Custom Buckling Spring Membranes

[wcass]

you might look into making custom membranes if folding doesn't work.

you can get the clear sheets in correct thickness from art/drafting supply (bottom is 5 mil, top is 3 mil). plan trace routing to new controller location and create a paper template. put the clear sheet on top of the template and trace the lines and dots with a conductive ink pen.

[mkawa]

samwisekoi and i have been talking about this, but the standard silver-based conductive pens (see: CAIG circuitwriter, mg chem equivalent, etc.) don't work very well on plastic. i don't understand enough materials/chem to know why, but i've found that it's only conductive over very short distances and has very high resistance even then. the bare conductive ink i just received is significantly better, but has very odd resistance characteristics that depend on geometry. in short, long skinny traces have ridiculously high resistance.

[hasu]

I had same experience with  circuit writer and shield paint for guitar cavity .  They are very brittle on curved surface once dried. cupper tape worked very well . Maybe 3m emboossed tape is the best . You can even solder on tape before apply on membrane .

Anyway, great progress!

[mkawa]

I had same experience with  circuit writer and shield paint for guitar cavity .  They are very brittle on curved surface once dried. cupper tape worked very well . Maybe 3m emboossed tape is the best . You can even solder on tape before apply on membrane .

Anyway, great progress!

what do you mean by cupper tape? can you elaborate? is it just that the silver inks need more texture to stick to the surface? thanks!

[regack]

I had same experience with  circuit writer and shield paint for guitar cavity .  They are very brittle on curved surface once dried. cupper tape worked very well . Maybe 3m emboossed tape is the best . You can even solder on tape before apply on membrane .

Anyway, great progress!

what do you mean by cupper tape? can you elaborate? is it just that the silver inks need more texture to stick to the surface? thanks!

Hah, this is funny, I didn't read it as cupper, I read it as copper tape... only after re-reading it did I see my brain automagically interpolated it for me...  Anyway, he probably meant Copper Tape.

[hasu]

I had same experience with  circuit writer and shield paint for guitar cavity .  They are very brittle on curved surface once dried. cupper tape worked very well . Maybe 3m emboossed tape is the best . You can even solder on tape before apply on membrane .

Anyway, great progress!

what do you mean by cupper tape? can you elaborate? is it just that the silver inks need more texture to stick to the surface? thanks!

Hah, this is funny, I didn't read it as cupper, I read it as copper tape... only after re-reading it did I see my brain automagically interpolated it for me...  Anyway, he probably meant Copper Tape.

Haha, Yes I meant copper tape. I wrote that on phone without dictionary
I used copper tape to fix membrane switch trace instead of conductive pen. At first I used normal not embossed copper tape but it was difficult to get conduction with trace. To make contact with trace I needed to needle tape. But 3M embossed copper tape(1245?) worked well without needling.

You can see some photos of my membrane sheet here. It is better than my English, hehe.
http://geekhack.org/index.php?topic=29483.0

[rknize]

Nice work, hasu.  Someone really determined could make reliable traces this way.

[mkawa]

I had same experience with  circuit writer and shield paint for guitar cavity .  They are very brittle on curved surface once dried. cupper tape worked very well . Maybe 3m emboossed tape is the best . You can even solder on tape before apply on membrane .

Anyway, great progress!

what do you mean by cupper tape? can you elaborate? is it just that the silver inks need more texture to stick to the surface? thanks!

Hah, this is funny, I didn't read it as cupper, I read it as copper tape... only after re-reading it did I see my brain automagically interpolated it for me...  Anyway, he probably meant Copper Tape.

Haha, Yes I meant copper tape. I wrote that on phone without dictionary
I used copper tape to fix membrane switch trace instead of conductive pen. At first I used normal not embossed copper tape but it was difficult to get conduction with trace. To make contact with trace I needed to needle tape. But 3M embossed copper tape(1245?) worked well without needling.

You can see some photos of my membrane sheet here. It is better than my English, hehe.
http://geekhack.org/index.php?topic=29483.0

i remember that thread! in fact, i've been looking for it forever! i've just de-archived it. would love more pictures of your membrane.

[The_Beast]

would love more pictures of your membrane.

I thought this was a family site?

\

[wcass]

if we are talking group buy already, how about getting a custom matrix made:
http://www.alibaba.com/product-gs/465289669/Flexible_Membrane_Circuit_Cables_with_Conductivity.html

instead of going from Unicomp controller to Teensy, you might go directly from membrane to a Teensy++. for connecting to the membrane, maybe an FFC like the older M's but with a right angle like this:
http://www.te.com/commerce/DocumentDelivery/DDEController?Action=srchrtrv&DocNm=520314&DocType=Customer+Drawing&DocLang=English
you could mount it behind the back plate, right at the top of the board and not need to bend or extend the membrane - just make a notch.

