Mastodawn

A shift register would pair nicely with this VCA matrix. Been a long time since I've designed any circuits, and I've never done anything with eurorack-level voltage, but I *do* happen to have a couple 5V 8-bit shift register ICs gathering dust in a drawer...

#eurorack #synthdiy

Looks like I'm going for it.

This is rough. I've forgotten a lot, and I didn't know much to start with.

After a couple rewires, my perfboard is missing quite a few copper pads. Some of the components are attached only to each other, not the board. I wonder if epoxy would be sufficient to hold them in place.

This project quickly went from being one of my cleanest perfboard prototypes to one of my messiest. All because I decided I wasn't *actually* comfortable with cutting the corner I was cutting.

Less-recent self: "I can skip the resistors for the LEDs, right? They're salvaged, they'll be blinking on and off anyway, and I only have them on here in the first place because these jacks can't fit in 2hp and I didn't like having all that unused space when I went to 4hp. Plus it'll be interesting to see how it works out, right? Yes. I will leave them off For Science! "

More-recent self: "Why did I wire these up like this? This is Wrong and Embarrassing. I must correct this IMMEDIATELY."

There are two wolves inside me. One cares so little about correctness that they must never be let near a soldering iron or production code. The other will bootstrap their own Linux on the only fully open source hardware available at the time just for the assurance that everything is Correct. When they fight, the perfboard loses copper.

I also momentarily got Ohm's Law reversed somehow and spent a minute thinking I'd have to put the LEDs in serial with the shift register pins when I'd already soldered them in parallel, but fortunately I got my head on straight again before any perfboard was harmed.

No shorts. Gonna design and print a plate before applying epoxy because the board is a little curved and it definitely won't be happy about bending at all once it's epoxied up.

No epoxy yet because I don't have proper PPE for that shit at the moment, and won't until December 3rd at the earliest. Still gonna wire it all up and test it out a bit. Here's a progress photo.

Hope I don't fry anything!

Nothing caught on fire or blew up, but the module also isn't working. Debugging will have to wait until tomorrow, I guess.

I've figured out the problem and I'm not sure how this design made it off the breadboard. I thought it was working poorly because of loose connections; it was working *at all* because of loose connections!

There's enough room left on the perfboard to correct my mistakes, I think.

Well. I modified the perfboard prototype, but still no luck.

Problem had been that the serial clock needed to be the inverse of the register clock for my serial-in, parallel-out shift register to behave like a serial-in, serial-out register. Breadboard prototype had been unstable enough that it kinda worked with them just tied together.

Fails differently now, but still fails. Back to the multimeter.

Huh. There is a short between pins 10 and 11, but only when there is a power cable plugged in.

It does not matter if the power cable is plugged into anything on the other end.

This is a mechanical problem.

I've made sure, with soldering iron and hobby knife, that the solder traces between those two pins are well-separated. I am beginning to suspect the chip socket. 😛

Chip socket seems to be fine. No signs of melting, and the power header pushes on other nearby wires and components, but not that. No obvious route from those wires and components to a short, but this perfboard is a mess. Used Dupont headers as an extender for the power cable to power it up without creating the short successfully, but still no dice. Gate inverting transistor isn't damaged, but I'm still not seeing what I should be from it.

Maybe I'll set this project aside for a bit. My workspace is a disaster and the housework's getting neglected again. I should make an actual schematic in the meantime. Most likely I'm going to have to discard the current perfboard mess. I don't have high hopes regarding my ability to salvage something workable from this at the moment.

Probably time to get some sleep, too.

Right. Using PCBs seems to be popular. I'm not an electrical engineer by training. Here's the schematic I've come up with. Anyone feel like taking a peek?

http://lyk.so/eurorack/shift-register/schematic.pdf

Gonna start designing a board with this schematic and see where that lands me.

#synthdiy #eurorack

I've designed a board, sent off for fabrication, and... within less than 12 hours learned enough to realize my circuit has some big flaws that, while they won't keep the module from functioning, aren't the sort of thing I'd personally like to have in my eurorack. So I've canceled the fabrication order and am back to the drawing board, this time with a slightly better understanding of what I'm doing!

In the meantime I've set up a website where I might, in the future, sell a couple copies of my modules to defray my parts and fabrication costs, provided I can come up with something that seems sound enough.

http://eggmodular.com

Egg Modular

Continuing to work on the circuit design. I think, once again, I've arrived at something that works which also addresses the previous design's deficiencies. I am using an op amp for the first time, because buffered outputs are important!

Still unsure of whether to use 8V or 5V gates, but leaning toward 5V just to keep things simple. If it's good enough for ALM, it's good enough for me, right?

Considering learning to hand-solder surface mount parts in order to make designing the new PCB easier.

This browser-based circuit simulator has been very helpful for testing out basic concepts. Transistors and op amps are still shaky territory for me, it seems.

https://www.falstad.com/circuit/

Circuit Simulator Applet

New schematic dropped. Old schematic had really obvious, embarrassing errors. New schematic (probably) has less obvious, less embarrassing errors!

http://lyk.so/eurorack/shift-register/schematic.pdf

I would be shocked if it has *no* errors, but I'd be okay with being shocked in this particular case. 😸

I now have 9 bypass capacitors on this board. I think this may be overkill, but I'm erring on the side of overbuilding because I really kinda don't know what I'm doing yet.

3 for +12V, 3 for -12V, and 3 for 5V.

I think the design is as good as I can manage at my current level of knowledge. Just updated the schematic at http://lyk.so/eurorack/shift-register/schematic.pdf and sent off for PCB fabrication.

