|
|
Post by young Protoboard on Aug 29, 2020 22:40:35 GMT
Hi, all, I found myself wanting an expressive controller fairly quickly into my AEM career. Not one to purchase something well-known and vetted (usually due to not having enough cash saved up at any given time), I elected to DIY. And as a nearly blind person, I opted for mechanical control rather than touch. I enjoyed the look of custom industrial control panels and took to mechanical keyboard switches. Alas, they were not compatible with 0.1" Protoboard pitch, and while Sparkfun sold a breakout board for Cherry MX switches, they considerably increased the per-key cost of all of my designs. After running the cost/breakeven analysis, I took the plunge into PCB design and construction. Rather than add complexity, height, board space, and cost to my Protoboard builds by making breakout boards, I simply switched all module production to PCB. I'm working on single-sided 100x75mm 1.5mm thick PCB laminate, a dual-heating-element pouch laminator, ferric chloride, and a specialized teeny tiny drill press that has swiftly become my favorite tool of all time. Currently I can manage a .4mm trace width and 1mm diameter holes for component leads; any smaller and I'd need more magnification to work comfortably. I haven't yet attempted a double-sided board; toner transfer is the least accurate step in the whole process and I don't possess the confidence or precision to get the two sides lined up properly. Taking a cue from the DIY module, I worked around this limitation using a daughter board. My process has gotten cleaner and more precise over time, so some of the boards below aren't as pretty as I could remake them today. The keyboard proof of concept: Here's an imgur link; I don't have a YouTube account for uploading.This module is very simply a set of switches that output 5V when pressed. In this demonstration patch, an LFO is clocking the MM-DIV, which triggers some line of the DRUMKIT and KICK modules. The three buttons control three other DRUMKIT instruments. (My DRUMKIT shipped broken and attempts to rectify it with Robert petered out many months ago, so this was the best demonstration I could rig up with these modules that didn't split your ears with high frequency noise.) Those switches are Cherry MX Browns; I bought a Cherry switch tester and enjoyed the feel of the Brown switches the best, but I think I'll move to Greens or Blues with my next order; the Browns occasionally double strike if not pressed with a certain delicate typist's fingering. You may catch a few missed beats in the above video when this occurs. For a proof of concept, though, I'm pleased. Here's one of my first PCB builds-- my SC/OFF CV manipulator module (identical in function to one of the new CV modules):  This particular module didn't function due to an error in board layout, but it's a good example of my first attempts. Clearly, these traces aren't .4mm; I started at 1.27mm and narrowed as I gained confidence in production. IIRC, it's sitting atop a paper on Graph Theory discussing the tradeoffs of adding handles to a surface and reducing the crossing number of an embedded graph, but I've fortunately finished that math project and moved on to greener pastures. (Too much topology for me, personally.) Oh, and it's mounted in a little circuit test bench that I made and didn't share because it's kind of ugly. Here's my latest creation, which I just finished last night. It's a dead simple 8-step gate sequencer based on the CD4017 IC; nothing special and absolutely barebones in features for ease of manufacture. Jumper wires were necessary in a few places due to some spillover of toner mashing signal paths together, and the only solution was to remove some of the offending trace and jump:  An imgur link to its trial run.
Here, the gate sequencer controls an ENV; I demonstrate its continuous gate nature by toggling the gate/trig switch on the ENV. I much prefer the process of PCB layout, etching, and drilling a board, when the alternative is winging it with Protoboard and creating a spaghetti monster. PCB also allows me to attach those headers and bus connectors in the exact same place on every module, whereas Protoboard placement varies based on how the slice is cut, and is difficult to reliably replicate. I've designed a small handful of other modules but have yet to build them. Other projects always seem to jump in the way of my yPb AEM format creations, but that's okay. I've got lots of PCB laminate and plenty of time. Cheers! yPb |
|
Jihel
Full Member
knobs, knobs, and knobs !
Posts: 241
|
Post by Jihel on Aug 30, 2020 8:00:50 GMT
Fine projects !
As a noob in electronic design, I'm impressed  |
|
|
|
Post by slowscape on Sept 1, 2020 21:17:06 GMT
I agree, these look really cool! Are you just cutting the circuits into the boards with a desktop mill?
|
|
|
|
Post by rodney on Sept 4, 2020 13:41:30 GMT
It looks so great, and I can imagine how they feel with all the components exposed.