[mkawa]

wcass, if you can find a supplier that can do M sizes and the kind of complex artwork we need with reasonable MOQ and pricing i'm all ears. my understanding though is that small flex cables are a completely different animal from the large M membranes; the latter requires significantly more expensive tooling

[wcass]

wcass, if you can find a supplier that can do M sizes and the kind of complex artwork we need with reasonable MOQ and pricing i'm all ears. my understanding though is that small flex cables are a completely different animal from the large M membranes; the latter requires significantly more expensive tooling

i'll give it a shot. and you are right, small flex cables are very different than keyboard membranes. membranes are made using standard silkscreen technique only using conductive ink like DuPont 5021. the ink is sold by the KG, so need to find someone that is already doing this type of printing to make it cost effective. the membranes would be about 12"x5" - i can do the design, but will need a sample membrane in order to put the holes in the right place (all of my M's are of the 101 type).

[wcass]

do you think that the rivet positions are the same? either way, it might be good to have the measurements for all known models documented in the Wiki. i'll do the 101 membrane that i have and save as a PDF. if you print it out at 100% you could lay your membrane on top and see if the rivet holes line up. trace patterns will have to be changed of course, so i might leave them off for now.

[rknize]

The rivet positions between SSK and 101/4 are the same.  I used a new Unicomp mat on an SSK without any mods, other than cutting it down.

[mashby]

Thank you for splitting this from the original thread.

I'll track both, but I like that each topic has it's own thread.

[wcass]

would someone with a 104 membrane handy please check the contact points and rivet points on this drawing?
print it at 100% - best if you have legal paper - and lay your membrane on top. let me know what needs to move.

thanks

[mkawa]

The rivet positions between SSK and 101/4 are the same.  I used a new Unicomp mat on an SSK without any mods, other than cutting it down.

there's one rivet that appeared over the years and isn't present in early 90s era mats but is present in modern mats, but afaik this happened in both the SSK and the 101 key at the same time.

also it's not a huge compatibility issue. basically, include all the rivets from a 93ish M and you're good.

[wcass]

i submitted RFQ's to three membrane switch manufacturers requesting ...

I am trying to get replacement parts for a computer keyboard; very low volume. I do not need color, adhesive, embossing, domes, or connectors – I just need membranes with conductive print.

Material = polyester, DuPont 5021 or similar
Thickness = 0.075mm
Size = 300mm x 130mm
Print area = ~10%
Quantity = 5

i'll let you know if i get a bite.

On an unrelated note, i did a drawing of an idea i had for an M60 case. the basic idea was a milled upper and sides (aluminum or hardwood) and a sheet metal bottom. i was not planning on sharing it, but ...

• 5.74 inches is nothing to brag about.
• ribbed for her pleasure.
• as you can see, the testicles have not descended.

[mkawa]

i worry about the feel of the board with the backplate resting on a piece of aluminum, especially given that there will be protruding bolts (~1.5mm)

interesting idea though... hmmm......

[wcass]

 A reminder that this thread was split off from samwisekoi’s idea here. If this stuff looks interesting, please support his efforts to make a folded membrane and case for a 60% buckling spring keyboard.


The smaller the keyboard, the more important it is that it will accept multiple concurrent key-presses. NKRO is usually achieved in conductive keyboards by creating a one-way circuit using one diode per switch. But solderable flexible substrates are exotic and a diode anywhere near the switch might create enough of a bump to prevent the switch from working reliably. Can you tell that I have thought about this a lot?
I explained this idea in a few threads elsewhere; use a very thin PCB for the bottom membrane and route all of the traces to diodes beyond the border of the top membrane. As it so happens, there is an inexpensive surface mount diode available with specs similar to the common diode used with Cherry switches. But because they are 1x5 diode arrays, only 14 are needed for the M-60.
http://www.diodes.com/datasheets/ds31437.pdf

I already have very thin flexible blank PCB – enough to do 6 of these boards. It is 8mil thick – only .003” thicker than the original IBM bottom membrane. I have the skill and tools to DIY etch these boards, but don’t have the skill or tools to mount the diodes. If anyone has reflow experience out there, I would be happy to supply the parts to make several and ask for a subset back. If I don't get any volunteers then i will likely build a prototype without the diodes and check into the cost of having the the boards etched and assembled by Sitopway Technology or other vendor.

