Relay Switching Matrix Shielding Best Practices?

[a soBer Newt]

Hello All I have a request from a client about making a switcher that will switch between his two tape machines and radar system and two different consoles. Image attached showing signal flow. My question is about how to handle grounding in this situation. Can I just tie each machines ground together and then switch that across to the output? Or am I better off to dedicate a relay for each channels ground.

 

[sodderboy]

First, you are dealing with cable shielding and not equipment grounding. Think of it that way and things get clearer.
This is typically done with an EDAC panel that gets designed, financed, built, and never used after a couple of months. I figure the cost of relay boxes will be 2-3x the cost of a panel because the terminations are not as simple and brainless. You don't have to think much until it is time to block.
Shielding depends on what you have now. How are shields tied currently? How I wire, the shields would be tied at all consoles and lifted at the opposite ends at equipment.
The bestest gold star way would be to switch the shields as well, and load unconnected tape machine inputs with 600Ω. I'm dealing with older stuff, and some things do not like to sit with open inputs. Especially in NYC. This would 100% mimic an EDAC panel with termination plugs for unused machine inputs. Anything less would have to be tested in situ first before scaling up. The first compromise would be linking all the CH1 shields together, CH2, and so on in a single box. You want to have single shielded twisted pairs connecting the equipment as best as possible. When you start to link shields between different equipment the quiet lo-noise floor balance can be disrupted.
If you are thinking of taking the existing wiring, cutting it, and wiring it to a box filled with relays don't do it. Just don't do it. Any serious place, like commercial or rockstar home studios would never do this. They have EDAC panels. Dusty EDAC panels.

Mike

 

[ccaudle]

The bestest gold star way would be to switch the shields as well

The shields would not successfully operate as shields for high frequencies if you do that.
The bestest way for decently designed equipment is that shields are shields and fully surround all of the signal connections all the time, i.e. tie the metal of all the chassis together.
Whether your equipment is decently designed or not will have to be inspected on a case by case basis.

 

[Tekay]

How about using ULN2001 or a similare circuit to drive the relays? No extra protection diodes needed and one port can control more than one relay.
I've used this circuit in several products together with 12v relays and a LM2596 adjusted to +10v to save heat and relay power.
I've used a 2k7 resistor in serie with the control inputs.

 

[ccaudle]

Can I just tie each machines ground together and then switch that across to the output?

Those devices all have balanced I/O, so the connections are signal hot, signal cold, and shield. Assuming that none of the equipment has a pin 1 problem (i.e. shield incorrectly tied into circuit reference node such that shield interference currents become noise in the circuit) you can tie the shields directly to the chassis of the switching matrix and switch the two signal connections with a dual pole relay.
If all the equipment has 3-pin power entry, you could selectively not connect the shield at the receive end for equipment which has a pin 1 problem.
Hopefully all of the equipment has the shields properly connected, because you might have to evaluate inputs and outputs separately for pin 1 problems.
Connecting any equipment which does not have 3-pin power entry would require separately tying the chassis of that device to the same earth potential that all of the other equipment is referenced to so that it does not have a high common mode voltage from power supply leakage.

That is essentially just restating what sodderboy recommended as well, but the only disagreement I have is the suggestion that also switching shields would be a good idea. If shields are useful it implies that the shield is protecting the audio signal connections from some interfering signal. If you then switch the shields, now you are taking that interfering signal and bringing it inside the box in a semi-uncontrolled way. That has a good chance of rebroadcasting the interference signal around the inside of the chassis, and keeping the interfering signal out of the chassis was the entire point of shielding to begin with.