PCB fabrication paused again as I've had a change of heart regarding using the 5V rail.

Further meditation on the schematic has revealed more problems. At least I'm learning a lot.

I think my problems have gone from "this mistake could destroy something" to "I'm not sure if this design element is actually doing exactly the right thing," which is nice.

I went and asked people on the Internet for a review of the schematic I could not find anything wrong with and now I have a completely different schematic which I am still working on. I'm glad I asked. Even though my original schematic would have worked, what I've got now is more elegant and has a wider output voltage range.

Turns out that my design cannot work unless I also design and manufacture my own audio jacks. I'm a few days into that now. The only tools and materials on hand which seem applicable to the problem are my FDM 3D printer, wire cutters, needle nose pliers, and paperclips. I've gotten close to something working, but no dice yet. Ruled out two general approaches and am currently printing the first version of a third design.

I do have some significantly finer solid core wire that *might* work, but it's insulated so using it as bare wire seems like a waste, it's most likely *too* fine, and I don't have much left right now. So I'm sticking with paperclips until I have something that might be served better by much finer wire.

Success! The third approach is significantly better. Going to proceed with iterating on it. The biggest issue with the previous two had been the strength of the tip connection. Now the biggest issue seems to be how much the wire tends to shift in the channels. This doesn't seem to be an issue for *basic* function, but I decided to also add jack sensing to the third design; that now seems to be the weak point.

Also, allowing the mechanical stress to transfer to the solder joints seems bad.

Looks like y'all are somewhat interested in what I'm doing here, so here's a photo of some of my attempts.

The two big rows in front were my first, as I plan on making this an LED and jack assembly and foolishly though this would be simple. The one on the perfboard had severe pin mobility problems and was not very forgiving of imprecise wire work. It did at least allow me to confirm my ability to solder a pin connected to PLA without melting it, though.

More details in the alt text.

Pondered a little, considered twisting the wire on top and/or bottom and leaving a cutout to tuck the twists away, buuut... that seems fiddly and bulky.

I constructed the pin channels in the second row assembly from two cylindrical cutouts bent about 10 degrees away from each other. My intent had been to make a somewhat springy arch inside the jack sleeve, but it seemed to keep the wire in place pretty well too... 🤔

That helped. Not enough, but it did help. If it were soldered in place, it wouldn't move, but it'd be better if it didn't move without the solder as well.

Might try a more extreme angle next. 20 degrees was *too* extreme the last time I tried this approach. Either the correct angle is between 10 degrees and 20 degrees, or I need a fundamentally different approach.

Heading in two different directions now. One involves printing an insert for my 22 AWG solid core wire to wrap around. The other involves buying a bit of tinned copper ribbon, like the kind used to make battery packs, and using that as the leads on one side instead of wire. The latter seems more robust but less accessible.

I'm going to continue developing the former approach until the tinned copper ribbon arrives. I suspect the latter will be more robust.

Not off to a good start with the "plastic insert" approach. Already considering abandoning it.

This design will require bigger through-holes than I'd wanted. Little room for bosses here. But the tip insertion sensing is working reliably so far, at least with this one prototype. The sleeve insertion sensing pin seems to be getting pushed too far up to return to the right position, so I'll have to try adjusting that.

Uses two paperclips per jack.

Wish I could figure out a more elegant wire retention technique, but the only thing I can think of is a channel barb/"one way valve." I don't know the proper name. Anyway, the dimensions I'm working with are too small for the FDM printer to do anything like that for me, AFAICT.

Cutting out "wheel wells" helps alleviate the bulkiness by moving the twists up a bit. I've also switched to 22 AWG wire. Takes less force to insert and remove the plugs, feels a lot better, probably better for the plugs as well. Less reliable plug insertion detection, though. Going to have to adjust the design some more.

I feel like tinned copper strips would be easier to design for. Have a couple small spools ordered. I do like that this design works with paperclips, though. More accessible.

Too easy to overdo it on the twists and mess everything up, and still pretty bulky. Shorter channels also mean more pin mobility unless everything is just so. Reverting the "wheel wells" and switching to superglue. I'd wanted to avoid using adhesives because adhesives fail, but strictly speaking this stuff only needs to hold the pins in place until they're soldered down.

Also going to see if it's possible to bend the wire into a torsion spring, held in place with a top cover.

I've discovered a commercially available jack, in the meantime, that fits my dimensional requirements. Had passed over it initially because I was still focused on transparent jacks at the time. But now I'm locked in to the challenge, and there's a non-zero value, in terms of design flexibility, to being able to create my own weird jack constructs anyway.

I think I cracked it, but I'm going to need some more testing and tweaking. No luck with jack sensing using wire or paper clips, but the flat tinned copper I got is working. Will post more details after sleep.

Latest print has a reliable sleeve switch but the tip switch doesn't work. Also very fiddly and annoying to get the copper routed. Printing a version which will hopefully alleviate these issues a bit.

Unsure whether to continue this development or just buy my jacks like a normal person. I've dropped my need for transparent plastic and have found an alternative that meets my size requirements, so what am I even doing this for now?

Never fails. As soon as I put my design in the panelization queue, I have another idea about how to tweak things. Just realized I could make the shift register and its expansion module one and the same if I just rearrange the expansion header a bit and "normal" the jacks.

In this case it was largely due to my putting off thinking concretely about the expansion module until after the "primary" module was "done," so I suppose I should have guessed this would happen.