http://www.skycraftsurplus.com/coppercladboard23x17x00080.aspx

I have a very simple design for the bottom membrane that is similar to the original IBM design. I figured that “single-sided” would create less stress on the traces when clamped to the curved back plate. The controller would be made from a Teensy++ (no pins) with two FFC connectors mounted on it. The controller would end up top-center under the back plate. I’ve included some drawings below and will work on more for anyone having trouble visualizing it.
 M60_bottom.pdf
As I said in my previous post, I send out 3 RFQ for a custom top membrane to companies that advertise they will do jobs as small as one piece – and today one of the vendors contacted me asking for a drawing in order to make an estimate.
The final piece of the puzzle is programming the Teensy++. I would really like to leave that to someone better qualified. I hope to be able to offer assembled parts as a bribe.

[mkawa]

can you just send them your M60 drawing and see what they say? that's a nice, non-trivial starting point.

[wcass]

membrane vendor #1 update:

Dear William,
Thank you for your fast letter.
We have checked your PDF file , we can produce it for you .
For 5pcs, the mold fee is USD$100,  plus sample fee USD$120, total will need USD$220, and you will need to pay for the shipping.
If interest , pls feel free to contact us ,thank you .
Best regards,
Amy

 
i am sure that it is like PCB orders and cost would scale down significantly as quantity increases, but $44 a membrane is not what i was looking for.
 
i have yet to hear from the other two vendors and i think i will also ask local print shops to see if they have done (or would like to do) screen printing with electricly conductive inks. i am going to get a sheet of 3mil poly from a local art supply and bake it for a couple minutes at 90C to simulate curing process and see how it holds up.
 
mkawa, IIRC you said that you have tried the carbon ink and that resistance increases with the length of the trace. i think that resistance would be too high with that alone - but what do you think about using carbon ink for the contact points only and running magnet wire from dot to dot?

[mkawa]

here's the datasheet for the carbon ink: http://bareconductive.com/file/technicaldatasheet-bareconductivepaint-pdf

the short answer is yes, the resistance goes up with the length of the trace. a 1.5in" trace made by pen was about 1.2ko. the longer answer is that big traces can be low-resistance, but the ratio of length to height must be small. it's weird and counterintuitive.

eta: i have thought about magnet wire. you don't even need the carbon ink to affix the magnet wire at each point: tape works fine. the problem is basically that it's a pita to produce sheets like this in any kind of volume at all.

[hasu]

I came across this interesting stuff today. Snaptron domes. http://www.snaptron.com/
I wonder this dome switches with thin flex PCB works well instead of membrane sheets. Their actuation force is beyond 200g, I'm not sure model M hammer can push down these domes.

[wcass]

i do not think that metal dome switch would work any better. the base PCB has two contact points per switch - rim and center. the metal dome sits on the rim contact and over the center contact. pressing hard enough on the dome will deform it to complete the circuit.
 
what might work better is if we could make the hammers conductive and place one contact point at the hammer "pivot" and another where the hammer "head" strikes the board. there could be no padding so it would likely be noisier than an F.
http://www.thomasnet.com/articles/custom-manufacturing-fabricating/how-to-metalize-plastic
 
i should have a test PCB and membrane done in the next ten days or so. the barrel frame i have is cracked from age. i will try to repair it with epoxy unless someone has a spare barrel frame that they would like to donate to this experiment.
 
hasu, i acknowledge your programming skills. could i ask your help with a test controller? i have a Teensy++ and the matrix will be 15 columns by 5 rows. the rows would be on the cathode side of the diodes. you can pick the pins.

[rknize]

Model cement works better, as described here:

http://wiki.geekhack.org/index.php?title=Modifications:IBM_Model_M:Cracked_Barrel_Frame_Repair

Alternatively, they are available from Unicomp in both full and SSK sizes for $10.

I've thought about the conductive hammer idea before.  There are several problems with it, but they are not insurmountable.  Any conductive coating or paint would be prone to wear where it matters the most: at the hinges and the contact point.  A better solution might be a very light spring made from a "U" of fine piano wire or similar.  This would be a lot of manual labor, though, and does not scale well.  Another idea I had was a piece of copper or aluminum tape on the hammer with a fine wire "brush" to bridge to the moving hammer.  Again, this is a lot of manual labor and there may be key bounce issues with such a setup.

[hasu]

OK. I'll try to code for this project this weekend.

i should have a test PCB and membrane done in the next ten days or so. the barrel frame i have is cracked from age. i will try to repair it with epoxy unless someone has a spare barrel frame that they would like to donate to this experiment.
 
hasu, i acknowledge your programming skills. could i ask your help with a test controller? i have a Teensy++ and the matrix will be 15 columns by 5 rows. the rows would be on the cathode side of the diodes. you can pick the pins.

