Does anyone know how expensive or difficult it is to create membranes?

courtesi · 20 Dec 2015, 02:34

As a potential project for 2016 I thought about doing something cool and unique like a 60% or 80% scissor/chiclet switch keyboard.

The case, controller, key caps, and even the scissor mechanism should be easily obtainable.

However the big one is going to be the custom membrane underneath. I've noticed that there are quite a few companies that provide services to make them but how do you know which to go with - and how much could it cost?

It completely baffles me that not one OEM - not one has chosen something as simple as a tenkeyless layout.

ohaimark · 20 Dec 2015, 06:03

Flexible PCB exists already... Etching that and putting a thin layer of plastic in between seems very doable.

http://www.instructables.com/id/DIY-Fle ... -Circuits/

Not sure what the durability would be like, but worth a shot.

tigpha · 20 Dec 2015, 19:21

Hi Courtesi,

Back in the 90's I made a membrane for a salvaged keyboard. The tools and ingredients are:

Polyester drafting film
Conductive silver paint
Cartographer's ruling pen
Swann Morton handle No 3L
Swann Morton blades, types No 10 and No 10a

The ruling pen is the best instrument to draw well defined, constant width lines, with a heavy deposit of silver paint. The drafting film is almost exactly the same as the mylar used in membrane keyboard. With a pad of 15 A4 size sheets, there is enough to make ten 60% membrane sets, i.e. two halves of A4, three sheets per keyboard -- the two matrix layers and the perforated separator layer. A pad of A3 size can accommodate an IBM Model M Battleship membrane set. I recommend the Swann Morton blades and handles because you will need plenty of easy grip, and very sharp blades to do a good, safe job. The straight blades cut neat holes, and the curved blades serve to scrape off mistakes. Take care with the blades, though, they are unforgiving to human flesh!

wcass · 23 Dec 2015, 03:14

I've tried my hand at DIY membranes and i had some "professionally" made. Tigpha's solution sounds like a better option than what i did.

I tried using CircuitWriter pen. That ink will work on FR-4 or glass, but it dispenses poorly and dries non-conductive when used on polyester. If the silver ink tigpha linked to is more like the stuff that silkscreeners use, it should work much better.

The membranes that i had "professionally" made, i paid $120 for 10 copies. I requested 3 mil polyester but they gave me 10 mil and the laser cut was very rough (one person said it looked like they cut it with a blowtorch). The print accuracy and conductivity was very good, though.

tigpha · 23 Dec 2015, 15:00

I must add a word of caution: ventilate very well while drawing.

Your nose will be right up against silver conductive paint which contains strong organic solvents, xylene and toluene etc., and these will gently intoxicate you. You will only notice the effect when you try to take a bathroom break, and not be able to walk a straight line, or you might even collapse!

courtesi · 23 Dec 2015, 15:48

Instead of doing this all by hand would it be possible to use a printer? I know I'm dreaming at this point but if I'm successful and others are interested it would be much easer and quicker to reproduce membranes on a printer, no? Not just that, but if frequent small changes are needed turn around would be quicker - not to mention giving the opportunity to collaborate.

Would this even possible at the consumer level with professional results?

Also, isn't it possible to get larger membrane sheets than A4?

I'm very interested to hear your answer regarding these questions. I think if I can do this on a printer what I may end up doing is buying something like a Dell Quietkey and experiment with trying to replicate the membrane. If that is successful then smaller version would be trivial.

I guess that would be my first real success ... a 60% or 80% Dell Quietkey using all the original keycaps/sliders/domes - just add a slimmer modern case and USB controller.

tigpha · 23 Dec 2015, 16:50

Hi Courtesi,

Using a laser printer directly is not an option, the toner is made of plastic. You might be able to heat the transferred toner to a temperature which enables silver dust to adhere, but it quickly gets very messy.

Using an inkjet with special ink made of silver salts might be more successful. Once the print is complete, the salt is converted to metallic silver in a chemical bath, making it conduct. I found a company a while ago (it has since disappeared, acquired and silenced) that offered to print circuits on mylar using this technique. It was elaborate, because the metallised silver was barely enough and needed plating with copper in a second chemical bath to ensure that the result was practical.

The chemistry is explained "Sketchable, Stretchable Circuits"

As for "professional result", with a steady hand I was able to produce satisfactory results. It takes time and patience, but the result is as good as machine printed membranes.