It's not AE modular compatible yet (unless via the laptop via MIDI), here's a thing I've been messing with using a cherry switch keyboard. I just printed some hexagonal keys to replace the normal keyboard keys.
It's 19-tone equal temperament, which I like because you can still notate it in western notation, just the sharps and flats are different notes, and you get B# and E#. Also, 19 is a prime number so you can run cycles of any interval or diatonic sequence and it will not repeat until al 19 permutations have been used up.
White keys are a sort-of natural Major scale, Black keys are sharps, white keys are flats and red keys are B# and E#.
I have some ideas for how to get MIDI velocity out of a more drastically modified PC keyboard.
This actually came from an idea kicking around in my head 20 years ago to make an expressive word-processor and a velocity-sensitive keyboard so fonts, sizes, colours of text would change, the harder you hit a key. It feels like its time has come. 
|
|
|
|
Post by young Protoboard on Sept 4, 2020 16:11:28 GMT
I agree, these look really cool! Are you just cutting the circuits into the boards with a desktop mill? Thanks! I wish a mini mill were in my budget, and maybe in a few years it will be. Currently I'm acid etching with ferric chloride. Not the funnest stuff to work with; it stains pretty much everything and is a pain to dispose of (neutralizing with excess baking soda until 7 pH or higher, then throwing away everything in sealed containers, paper towels and gloves included). Also, turns out ferric chloride will wreck any metal plumbing in your home, and acetone will melt PVC tubing, which was a double whammy to my bathroom sink during my very first etching. Wear gloves or your hands turn a very pretty shade of yellow. |
|
|
|
Post by young Protoboard on Sept 4, 2020 16:19:34 GMT
It looks so great, and I can imagine how they feel with all the components exposed.
It's not AE modular compatible yet (unless via the laptop via MIDI), here's a thing I've been messing with using a cherry switch keyboard. I just printed some hexagonal keys to replace the normal keyboard keys.
It's 19-tone equal temperament, which I like because you can still notate it in western notation, just the sharps and flats are different notes, and you get B# and E#. Also, 19 is a prime number so you can run cycles of any interval or diatonic sequence and it will not repeat until al 19 permutations have been used up.
White keys are a sort-of natural Major scale, Black keys are sharps, white keys are flats and red keys are B# and E#.
I have some ideas for how to get MIDI velocity out of a more drastically modified PC keyboard.
This actually came from an idea kicking around in my head 20 years ago to make an expressive word-processor and a velocity-sensitive keyboard so fonts, sizes, colours of text would change, the harder you hit a key. It feels like its time has come.
Man, this is just about the coolest controller I've seen! Great keycaps, too. I was personally inspired by my Eurorack friend's 4-step, 4-channel CV sequencer and thought "Hmm, if that were a 3x3 grid of buttons and six pot knobs, that'd be the most expressive control possible in a small package." I very nearly started converting mechanical computer number pads circa. 90's or so, then realized that making my own arrangements would be more suitable and involve less firmware juggling. |
|
|
|
Post by young Protoboard on Sept 4, 2020 16:28:09 GMT
Fine projects !
As a noob in electronic design, I'm impressed It's only scary until you try it and it works. I believe in you! Seriously, though, that initial test run of any circuit or module is the most terrifying and difficult part of the process; you're about to see if the past 8 hours of etching, drilling, and soldering was done either 1. completely in vain, because the module 2. doesn't work, which requires 3. potentially several hours of error-hunting, and that's assuming the circuit didn't 4. blow your system up and leave you in tears. Or, perhaps all that work was 5. worth it in the end because somehow the thing works the very first time?  (Note: #5 never happens.) Start small and simple, gain confidence and sharpen those construction and design skills, then design and build until you hit a limiting factor, then break that barrier down or work around it and keep dreaming and growing bigger. |
|
|
|
Post by young Protoboard on Sept 4, 2020 16:36:21 GMT
ETCHING PRO TIP: Get some embossing foil, the kind that transfers with heat and pressure and is meant for making shiny, pretty, embossed lettering for invitations and stationary and such. After transferring toner to the board, secure down smoothly a layer of this stuff and give the board a few more passes through the laminator.