 

[sodderboy]

The shields are already in the box, at the relays, what are you going to do with them ccaudle? How do you keep shield purity?
In a completely fresh design with these relay boxes in the middle, I would break all shields at the boxes and tie at sources and loads. All noise current drains to either end. Not my fave, but it works. But this wiring is already in place to some extent.
In the case of either box, switching shields always has the load end of a source shield(connected to source chassis) either connected to nothing, or to the shield of the cable going to the selected load. Both boxes. Always discreet cable runs. With my first-tier compromise the source shield of one channel would connect to the shields of the three load devices, vice-versa for the other box. This would have to be tested with modded mic cables connecting a few channels to make sure the existing noise floor does not change.
I absolutely disagree that relay switching cable shields does not effectively shield “high frequencies”. Please educate.
Mike

 

[ccaudle]

The shields are already in the box, at the relays

Where does that come from? The first post referred to a request to make a switcher, so I assumed that it does not exist yet. If it doesn't exist yet then there is no a priori reason to assume the shields are already anywhere, the yet-to-exist design will determine that.

I absolutely disagree that relay switching cable shields does not effectively shield “high frequencies”. Please educate.

Shielding effectiveness is determined by the longest dimension of an aperture in the shield and the wavelength of an interfering signal. Once a hole is large enough in a shield (in any dimension, can be a narrow slot) the interfering signal goes right through, shielding effectiveness goes to zero, or nearly so.
If you are switching through a relay, that would imply that the shield (which up to the switch box has been a continuous tubular shell around the signal conductors) has to be routed as a wire or PCB trace to the relay matrix. That means you have created an aperture which is at least the entire width of the relay matrix, which lowers the frequency at which the shield stops being effective to a very low frequency.
Additionally, any RF energy which is traveling on the shields is now inside the box on a long thin wire. If you look at a hand held radio/walkie-talkie you will be reminded that a long thin wire is how you make a whip antenna. So now the local broadcast tv stations which the signal wires had previously been shielded from are now inside the chassis being re-broadcast from a whip antenna right next to the signal wires, and reflecting around the chassis because now the RF waves are on the inside of the conductive box instead of the outside.

 

[a soBer Newt]

As of now no cabling has even been run at the studio so terminating the shielding can be done however will work best. The tape machines are an Ampex MM1200 and a Stephens 24 track. The main tracking console is a Calrec J series, the monitor console may be an Allen and Heath but this may change. I have only made up a small proto relay card to test the latching relays so its still very early in the design stage.

 

[sodderboy]

Oi vey!
Bro, you’ll never build anything connectable in your world. Hardwire stuff together for the sheeld integrity?!? Get out of the theory and into the real world.
Thanks for the cool theory! Now, how do you wire the shields within your limited constraints?
Mike
Added edit: now given the equipment list I did not see on my phone, I would do everything to steer the owner away from a custom relay situation. Absolutely. EDAC panel or I'll go fishing instead. And that would maintain the tightest shield integrity, reliability, etc. Cauddie approved!
And when they ask, "Um, we said we would never want it, but, like can we mult a machine to two desks?" you can build an EDAC mult and an extra EDAC patchcable and make it dunn. Yay hero! With the relay box, cool as it is, switched shields or not, you gotta mult at patchbay or XLR's like they do now kinda. Doable but clumsy.

 

[ccaudle]

Hardwire stuff together for the sheeld integrity?!?

Now you are putting words in my mouth. Properly connecting shields to connectors and connectors to chassis is not the same as hardwire (at least if I am interpreting your use of the term correctly to mean permanently attached together). I'm not advocating for anything more than recommended in AES48.

 

[sodderboy]

More talking and no answering.
“The wires are in the box” because I built two prototypes in my head already.
You’ve got aes48 in yours.
Do you also have an idea of what you would do with the shields? There are 120 of them in the 3x1 to get you started.
Don’t cheat and copy ISO 16772.

Mike

 

[RSRecords]

I built something like this for a mixing engineer who shared a space with a mastering engineer. Worked great. Been in service for about 3 years now. Used Panasonic TQ series relays.

Swapping between patchbay for mixing and mastering insert router. Shields were all tied together at the chassis. I did build it in a way to be flexible with shields in case an issue arrived.

