migrating from Minimal mixers
OP:
Seems like it would require a ton of VCAs with digital control if you wanted to keep the signal path analog or some sort of digital mixing system - you’d essentially be making a K-Mix into a module with external control!
Not saying it wouldn’t be amazing, but it might be minimal only in aesthetic 
Yep. I think 4 dual op amps on the input stage to split and buffer the signal, 4 VCAs, 2 more dual op amps for the outputs, a 16 channel digital output arduino, and a couple DACs? I’ve never designed anything, only built a lot of kits. Part count would be high but the circuit and programming very straight forward I would think (no clocks, ADCs, or actual processing of audio).
I have similar experience (/lack thereof), but this idea is really interesting!
Do you mean 4 quad VCA ICs like the V2164? Maybe I’m missing something, but it seems like you would need 2 VCAs per input channel (16 so far) to handle panning if that would be done by just splitting the input, plus a VCA per aux send (8 for 1 mono aux send, since you’d want to control each channel’s aux level), so that would be 24 total VCAs and a DAC gain control for each one. Mutable’s Frames does 4 channels of digitally controlled VCAs, so maybe something could be learned from how Olivier handled the DAC and analog section of that.
The multiple page idea would seem to translate best to a grid, since there wouldn’t be any visual feedback as to channel settings on the 16n once you switch pages.
1 Like
You are right. Actually, I think there may be more needed. There’s technically 4 buses: Main Left, Main Right, Aux Left, Aux Right. So each channel needs a VCA for each bus which means 32 total VCAs or 8 quads 
plus DACs for each and an arduino that could handle 32 digital channels of output - which doesn’t appear to exist. Maybe someone who actually designs circuits could chime in or if enough people care I’ll take it to a new thread.
I already thought about how to give feedback for the Aux page for a 16n user - I was thinking some sort of lighting scheme with the 8 channel LEDs to indicate the recorded position and a “move-thru” protocol for the slider to be set. The idea being that it would remember positions from page to page but also if the Grids or 16n were disconnected as is the current functionality of a Monome module. I forgot that there are two extra buttons that could potentially be used as well.
Another possibility would be cutting it down to 4 channels and using the other 4 jacks as audio pass thrus which might be nice for someone wanting to multi-track record at the same time. This cuts the part count in half and allows for all the channel functions to appear on the same Grids/16n page. 8 channels is more appealing though.
2 Likes
Agreed about a thread if any discussion of this should continue just to avoid further derailment, but one thing that perhaps someone with DAC/hardware design could reflect on is that you likely don’t need independent digital outputs from the MCU to control independent DAC channels. Frames only has 3 lines connecting the MCU to the 4-output DAC, for example. I don’t understand how that works, but someone does
DACs typically controlled by SPI (serial) protocol. for example, 24b control word might include 4b designating a command (e.g. “update a channel”), 4b designating channel number, 16b for value. all slaved parts using shared clock/data lines, each part with its own chip-select line (GPIO.)
whether a bunch of vcas are necessary for a line level mixer, is another question. maybe consider simple attenuators with digital pots if you don’t need individual buffered outputs per channel.
if you want to make a new thread i can move these posts.
1 Like
Thanks @zebra!
Just to recap and update from a few thoughts from PMs that aren’t in the thread…
- Teensy appears to be a viable platform for the development that would support more than enough channels - the Teensy LC should be adequate?
- The V2164 seems to be the best choice for VCA although as was pointed out, a lower cost digital attenuator may work just fine and reduce part count… this leads me to another thought
There’s two iterations I’m thinking about - maybe one is better/more popular than another:
-
8 channel version: channels use a reasonably priced VCA like the V2164 or a digital attenuator for lower parts count; good sound quality and dynamic range.
-
4 channel version: 4 inputs with 4 through channel outputs for multitracking; channels use a very high quality VCA like THAT 218x paired with an opamp; very high sound quality and opens up the possibility of incorporating the Teensy ADC for a basic but very usable limiting circuit on the bussed outputs. (I guess the limiting circuit would work for the 8 channel version but would probably sound best on the higher quality components).
1 Like
Just a note that y’all might want to check out this thread if you hadn’t seen it yet.
This SSSR Matrix mixer is digitally-controlled and talks over I2C. Since it’s DIY, I bet you could gather some idea of what that project uses for components (though just a note to be careful and double-check to see if it is open-sourced and what the license is should you make use of the design in some meaningful way). I bet there’s probably some pretty similar things inside.
1 Like
Thanks for this! I was literally typing out that the Teensy can communicate over I2C so may be able to be controlled via TT. Not sure how useful that is but its something to think about. Also, I think I’m running out of PCB real estate. I’ll head over and check it out.
1 Like
also I just realized that the LED pattern is 8x16 (same as the 128), and since there’s now some grid-integration stuff built into Teletype, it’d probably be a not super daunting project to hack together the 128 to actually control this matrix mixer if someone was interested in that as a project!
1 Like