You're thinking this kind of configuration, yes?
1 in to 1 or more outs, with up to 32 ins and 32 outs?
(3rd wire of TRS not shown, that is the common Ground for all devices?)
I'm actually thinking even simpler than that. One output would never go to more than one input. Everything on the audio side would be wired in series. So like in the drawing you have output one would only to go to one input(1-4).
In your drawing, would you also have relays at each of the crossing points?
Yeah, I guess that would make more sense. Otherwise every input is connected to every output all the time.
Harder to draw as a schematic. Need a relay at every dot.
Yeah then that's definitely the idea. I was thinking the same thing just do a shared ground. All seems good to me!
Yeah, lot more parts that way. 1024 relays, 1024 diodes, 128 8-bit shift registers. 11 of my card for controllers.
Can you buffer the input to make it single line in to the relays, then buffer again going out to feed the TSR output?
Do you need 48V phantom power going out also for mics?
CrossRoads:
Yeah, lot more parts that way. 1024 relays, 1024 diodes, 128 8-bit shift registers. 11 of my card for controllers.Can you buffer the input to make it single line in to the relays, then buffer again going out to feed the TSR output?
Do you need 48V phantom power going out also for mics?
I could buffer the input. You're saying just so I could use the single pole relays? I wouldn't see any advantage other than that.
I wouldn't need phantom power, I'm always gonna run mics direct to the preamp.
You will eventually go to a PCB, yes?, to alleviate a ton of hand wiring.
If you buffer the inputs and buffer the outputs you can cut out a ton of wiring, design a board with ground planes, and I think end up with a quiet system.
So something like this, this is 8 inputs to 2 outputs so far.
I think the original SPST relay I linked to is thinner, I had a library symbol for this one already tho.
That PCB makes way more sense. I like the way you tied the common pins together for the relay contacts. I don't know why I was thinking any different than that.
Wouldn't powering the relays be much easier using latching relays? Only two relays would need power at any given time. I could put delays between powering them on and off.
Latching relays means a driver that can pulse current in two directions. Or two drivers if the relay has separate latch & unlatch coils - That'd probably be simpler, just double up the shift registers.
The least expensive latching relays are 1.456 each in 1000 lot, nearly three times the price of the non-latching SPST we started the discussion around. So there's your whole $1500 budget there in just the 1 part.
I don't understand the concern over keeping two relays powered - that only needs 20 milliamp for two of these
if you are sticking with dual signals per path.
latching relays come in two flavors. single coil (reverse voltage to alternate)
two coil, one to open, one to close, common ground.
to power 36 (2 coil) relays, you can 24 signals 6 individual to send the open signal and 6 to bring coil to ground
ditto for the closed coil side.
or for 36 single coil latching relays , you need 12 signals and and h-bridge (or two relays)
not going to get you close, but for any large quantity, I can see the desire to not have constant power drain.
Yeah I was running through a few things in my head about the latching relays. Without latching relays, when running the system at max potential, all 32 at once the total current draw for the relay system alone would be over 300mA. With dual coil latching relays I can get away with a total current draw at one point of only around 30mA. I can actually power that directly from an arduino.
Also when I was going over that with the latching relays, I was realizing that some gear would never have to be run to more than one input or output. If I used dual throw relays I can use the normally closed position to send direct from preamps to the AD. This would save me 12 possible relays in the matrix. So basically the AD inputs would never have to see an input from a preamp other than the one that corresponds to it's input. PRE 1 will always route to AD IN 1. This would make it so that one preamp could never be routed to another preamp, but I would be fine with that. I can also use a normally closed position from my DA outputs to route to my tracking monitoring. When tracking I would never need to use the DA for anything other than a direct feed to my headphone distribution.
CrossRoads:
Latching relays means a driver that can pulse current in two directions. Or two drivers if the relay has separate latch & unlatch coils - That'd probably be simpler, just double up the shift registers.
I guess I'm getting confused on the need for so many shift registers. Wouldn't each shift register bump my output pins by 8?
Even with dual coil latching relays I would only need to output 128 pins. I would send around a 500ms pulse down one side of the 32 relays and another down the other side. The coil they meet at would energize and stay latched when power is released. I could then do that with another relay...etc.
Okay, so a setup like this then.
MIC5891 is a shift register with high current source output.
TPIC6C595 (or 6B595) is a shift register with high current sink input.
Turning on one output each of Open+ and Open- will open a latch at their intersecting relay,
while turning on one output each of Close+ and Close- will close the latch at their intersecting relay.
This provides the latching function at the cost of more expensive dual coil latching relays. Each 'row' now needs two shift register outputs and each 'column' now needs two shift register outputs.
32 x 32 = 64 row outputs, 8 sink shift registers, and 64 column outputs, 8 source shift registers
You can power down the shift registers and keep the relay settings.
Probably want to store those in non-volatile memory first. Or clear all the settings and start fresh every time so you know what's what.
On the 'plus' side only DPDT relays are available at reasonable costs so you don't need to buffer the analog input & output.
(Reposted attachment, had the audio output signal names wrong)
Datasheet page 3, top left, I see two distinct sets of coils.
I suppose the -s could be connected, and the OPEN+ or CLOSE+ would determine which coil was driven wnen the - was driven.
Could save on TPIC6x595 shift registers. Would any isolation be lost relay to relay?
(ooh, that was an ugly mod to the schematic!)
Isolation lost in the audio path? Or through the coils?
I think that last schematic seems pretty much exactly what I'd like to do. I think given that I can use the normally closed to send preamps directly to the ad, I can cut down the relays to a 20x20 instead of 32x32. That way I only need 400 relays instead of 1024. That'll help keep the cost even lower than my initial budget.
Going back to the audio layer...
Using relays or other passive crosspoints in the matrix, you will eventually see some loading into high-z inputs when one source is routed to multiple destinations.
I'd strongly suggest testing some of your switching ideas before committing to a technology.
Also consider capacitance of the input wiring that unloads into the destinations when the switches are first closed - with a click/pop or other undesirable effect
"Isolation lost in the audio path? Or through the coils?"
I was thinking thru the coils, but when I posted that thought I didn't have the coils wired correctly (there was an intermediate file that I pulled & replaced).
Each input will feed the NO pin of 20 relays (down from 32 as you say), I don't know how that will affect the signal (maybe a little loss of high end? from the 75mOhm series resistance and the capacitance of the connecters, board traces, and the relay), and you may want to turn down any amps that are being fed before switching relays.
Hi,
Did anybody google audio switch matrix IC or switch matrix IC
They would make life so much easier. if you re just switching audio.
Tom.... 
The schematic Crossroads posted would work for a dual coil relay.
such a project of 4x4 or 6x6 would not be that hard to implement and test.
make sure you haven enough power to drive the coils. other that that, the mechanics of the electrics are good.
the audio side quality will be dependent on contacts and relay quality.