The way I’m looking at it, jumperless could function as a 128x128 audio switch. Is this likely to work? I don’t know if there might be any bleed or noise issues with this kind of use.
My idea is to use headers to connect out to some kind of audio termination for a music studio. This would alleviate all kinds of routing issues between synthesizers and effects and alleviate the need for so many audio inputs for mixing or recording. It’s also about 1/10th to 1/50th the cost of a flock audio equivalent solution depending on balanced vs unblanced connections. Technically this could carry midi, too, but there are other solutions for routing that from place to place.
So, what kind of audio quality do you think is possible. Also, am I getting the numbers right with 128x128? It seems too good to be true!
Yeah that’s kind of a perfect use case for one of these, I’ve been waiting for someone to do this (and tell me about it, maybe someone has.)
The switches are made for video so audio frequencies are practically DC as far as they’re concerned. I suspect a lot of audio equipment already uses these. The physical breadboard might contribute something, which would be the same for any breadboard so you could test that right now if you want, but I doubt it. If you want, I do have a few where I’ve asked them not to solder on the clips/breadboard if you want to put it in some permanent thing.
Here’s the datasheet for the CH446Qs, on page 7-8 is the chart with things that might be relevant here, so you can see if any of those things are deal breakers. Any signal will pass through a crosspoint switch twice (and sometimes more if you have a lot of connections).
Also note I’m driving these ~4V higher than the max rating, which due to the way CMOS works (and a bunch of testing), makes all those specs a bit better than shown there.
As far as being 128x128, it’s not quite that big, it’s closer to 81x81, but with the added advantage of those 81 nodes being both the inputs and outputs, so it can route things back to itself. So there’s not separated X and Y port like the individual crosspoints have. (lmk if that explanation is terrible and I should try again, ha.)
The Jumperless V5 also has daisy chain headers so you can stick these together and route 8 analog signals between them (at the cost of the routable GPIO), and 4 digital signals to tell the other Jumperless what’s happening.
Or you can connect the Arduino Nano headers together for more, so each one will add 60 nodes and 31 connections can be made between them.
For control, it shouldn’t be hard to make a simple program that basically makes all your equipment look like a node graph and then just spit out the string with the connections to make that happen.
And then that would just send a string over serial like:
f 2-35, 4-55, 6-24, 8-15,
and that will make all the connections for you.
I’d be happy to help make something like that if people wanted it. Maybe there’s a way to just make this a plugin for plugdata or whatever. But I’m kinda out of my element with both music and software for “real” computers, so I’d need some help.
Let me know if you have more questions/ideas. Clearly I enjoy talking about this stuff.
As far as specs I’m not an ee, the thing to do is just some audio quality tests, including just listening. The only thing I could think of is impedance, maybe?
I get the explanation about nodes. With the underlying matrix, an x can only be routed to a y, not another x. Nodes can be freely routed to each other.
A typical studio has far more outputs than inputs so this is a bonus unless you are trying to do pedalboard routing, which is symmetric. Even then you could connect dozens of pedals to something like this.
The daisy-chain is interesting. In the case that you use the nano headers, do both boards contribute 60 nodes or does one have 81 and the other 60, connected by 31 lines? Do the 31 connections just look like additional nodes for each board? Also, could you just use the nano header on a single board to get more nodes?
I could see having all of my output only devices on one side and putting I/O devices on the other side. It’s certainly possible to accomplish a ton with just a single board.
As far as software, I’d want to make an intermediate logical configuration that could be used through a GUI or some standalone hardware editor.
Also, I just noticed your appeal to synth and music people. :-) One thing to keep in mind about modular is that signals can be 10v, which seems like it would be a problem.
So, each board has 60 nodes on the breadboard part, 23 routable nodes on the Nano header, and 8 on the daisy chain headers. All of those just act like any other node, I could have not put the Nano header on there and just made a slightly bigger breadboard area.
You could split that up any way you want by using some as connections to another Jumperless, if you have 2, you could connect all 31 of those extra nodes to the other, so you’d have 60x2=120 accessible nodes and 31 paths between them.
If you had 3, you could connect 15 extra 31 nodes to each of the others, giving you 60 x 3 = 180 “accessible” nodes and 15 paths between any 2. (I would draw a picture but I doubt anyone will actually do this)
Yeah it might clip the signal at around ~±9.3V, but at least on the V5, it wouldn’t cause any damage.