Why? Because embossing foil is nearly the same stuff as the much more expensive, hard-to-find, purpose-built, "toner transfer foil." This foil will provide a nonporous liquid barrier over the toner. It makes the difference between pockmarked and porous-looking etches and traces (some of which will be broken if you don't remove the board from the etchant at just the right time between "all of my traces are electrically distinct where they should be" and "too much soak time = the toner is soaked through and now I have to start over") and big, shiny, wonderful boards. The very first board is done without this foil, and the 8 step sequencer is etched with it. Highly recommend!
|
|
|
|
Post by rodney on Sept 4, 2020 23:30:07 GMT
I temporarily sacrificed my lovely Filco mechanical keyboard in order to show people the basic idea and garner interest among microtonality freaks out there.
I have a notion that, with two identical cheap membrane keyboards, I can transplant the clear plastic switch matrix sheets from one into the other, along with its scanner and USB board. The outboard processor would see this as two separate USB keyboards. Then, hopefully, I can compare the time between, say, letter 'Z' happening on one USB input and on the other because the two plastic layers of switches are stacked one above the other. This might give me info to use as MIDI velocity.
I can see several potential problems with this, particularly the physics determining which switch will activate first for a given keystroke. However, it's too complex to predict in advance, so the best thing is to knock it together fast and get a better picture of what is needed.
It's my first impulse to try to anticipate everything and make it perfect first try but it's a moving target and I never get things finished. Perfectionism can be really crippling.
I know I say this a lot but it's good to make a crappy prototype as soon as you can then just try it out and ask it what it wants to become. |
|
|
|
Post by young Protoboard on Sept 6, 2020 0:38:53 GMT
I have a notion that, with two identical cheap membrane keyboards, I can transplant the clear plastic switch matrix sheets from one into the other, along with its scanner and USB board. The outboard processor would see this as two separate USB keyboards. Then, hopefully, I can compare the time between, say, letter 'Z' happening on one USB input and on the other because the two plastic layers of switches are stacked one above the other. This might give me info to use as MIDI velocity.
I can see several potential problems with this, particularly the physics determining which switch will activate first for a given keystroke. However, it's too complex to predict in advance, so the best thing is to knock it together fast and get a better picture of what is needed. These mechanical implementation issues make sense to me-- as another example, the wear and tear on both 'Z' keys might be different than the wear between any two other pairs, which would create inconsistencies across the keyboard if they perform differently. If only it were as simple as throwing a pulse generator in between the keyboard and MIDI recipient. Or, 26 pulse generators, one for each key... Hmm. If we could determine how the USB keyboard talks to a computer, perhaps we could code up an inline microcontroller to convert to MIDI. As I type that, I wonder if such a product already exists. Regardless, it would boil down to a matter of programming; the physical implementation of such a circuit would be easy. I'm going to see if I can brainstorm forth a mechanical solution to your problem, regardless; physical builds are so much more fun to work on than software for me. (I was so sad when I removed the front panel of my 2ENV only to discover a big fat microcontroller doing the heavy lifting.  ) |
|
|
|
Post by young Protoboard on Sept 6, 2020 0:40:14 GMT
I know I say this a lot but it's good to make a crappy prototype as soon as you can then just try it out and ask it what it wants to become. Say it more! Preach the truth, brother. |
|
|
|
Post by rodney on Sept 6, 2020 23:12:24 GMT
Re: keyboard
after dismantling two cheap keyboards from China the other evening, I am leaning towards making a custom MCU-based keyboard scanner to poll both keyboards and do the MIDI mojo all in one place.
Then, it would not be such a big step to move to a from-scratch keyboard with plated circuitboard contact pads and a moulded silicon rubber set of keys to contact the pads.
For each key, commercial MIDI keyboards have two such switches with a buckling silicon rubber upper part with a conductive pad to make contact when it hits the circuitboard.
The key has a slightly diagonal piece that ensures that one switch closes before the other one. That's what gives the MIDI velocity data by timing how long it takes between the two switching events.
