laminator.

here are some pictures including some extreme close-ups of the best and worst etched spots. the diodes i will be attempting to mount are 1.5mm square.

[mkawa]

PM boost for printing and lamination.

this looks like it will be really high quality with right equipment. how good are you at negotation? i think this is sufficient as a proof of concept, and i don't see any other reasonably priced options. we're going to need a lot of pyralux

[wcass]

the Pyrolux was a little hard to work with, but i think it will work well. the stuff i have has extra thick copper (2oz) so it took forever to etch it off. i started with my usual HCl + H²O² and that went from clear to green to almost black. i had to re-oxygenate it and start again twice (re-applying toner on top of the old toner).

[mkawa]

looks like we're going to need a minimum width per trace rule of thumb.

hmm.. i wonder if BOST's printers can even handle 2oz..

wait, what am i thinking, they're just printing on the screening material. durrrr

[wcass]

the controller is complete. i need to etch a new bottom PCB and it should be fully functional; i just need to make it pretty. i got some aluminum samples in the mail from Brunner Enterprises - very high sheen and the profile is perfect for the front edge. i will be using standard 1" angle for the rear edge.

[wcass]

almost done. 3 new pictures.
controller top

controller back edge

caps lock and num lock LEDs

[dorkvader]

This is excellent! It's also using my favourite case! This is pretty much (almost  my ideal "small" keyboard.

I'm really excited about this project, now. I'll finally have something I can use my F XT for!

[pasph]

Wow!

[rknize]

Looking really good, wcass.

[mkawa]

holy crap this is effing fantastic. YOU HAVE DONE IT WCASS

crap, i may have to start a new account on ebay (deleted mine because of that horrible spam email they sent out recently -- no joke) to pick up a pack of this stuff and have boos- sorry BOST print and laminate/etch out a couple circuits..

[The_Beast]

Nice, I'll be doing the same to one of my fullsized M's when I get some time this summer.


Wood cased 60% BS Model M...



hold on, I need to change my pants....

[mkawa]

wcass, could you start a negotiation with these pyralux guys and see if we can get good bulk discounts? something on the order of 20-40 pieces of material

[wcass]

i'm not sure this type is the best fit. the substrate is 1 mil instead of 3 and the copper is very heavy (2 oz). Dupont makes a different type with 3 mil substrate and 1 oz copper that should be a better fit.

with this keyboard i am getting phantom key presses that i believe are caused by the substrate bending too easily. doubling up the separator sheet might resolve the issue, but i don't know.

i bought some Circuit Writer - i think you said you tried this with poor results, but i will also try sintering (which is part of the silkscreen process).

and if that doesn't work then i will try the "rapid prototype" service from InkJetFlex.com

let me know if you would like to try the Pyralux and i will send you a sheet.