I’ve never gotten a straight answer (I doubt there even is one) about whether anyone actually uses the full 20V swing. But it seems to be somewhat uncommon.
Just an interesting tidbit, there’s also another variant of the crossbar switch, the CH446X, which is 5x24 instead of 8x16 like the CH446Q I use. So if you wanted to make purpose-built hardware for something like this, these would probably be much more effective for acting like a mux (but more way flexible).
I’m so excited to find this thread! I’ve been thinking about the potential of programmable audio signal routing for years - both for modular synth and for effect pedal chains. Any time I’ve searched for it, I’ve just found relatively small patch bays or the ridiculously expensive Flock Audio stuff.
Yesterday, I was chatting with a friend about breadboard prototyping and I thought… Wouldn’t it be cool if there was a way you could just pop your components into a breadboard and then drive the analog traces through code instead of having to fiddle with jumpers endlessly? Imagine if the breadboard even had some display elements to show you the current state of the connections. I did some searching, which brought me to BreadWare and then to Jumperless!
I wasn’t even thinking about how similar this concept is to programmable audio signal patching. I mean - one has little cables and the other has big ones!
So, suffice it to say, it is really inspirational to see how far you’ve pushed this concept and to find you all already discussing this application. I’m an experienced developer and a novice ee.
Assuming the jumperless could handle the voltages here, which it sounds like it probably can, my next question would be about making a stable solution for the audio jack connections. I’m thinking you’d need a solution for connecting a ton of 3.5mm modular synth minijacks (1/8") or 6.35mm (1/4") TRS cables, both for the ins and the outs, with the jumperless(es) sitting in between.
There are various formats for batched audio signal sending, though they definitely get spendy. DB25 is a standard, for example. e.g. - the Redco R196-D25PG.
@ArchiteuthisFlux , like you were mentioning, it’s possible that making something purpose-built using the proper crossbar switches might make more sense for this setup than a jumperless itself, but I’d be curious to get your take.
I frequently make the joke that I’m going to sell a special edition Jumperless for modular/audio. Which would be exactly the same but twice the price.
Yeah it’s funny how often the idea for something like this comes up. When it was my turn to come up with that same idea, it seemed really weird that no one had done this before (in an actual, useable product I could buy, there were some graduatestudentprojects that did similar things and were very helpful.)
It can handle ±9V, the whole “spec” for Eurorack is a confusing mess, but it sounds like they sometimes go to +10V but that’s hardly used? Idk, I’m not tuned into that world enough to have solid context. It should be fine though.
I did kind of half-assedly play with that idea a while ago, just an 8x8 grid of 1/8" audio jacks in place of the breadboard. (Ignore the random 3D models on this, I just used one I had)
Then there would be a cover PCB with holes and LEDs to show the connections.
I do have a few Jumperlesses without the breadboard part soldered on, so you could make something like this and just solder one on.
Once Jumperless V5 ships, I might look into getting back on the Barnacle Board
thing and make a crossbar matrix that can just plug into a 70-pin card edge connector so people can make their own boards like this and not have to worry about the really complicated routing stuff.
idk, I’ve played with the idea of making a purpose-built matrix for some other use case, but it always ends up converging into the same setup as a the Jumperless. It’s just super optimized at this point to connect anything to anything and make no assumptions about what the user wants to do with it, any other setup I play with starts to impose weird limits on what can go where.
Not to say my design is perfect by any means, but I have been staring at it and tweaking/revising things for quite a while. If I had seen a better way I’d probably have done it. And the pathfinding algorithm is absurd complicated and been battle-tested by hundreds of bug reports.
So starting over from scratch would be a whole thing, especially because it’s super open source so there’s really no need to reinvent the wheel.
I guess if it was just for audio and you were willing to pick a set of nodes as inputs and another as outputs, you could put buffers on all of them. So you’re not passing any current.
Anyway, if you want to come chat on the surprisingly active Discord server, there’s a bunch of cool people there who would love to play with this idea.
Ha I knew that name sounded familiar when the order came in. I’m going to throw in an extra one without the breadboard soldered on (I have a few left from back when I was doing kits), in case you decide to hard wire it into some audio jack patch bay or something.
