ok so I get and value the purpose of the matrix mixer, and intend to build one shortly.
but it got me thinking about building a matrix switcher. NOT a patch-matrix like on the ARP2500, but a matrix-mixer just of switches. I couldnt find anything at all until literally just then when i finalyl hit the right keywords and got this one back:
Im thinking of using it predominantly for logic-addiing gate trains for easy drum triggering stuff. I don;t really like programming paterns drums in multiple chains of 16 steps etc, I like effin about with shit and seeing what happens. This seems like a really potential good way to eff about rapidly…
I was thinking that each vertical row would have 2 outputs, one taking an ‘OR’ of whatever switches I had activated - ie a unity gain mixer of gate/trigger signals - and one taking a differnet logic gate (ie AND,XOR, etc) so that I could patch in 4 gate trains, then easily switch on the fly to give 4 x OR outputs and 4 x other logic outputs.
plus of course it could unity gain any analog CV or audrio rate.
does anyone think that this idea has any value?
it seems like a pretty basic stripboard build plus lots of panel wiring…
Absolutely has value!
Hinton has a very fancy (and very expensive) switch mixer: HINTON INSTRUMENTS: SwitchMix™ II (hinton-instruments.co.uk)
Great video with Steevio using it at the bottom of that page. There is also a manual which might give some design ideas.
I’ve got a very simple passive module that I use as you describe:
The top three inputs are ORed, and the switches next to them take the signals in or out of the mix. Underneath that is a passive mult, and the bottom right is just an On-Off-On switch, which switches the central jack to n/either of the others. All said it’s a useful combination for the task.
The advantages of the matrix, as I see it, are that you would minimise the need for clock mults, and you would have a better visual representation of what you were doing. My module is more flexible, but sometimes I forget which clock is where. The potential disadvantage of the matrix is that you’re stuck with the clocks that you put in. The EMW module, for example, has four inputs. Let’s say you put in clock divisions of 2, 3, 4 & 5. That limits you to the rhythms you can obtain from those values. My suggestion is that you just go with the OR summing on your module. That leaves more vertical space for inputs. If you were to go with the classic matrix layout that the EMW module uses, and if you are building in 3U format, 7 jacks fit comfortably vertically, so you could have 6 inputs (and as many outputs as you want). I think that most of the time you would find yourself using the OR output, and there’s nothing to stop you from multing an output to a general purpose logic module, which is a very useful thing to have on hand in any case. The formula for how many possible combinations you can get is (2 to the n) - 1, where n is the number of inputs. For 4 inputs, that gives you 15 possible outputs. For 6 inputs, it’s 63. That’s assuming that none of your inputs share a common factor, otherwise some will be duplicates.
thats very encouraging thanks. I had a look at the large module and that is well beyond what I will be building!
I had a bit of a think about it and some sketches in transit this arvo, and I think I can use a bunch of 2P2T swiches I have (pushbutton) so I can have them indicated by LEDs rather than just switch postion. If I didnt have em I’d go postion but I bought 50 so gotta use em for soemthing.
I think I can make up a design on homefab PCB as theres gonna be a lot of wires otherwise.
As I was designing I thought I could potentially buffer the input signal and then use the signal iteslf to drive the LEDs. This would save some wiring/complexity but would come at the cost of only being able to use square positive cycles - ie gates and triggers. If I use both halves of hte switch and keep the LEDs on a separetepower line then its a bit more finicky, but I can resistive-mix it to an Opamp hence also allow for unity gain mixing as well.
In your experience do you think having a unity gain only audio or CV matrix mixer is actually useful at all? It seems like it would be to me…
this is my rough idea. Ive used the mixing resistors as the top trace output line so I can still do all the input lines and led power rail to the switches on the main pcb. Will probably need a few links for the top row or two, but he bottwom ones will reach. there would be some really long exposed resistor legs but im ok with that. The 6-dots are the footprint for my 2P2T switches.
Drew in LED and resisitor for the top few only to get the idea.
I think I can fit 5-6 rows of these into a euro panel size.
posisbly depending on how hte LEDs fit in too I could potentially fit 2 output jacks side by side at the bottom per row for the OR and a logic. So each colum would be slightly further apart to accomodate a second output jack and LED row, thus trading width for hieght.
I would probably only do 4 columns unless I had it very compact. IF the pcb workedwell I couldprobably daisy chain them for more.
I think that a unity gain CV mixer is incredibly useful – lets you do stuff like transpose sequences. Not so sure about its utility for audio signals – you’d need to have a way to attenuate them first, otherwise your mixer op amp will saturate fairly quickly, and that more or less implies another mixer.
You’ll definitely need to buffer your inputs if you’re planning to use this to mix CV or audio. Depending on how fussy you are about pitch, you might want to use precision op amps, and irrespective of that you should be careful about the way that you compensate for capacitive loads on the way out.
Which logic chips are you planning to use, and how are you planning to convert your signals to logic level?
You can just have a variable gain on the mixer op amp, right? The mixer input should be fine, especially if the mixer stage is inverting. Of course that’s rather incompatible with using it for V/oct signals.
The potential disadvantage of the matrix is that you’re stuck with the clocks that you put in. The EMW module, for example, has four inputs. Let’s say you put in clock divisions of 2, 3, 4 & 5. That limits you to the rhythms you can obtain from those values. My suggestion is that you just go with the OR summing on your module. That leaves more vertical space for inputs. If you were to go with the classic matrix layout that the EMW module uses, and if you are building in 3U format, 7 jacks fit comfortably vertically, so you could have 6 inputs (and as many outputs as you want). I think that most of the time you would find yourself using the OR output, and there’s nothing to stop you from multing an output to a general purpose logic module, which is a very useful thing to have on hand in any case. The formula for how many possible combinations you can get is (2 to the n) - 1, where n is the number of inputs. For 4 inputs, that gives you 15 possible outputs. For 6 inputs, it’s 63. That’s assuming that none of your inputs share a common factor, otherwise some will be duplicates.