[Hubbert]

I encourage the conductive hammer approach.

[mkawa]

cracked barrel frames are no fun. unicomp can have a new one for you lickety split

[hasu]

wcass, can you post your matrix schematic or top and bottom sheet pics?
It is needed to define keymap of your firmware.

I found bottom pcb pic above but it has only 14 lines, not 15. I guess it is not final design.

OK. I'll try to code for this project this weekend.

i should have a test PCB and membrane done in the next ten days or so. the barrel frame i have is cracked from age. i will try to repair it with epoxy unless someone has a spare barrel frame that they would like to donate to this experiment.
 
hasu, i acknowledge your programming skills. could i ask your help with a test controller? i have a Teensy++ and the matrix will be 15 columns by 5 rows. the rows would be on the cathode side of the diodes. you can pick the pins.

[wcass]

i will be making the top sheet from 3mil poly and copper tape (thanks for that too!). 5 tape lines strait across the sheet - one for each row.

attached is what i will be etching this weekend. i dropped a block on top of where the 14 diodes will go to complete the circuit. if this works as well as i think it will then i will contact a couple PCB assembly shops to see what it would cost to get them etched and assembled for us with diodes in place. as it is an etch - it is a mirror of what will be on the PCB.  PCB3.pdf (22.71 kB - downloaded 383 times.)


actually, better swap the Baskspace for the Del because the diodes will not be in on this test board.

[hasu]

OK. Try this and let me know your result.
https://github.com/tmk/tmk_keyboard/tree/mm60
https://github.com/tmk/tmk_keyboard/archive/mm60.zip

Pin usage is PD0-PD7 and PC0-PC5 for column PB0-4 for row.
It doesn't support ghost-block feature at this time, so you may experience ghost key with some key combinations. Though I don't think this is a big problem for testing purpose.

I named this project mm60 tentatively, but fell free to rename it.

[wcass]

THANK YOU HASU!

i am no expert at this, but here are my notes and some pictures from etching.

previous experience has taught me that my home printer does not work for this. if you are thinking of doing DIY PCB fabrication and you do not have a high-end laser printer, save your drawing to PDF and take it to your local office supply for printing on "glossy" paper. this paper is available in 11x17 and is almost as good as specialty transfer paper. it is also going to be "in stock" and the cost is next to nothing - just 32 cents with my drawing on it. use the in-store paper cutter to trim to size.

the paper at my local Office Depot tends to shrink about 1% when heated, so i scaled my drawing to 101%. just before ironing, i heat my drawing in the oven at 200F for about 5 minutes to pre-shrink the paper before putting it on the copper. i know that i have ironed enough when the drawing has bled through to the top (about 10 minutes for a board this size) and then drop it in a water bath for about 10 minutes. as you peal the paper off, there should be little ink left on the paper (most should be on the board).



i soaked and etched in a drawer from a cheap storage bin. i gave it a too-quick inspection before i etched. looking at the pictures and knowing where they are, i can now easily spot two breaks. the third break can't be seen without significant magnification.

[mkawa]

these breaks should be easy to fix with a conductive pen (eg, caig circuitwriter). good job man!

[wcass]

another weekend and more progress.

i cut the back plate today. here are a few pictures of the back plate, a test bottom "membrane", and controller.

what's that? you say you can't see the controller? here's a picture of it from the rear edge.

still don't see it? OK, i'll flip it over.


the connector is a right angle 17 pin FFC connector that i got from Digikey. i will only be using 15 pins; PD0-PD7, PE0-PE1<not connected>, PC0-PC6. i might move it over a tenth of an inch to improve access to the mount hole under the USB connector or decide to flip the controller to the top, depending on how it fits in a case.

speaking of keyboard case, has anyone tried anything like this? suggestions/constructive criticism would be appreciated.
http://geekhack.org/index.php?topic=40006.msg816503#msg816503

[mkawa]

have you seen the prototype picture i posted in the ssk thread? we've had some ideas a bit like that, but without the top bit and using 3d printed material as the substrate

[wcass]

have you seen the prototype picture i posted in the ssk thread? we've had some ideas a bit like that, but without the top bit and using 3d printed material as the substrate

i have been following your thread and samwisekoi's as well, but had not see your new picture yet; thanks.

your comment about torque causing a failure - makes me think that my bottom, should have upturned sides and be 1.5mm or more thick.