And yes, polyester drafting film sheets of all sizes up to A0 are available, if you have a generous budget. I suggested size A4 based on cost and practicality, and mentioned size A3 too.

This conversation might inspire me to try making a 60% Model M. I was hoping to make a 60% Model F, but a membrane keyboard might be quicker and cheaper for me to make, and produce a working result sooner than the ultimate Model F.

tigpha · 23 Dec 2015, 17:51

I found the link to CIT Tech. The company website redirects to Carclo, an umbrella corporation which seem unable to give CIT the visibility it deserves on its website.

I found an ATMEL blog article describing the process.

I also found an article on EDN Network about CIT Tech.

If CIT is responsive to queries, and can offer a service at unbeatable value, I recommend you try them.

CIT Tech have a LinkedIn page which might enable fruitful communication, rather than dealing with the numpties at Carclo.

wcass · 23 Dec 2015, 20:19

I used Inkjetflex a couple of years ago. The substrate thickness was right and print quality were great, but I did have to cut the membranes by hand. The price was OK.

The next time I tried ordering them, they had been acquired by CIT Tech who was not taking orders because they were being acquired by Carclo. I don't think Carclo is offering this service now.

FYI, you can get NKRO with membrane.
http://deskthority.net/workshop-f7/new- ... ml#p227212

tigpha · 23 Dec 2015, 21:25

Of course! Hi Wcass, sorry I didn't make the connection. Thanks for reminding me of your work.

As for NKRO, I am beginning to wonder if there might be value to be gained by a fully demultiplexed set of e.g. 64 strobe lines that fire in sequence, and a single sensing line to detect the key patterns? Even 128 strobe lines seems practical, still. It could even be implemented in basic logic components, with a binary counter, a set of demux and 128 bits of state flags, each set low or high according to button presses.

The tradition of a matrix designed as square as possible is due to, I think, lack of i/o ports on controller chips. I found some I2C I/O Expander chips from various foundries selling for roughly a couple quid each. Maybe the original cause, lack of I/O pins, no longer applies? Even demultiplexing 128 lines still enables a scan rate in excess of 3KHz, it seems.

If the entire array is demultiplexed, then the opposite membrane can be a simple, single conductive sheet, without need for any printed circuit design at all.

I do like the membrane-over-PCB design of yours, but I regret the loss of the classic curve. Perhaps using 0.8mm thick FR4 board could help restore the curve? It's the type of PCB used in the IBM Model F, and it seems to take on the curved form without problems.

wcass · 23 Dec 2015, 21:39

LOL - I don't think anyone liked the lack of curve, but it did make the case design much easier. Case and firmware are my weaknesses. I'm always happy to help others as a PCB draftsman. Teams are good!

tigpha · 23 Dec 2015, 22:30

Hi Wcass,

I took another look at the membranes you ordered from the "professionals". Very disappointing quality. I am convinced I can do much better by hand.

Chyros · 24 Dec 2015, 01:12

tigpha wrote: ↑I must add a word of caution: ventilate very well while drawing.

Your nose will be right up against silver conductive paint which contains strong organic solvents, xylene and toluene etc., and these will gently intoxicate you. You will only notice the effect when you try to take a bathroom break, and not be able to walk a straight line, or you might even collapse!

I wouldn't say gentle intoxication is the biggest danger with toluene. I've had a toluene poisoning myself during my master's, so I'm very acquainted with it xD . It attacks your eyes. One day I breathed in too much after I spilled some in my fumehood (it came in very awkward metal jerrycans) and the next day I couldn't see properly anymore. Turns out I gained an extra 1.25 dioptre on my left eye (from +2.25 to 3.50) as well as a 0.25 cylinder, pretty much overnight. And that eye damage is permanent.

Xylene isn't as bad, that's why they replaced toluene in markers (remember marker sniffers?) with xylene.

tigpha · 24 Dec 2015, 11:42

Hi Chryos, my deepest sympathy to you, that really sucks.

Organic solvents need great care when handling. I learnt my lesson with much, much less lasting damage, but it surprised and shocked me enough that I felt compelled to add a warning to my instructions.

I never understood glue sniffers.

Chyros · 24 Dec 2015, 12:27

Well it's my own goddamn fault for not handling it properly. After spilling it I should've just left it alone until it was gone, go on an extended tea break or something xD . Oh well, could've been worse. In all fairness, most organic solvents aren't anywhere near as bad for you, and people used to do way crazier things (it used to be customary to wash your hands with benzene after leaving the lab - now a well-known carcinogen).