Hey. I'm trying to make a relay matrix. For starters I was going to give it a try as a 4x4 board with latching relays. I'm thinking about putting together the circuit based on the schematic I drew up in the attachment. Would there be a way that would work better? I'm a bit confused by the base current to the transistor in the upper left hand corner. It seems like I would need a larger resistor to be getting base currents in the microamps.
In the drawing, the first four relays would be the top row of relays, the fifth relay would begin the second row. This would repeat for four rows, but the schematic was getting messy. So the top four relays are connected to the same output pin on one side of the coil. The other side of the coil is connected to one of four other output pins. So based on where the circuit completes would dictate the relay that was powered on.
I'm thinking of using latching relays so that I can keep the overall current draw down by closing the relays in series with delays in between.
Does this seem feasible? I would also need a reasonable way to unlatch the relays either with a reversed polarity or using double latching coils. I eventually want to do this on a large scale (32x32, 1024 relays total)
Haha sorry I was focusing more on the best way to supply power to the relays than what I'm actually doing with them. I'm trying to make a digitally controlled patch-bay. So basically in this small prototype I want to have the option of routing audio signals from four outputs to four inputs. So sixteen total combinations
If you're routing audio through relays then you need high quality relays. Getting them in a latching configuration and then getting more than 1000 of them seems like it will be expensive. It also sounds like you're putting each signal through two or more relays to reach the output, so the impediment of each relay contact will add up.
Audio multiplexer chips are not expensive. You can get all sorts of combinations of inputs and outputs. Do the switching in solid-state where there's no contacts to wear out or gather dust.
These days I'd expect to do it all in digital processing. 32 inputs can be digitized and manipulated in software. If you need 32 analog outputs then 32 DACs will be required. That initially seems like a lot but the data rate for audio is not too high to do this on one single processor. (Not an Arduino.)
Id like to keep all the audio in the analog realm if it seems possible to do. A relay contact has to be fairly similar to a trs patch cable, no? Those are used in much longer runs in studios patching in and out between multiple pieces of gear.
Of all the research I've done it seems the best monitoring controllers use relays to route audio between different speakers and inputs. I figured the relay route would be the cleanest signal path since it would only pass through the two contact points, similar to patch cable.
Look for "analog crosspoint switches" Mouser have a huge range 15/20 pounds for 4x4 and 8x8. These can deal with video as well as well audio. Forget that relays exist!
If you're determined to continue with a relay 'matrix' - you already expressed one way of thinking about it.
Relay 5 is the first device on the second row... so just imagine your matrix unfolded to a single chain.
Then you can imagine a string of shift registers as the interface between micro and relays.
(Remember the first relay is typically #0)
How about a bank of TPIC6B595 shift registers? Each can drive 8 relays.
I offer a board now that will drive 96 outputs for up to 150mA coil drive, and have done other designs with 325 outputs.
@crossroads - have you looked at the 4066 or 4016 (it's 4 solid state relays) in one package at the cost of 1 of the transistors used to control the relay.
Providing your not trying to switch power then the link to crosspoint switches should give you a lot if ideas for products. But it's the old XY thing you have to ask "WHAT ARE THEY SWITCHING".
holmes4,
No, I wouldn't consider using 4066/4016 for switching audio signals around. They both have on-state resistance, and will add in power supply noise. If anything, I would go for parts with +/- power supply to eliminate the need for capacitive coupling and DC offset of the signals.
CrossRoads:
holmes4,
No, I wouldn't consider using 4066/4016 for switching audio signals around. They both have on-state resistance, and will add in power supply noise. If anything, I would go for parts with +/- power supply to eliminate the need for capacitive coupling and DC offset of the signals.
Opto FETs have an on resistance of ~200 ohms, no supply noise.
TomGeorge:
Hi,
Have you looked at how you control latching relays?
Yeah I have some NEC latching relays that reset with current from the opposing direction. I'm trying to work out the best way to use the least amount of pins and keep it as neat as possible. I did look into double coil relays, but would need a total of 128 output pins.
LarryD:
Opto FETs have an on resistance of ~200 ohms, no supply noise.
But we don't know what the budget is.
.
I'm trying to keep it around $1500 total.
CrossRoads:
What kind of connectors do you plan to use? That could be a big cost driver.
I plan on using TRS connections for input and output. DB25 connections would save money, but I have cabling already set for TRS connections.
CrossRoads:
How about a bank of TPIC6B595 shift registers? Each can drive 8 relays.
I offer a board now that will drive 96 outputs for up to 150mA coil drive, and have done other designs with 325 outputs.
Using the shift registers with the relays seems like a really solid plan. The relays you linked are super affordable, but being SPST I couldn't use them to send a balanced signal. They have a low enough coil current though to not have to worry about using a latching relay.
Ah - double pole, single throw (DPST) is what you need. 2 sets of contacts, Not connected, or Connected when energized. Either way, two sets of switches, a bit over twice the cost of using two SPST.
CrossRoads:
Ah - double pole, single throw (DPST) is what you need. 2 sets of contacts, Not connected, or Connected when energized. Either way, two sets of switches, a bit over twice the cost of using two SPST. HE722A0500 Littelfuse Inc. | Relays | DigiKey
Using the non latching relays in this way though, isn't the coil current too high to have all 32 energized at the same time? (32 would be the most possible given the audio IN/OUT of the unit)
32 outs connected to 1 in?
Current is fine. TPIC6B595 can sink 150mA on each output, the coils only need 25mA, 8 x .025 = 0.2A.
4 chips with all 8 outputs on will sink 0.8A. 2A, 5V supply for coil current and you're all set. I've used this one in projects http://www.mpja.com/5-Volt-DC-Plug-Power-Supply-4A-Regulated/productinfo/18520+PS/
In case you decide to connect all ins to all outs for some reason. You wouldn't ever connect 1 out to multiple ins, tho, would you? That would result in the 2 ins interfering with each other as there is no isolation between them.