[mkawa]

how are you planning on securing the pcb to the plate? bolts?

[wcass]

yes, a standard bolt mod. the PCB is only 8mil - the orginal bottom membrane was 5 mil, so i don't forsee any issue.
the top-most and bottom-most rows will secure the keys/frame/membrane/plate assembly to the case bottom (or top - depending on case design).

[mkawa]

nut torque will depend on what frame you use i think more than the plate. post some pictures before you throw the barrels on and i'll eye it

[wcass]

ooooooooohhhhh - you were talking nut torque! i thought you were talking case torque (4 corners of the case not on the same plane).
 
i've done a few bolt mods, so i don't expect that to be a problem.

[wcass]

more pictures.

this is a practice pcb with a sheet of clear 3mil plastic on top and i am cutting the mount holes. keeping the sheet and pcb aligned was a pain in the butt. in the end i used double-sided tape.


i started with 5mm copper tape that comes with paper backing. it is a pain to work with because it has the worst properties of wire and tape; it sticks to itself, bends, and kinks. as you can see i cut several of the traces too short and had a very hard time taping in a strait line.


it will be interesting to see how much of a difference the misaligned traces make to the keys registering. i'm waiting now for a barrel frame and a few key caps from Unicomp.

[wcass]

OK. Try this and let me know your result.
https://github.com/tmk/tmk_keyboard/tree/mm60
https://github.com/tmk/tmk_keyboard/archive/mm60.zip

Pin usage is PD0-PD7 and PC0-PC5 for column PB0-4 for row.
It doesn't support ghost-block feature at this time, so you may experience ghost key with some key combinations. Though I don't think this is a big problem for testing purpose.

I named this project mm60 tentatively, but fell free to rename it.

so i read the readme.md from the first link and downloaded and expanded the second link.
i installed WinAVR, opened a command prompt from the ...\tmk_keyboard-mm60\keyboard\mm60 folder ran make -f makefile.pjrc
it created some files and folders and spat out a bunch of of text including ...

Code: [Select]

Creating load file for Flash: mm60_pjrc.hex
avr-objcopy -O ihex -R .eeprom -R .fuse -R .lock -R .signature mm60_pjrc.elf mm6
0_pjrc.hex

Creating load file for EEPROM: mm60_pjrc.eep
avr-objcopy -j .eeprom --set-section-flags=.eeprom="alloc,load" \
        --change-section-lma .eeprom=0 --no-change-warnings -O ihex mm60_pjrc.el
f mm60_pjrc.eep || exit 0

Creating Extended Listing: mm60_pjrc.lss
avr-objdump -h -S -z mm60_pjrc.elf > mm60_pjrc.lss

Creating Symbol Table: mm60_pjrc.sym
avr-nm -n mm60_pjrc.elf > mm60_pjrc.sym

Size after:
   text    data     bss     dec     hex filename
  20156      24     153   20333    4f6d mm60_pjrc.elf

-------- end --------

i presume that now i need to use the teensy loader to load the created hex file. is that correct? do i need the other files and folders for anything?

[hasu]

You are right. You need only hex file to program your controller.

[wcass]

ok, the hex file is loaded and the OS picks up the device, but is the matrix completely blank? i have shorted various row/columns and no keystrokes are registered? how do i test functionality?

[hasu]

Short B0 and D1 with wire or something and see if you get '1' on computer.

EDIT: Added code to enable debug print. Get new source from 'mm60' branch of github and see debug output with 'hid_listen'.

[wcass]

hid_listen replies for the first short only - shorts after this do not display anything. also second keyboard and mouse clicks do not register properly until the computer is rebooted. i duplicated the keyboard issue with USB and PS2 keyboards.  also, when the teensy is plugged in, the LED on the teensy has a dim glow.
the hid_listen reply is ...

r23 +07 rF0 r23 -07

i do not solder well, but i did test with a multimeter before programming and it checked out with no shorts. here is a picture of my ugly solder results.

[hasu]

Matrix read code didn't work. Fixed now. Try new code. Tested and confirmed that left part of matrix works on ATMega32U4, but I have no Teensy++ in hand now to test full matrix.

hid_listen can only work with first Teensy, second one is ignored. Disconnect another Teensy to debug.

[wcass]

i still can't get it to work. it is not locking up the other keyboard, but it is not being seen by hid_listen. windows device manager adds a human interface device when i plug it in, but does not add a keyboard device. hid_listen says "waiting for device . . . . . " and keeps waiting as i short rows and columns.

[hasu]

Wierd. I think you should get some debug output when you short any column and row pin.
Are you sure about using latest source?