 

[ruffrecords]

More talking and no answering.
“The wires are in the box” because I built two prototypes in my head already.
You’ve got aes48 in yours.
Do you also have an idea of what you would do with the shields? There are 120 of them in the 3x1 to get you started.
Don’t cheat and copy ISO 16772.

Mike

It is very simple really. Assuming you have XLRs in and out then just connect each and every pin 1 direct to chassis at the XLR.

Cheers

Ian

 

[ccaudle]

More talking and no answering.

I don't follow what prompted this response. Do you have a specific question about what I proposed? The physical specifics of how you properly connect the shields (or lift if that is what is decided) will of course vary based on the specific connectors used (XLR, 24 pin Dsub, EDAC rectangular, etc.).

Do you also have an idea of what you would do with the shields?

Shields connect to the chassis if you have properly designed equipment without pin 1 problems. If you have to accommodate equipment which is not properly designed then connect the shield at the driver end and lift at the other end. As I said above the physical specifics of that would vary. With XLR connectors, pin 1 connects back to the shell; with Dsub it would depend on whether you have individually shielded pairs (that seems the most common), so same, the shield pins would get a short direct connection back to the shell/chassis (although Dsub can be a special case if it has an overall shield; connecting the shield as the receiver end is most effective at high RF frequencies, so if the overall shield is handling most of the RF rejection, it might be easiest to just not connect all of the internal shields at the receive end and only have the overall shield connected to the connector shells at each end).
I don't believe that there is an industry standard for EDAC rectangular panel connectors, so that is up to the individual system designer to pick the pinout, correct? The same principles apply, overall shield should connect the connector shells at each end, if the pairs are individually shielded and connected to pins then those shield pins should connect back to the chassis with a short and direct a path as possible.

 

[Brian Roth]

FWIW, Audio Accessories has “Standard Audio Configuration” for various EDACs:

https://patchbays.com/pinouts.htm

Bri

 

[sodderboy]

It would double-down on insanity to design a 24 ch 2x3 relay box with connectors on the back. And you open it to make sure that its not just empty inside, and find that all the shields are bused together at the connector panel? I would love to see 240 XLR's on the backs of two relay boxes. I think they will fit on 8U, which kinda gives space for the 120 1' cables inside. Each box. Definitely Mogami multipair and different colored shrink on the ends to keep the eqpt groups separate. Who am I to kill a dream? Go for it! I volunteer the qty 250 2.5" RS green 7-strand for the chassis lugs post paid.
Mike
PS: Ian, didn't the beeb wire rooms with UTP? Pre aes48 that is.

 

[Matt Syson]

The BBC did indeed use twisted pair multicore for balanced line level with no problems. So 25 pairs in an overal foil screen with one 'drain' wire. A certain establishment in London had a room around 3 metres wide and IIRC about 10 metres long with both long walls covered with 'post office' Christmas tree blocks (200 contact points) and all the multicores dropped the screens off where the cables entered the room and all interconnections were done in twisted pair (wire wrapped). They handled all inputs and outputs of all gear in 50 small studios. Tapes and grams and post amplifier mic feeds. After decades of use when the room was stripped as studios went digital they discovered 10,000 wire wrap posts that they had forgotten to solder over the years.

 

[amplexus]

The BBC did indeed use twisted pair multicore for balanced line level with no problems. So 25 pairs in an overal foil screen with one 'drain' wire. A certain establishment in London had a room around 3 metres wide and IIRC about 10 metres long with both long walls covered with 'post office' Christmas tree blocks (200 contact points) and all the multicores dropped the screens off where the cables entered the room and all interconnections were done in twisted pair (wire wrapped). They handled all inputs and outputs of all gear in 50 small studios. Tapes and grams and post amplifier mic feeds. After decades of use when the room was stripped as studios went digital they discovered 10,000 wire wrap posts that they had forgotten to solder over the years.

My understanding has always been that properly done wire wrap didn’t need to be soldered. Emphasis being on “properly done”