Below is my rough prototype that involved just rearranging the keys for a 19-tone-equal-temperament scale. I'm talking to a guy in China who can sell me a bag of white D keys and Black keys with which to do a whole keyboard. Frankly though, I think a full-size keyboard done this ways is not so satisfying because you can't reach anywhere near an octave with one hand and your fingers can fall into the gaps between white keys. So, I'll probably 3D print some custom wide white keys to at least reduce the latter.
young Protoboard sorry to hijack your thread! I'm a bit overexcited and obsessed with keyboard ideas at the moment.
|
|
|
|
Post by rodney on Sept 6, 2020 23:12:50 GMT
I'm also keen to try some capacitive touch-sense ideas as discussed elsewhere in the DIY channel. I LOVE the look of your circuit board. Too pretty to cover up and looks so steam-punk. I also applaud your use of a very generous ground-plane.. It saves wastage of chemicals and will probably help keep the hum from creeping in. I also really like the idea of using multiple circuit boards to stack and better manage space. And I agree that lining up double-sided ones would be a pain and best avoided until you figure out a way to register them perfectly, maybe with some kind of special rig? Of course, a run of 20+ units would make it worth getting it all fabbed. |
|
|
|
Post by young Protoboard on Sept 9, 2020 18:44:35 GMT
Re: keyboard
after dismantling two cheap keyboards from China the other evening, I am leaning towards making a custom MCU-based keyboard scanner to poll both keyboards and do the MIDI mojo all in one place.
Then, it would not be such a big step to move to a from-scratch keyboard with plated circuitboard contact pads and a moulded silicon rubber set of keys to contact the pads.
For each key, commercial MIDI keyboards have two such switches with a buckling silicon rubber upper part with a conductive pad to make contact when it hits the circuitboard.
The key has a slightly diagonal piece that ensures that one switch closes before the other one. That's what gives the MIDI velocity data by timing how long it takes between the two switching events.
Below is my rough prototype that involved just rearranging the keys for a 19-tone-equal-temperament scale. I'm talking to a guy in China who can sell me a bag of white D keys and Black keys with which to do a whole keyboard. Frankly though, I think a full-size keyboard done this ways is not so satisfying because you can't reach anywhere near an octave with one hand and your fingers can fall into the gaps between white keys. So, I'll probably 3D print some custom wide white keys to at least reduce the latter.
young Protoboard sorry to hijack your thread! I'm a bit overexcited and obsessed with keyboard ideas at the moment.
Hijack my threads ALL DAY with cool stuff, I'll never mind. Expressive mechanical controllers? Expressive mechanical controllers!!! Any chance on 3d printing a mini-keyboard a-la "faux black-and-white"? A one-octave footprint and a few control switches handling the lowest note's pitch (eg flip two switches and drop two octaves, hold down another and go up five half-steps) might wizard out a really cool controller. And mad props on uncovering the mechanical solutions already present. |
|
|
|
Post by young Protoboard on Sept 9, 2020 18:51:01 GMT
I LOVE the look of your circuit board. Too pretty to cover up and looks so steam-punk. I also applaud your use of a very generous ground-plane.. It saves wastage of chemicals and will probably help keep the hum from creeping in. I also really like the idea of using multiple circuit boards to stack and better manage space. And I agree that lining up double-sided ones would be a pain and best avoided until you figure out a way to register them perfectly, maybe with some kind of special rig? Of course, a run of 20+ units would make it worth getting it all fabbed. EVERYBODY says "dang that looks heckin' cool, throw a transparent front panel on that bad boy" and I just sit here, patch wire in hand, desperately trying to insert at the correct (and illegibly-labeled) socket. Lol. Well, I've not yet tried to make one. However, I'm going to experiment with a clear coat epoxy spray to give the board some protection against scratches, shmoo, and atmospheric moisture, as I also cannot yet line up a solder mask. Such a transparent coating would preserve the look of those sweet traces, which, I'll agree, I love the look. Perhaps someday I'll give a Ciat-Lombarde artistic flair to an exposed PCB. Etchant is my highest cost currently; without leaving all that ground plane on I couldn't DIY a 1U module for under $5, which is my current working budget goal. I have ideas for an aluminum plate press that gets chucked in a toaster oven for 20 minutes or however long experimentation yields best results. With the heat and pressure application relatively static to the board and subsequent fancy, tricky line-up layers, I just might be able to manage double-sided boards and solder masks. |
|
) 
