I am planning to add RCA and/or BNC output chassis connectors to a device that already has a TOSLINK connector. When doing so, is it preferable to install the connector uninsulated, relying on the chassis for grounding, and only wiring to the S/PDIF signal? Or is it preferable to insulate the connector from the chassis and wire to both signal and ground in the TOSLINK circuit? I can see pros and cons to both approaches.
Signal 0V (ground) is usually connected to chassis (if at all) at one point, I'd insulate it. Also digital and audio grounds are largely kept separate. What pros and cons do you see?
As I understand, chassis is only for shielding purpose. You should not connect the signal return path to the shield (chassis). This may also create a ground loop.
Looking for something a little more in depth. It's not a DIY device, it's commercial. The chassis comes into contact with the PCB ground plane in at least 10 places. I'm not so sure about the ground loop issue, as I bet somewhere in the chain there is a digital filter that will take care of that.
S/PDIF is still a two-wire connection. The ground/return of the interconnect should connect to the ground/return of the device and not be routed via the chassis.
My preference would be to connect the shield of the connector straight to the chassis (best for RF shielding), to put a little S/PDIF transformer right behind the connector (to eliminate ground loops that pass through the PCB) and to use a thin 75 ohm coaxial cable to connect the primary side of the transformer to the driving circuit.
The chassis comes into AC/RF contact with the PCB ground plane in at least 10 places.
Comparing options without transformer:
1. Single wire from interface circuit to connector, return via chassis, large loop area between wire and chassis:
The loop will act as a magnetic antenna and radiate the S/PDIF signal all over the circuit. Besides, the self inductance of the loop will make the impedance matching worse. I think this would be the worst option.
2. Single wire from interface circuit to connector, return via chassis, wire glued to the chassis where possible to minimize loop area:
much better than option 1, but still not very well-controlled
3. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector connected to the chassis:
Extra connection from the chassis via the cable to the PCB, may or may not give hum loop problems, depending on loop area and what 50 Hz or 60 Hz magnetic fields are there inside the device. Keeping the cable as close as possible to the chassis can help to reduce the loop area.
4. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector insulated from the chassis:
Shield of the coaxial cable can couple in RF interference picked up by the external S/PDIF cable. If the external S/PDIF cable is part of a ground loop, it will inject whatever it picks up straight into the PCB ground (rather than into the chassis ground).
5. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector insulated from the chassis, but decoupled to it with a 100 nF or so capacitor with very short leads:
Much reduces the RF coupling issue of option 4.
I don't see an ideal solution in this list...
1. Single wire from interface circuit to connector, return via chassis, large loop area between wire and chassis:
The loop will act as a magnetic antenna and radiate the S/PDIF signal all over the circuit. Besides, the self inductance of the loop will make the impedance matching worse. I think this would be the worst option.
2. Single wire from interface circuit to connector, return via chassis, wire glued to the chassis where possible to minimize loop area:
much better than option 1, but still not very well-controlled
3. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector connected to the chassis:
Extra connection from the chassis via the cable to the PCB, may or may not give hum loop problems, depending on loop area and what 50 Hz or 60 Hz magnetic fields are there inside the device. Keeping the cable as close as possible to the chassis can help to reduce the loop area.
4. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector insulated from the chassis:
Shield of the coaxial cable can couple in RF interference picked up by the external S/PDIF cable. If the external S/PDIF cable is part of a ground loop, it will inject whatever it picks up straight into the PCB ground (rather than into the chassis ground).
5. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, both shield and centre of the connector connected to the interface via the coaxial cable, shield of the connector insulated from the chassis, but decoupled to it with a 100 nF or so capacitor with very short leads:
Much reduces the RF coupling issue of option 4.
I don't see an ideal solution in this list...
I forgot this one:
6. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, centre of the connector connected to the interface via the coaxial cable, shield of the connector connected straight to the chassis and via a 100 nF capacitor to the shield of the coaxial cable:
Similar to option 3, but with an added capacitor between the shield of the connector and the shield of the cable. Solves any potential hum loop issues of the internal wiring of option 3. Cable should still be kept close to the chassis if possible.
6. Thin 75 ohm coaxial cable (or just two wires very close to each other) from interface circuit to connector, centre of the connector connected to the interface via the coaxial cable, shield of the connector connected straight to the chassis and via a 100 nF capacitor to the shield of the coaxial cable:
Similar to option 3, but with an added capacitor between the shield of the connector and the shield of the cable. Solves any potential hum loop issues of the internal wiring of option 3. Cable should still be kept close to the chassis if possible.
You may want to look into these BNC connectors if you are going to use an approach which does not tie the shield to chassis.
https://cdn.amphenol-icc.com/media/...on/datasheet/inputoutput/io-bnc-filter-rf.pdf
https://cdn.amphenol-icc.com/media/...on/datasheet/inputoutput/io-bnc-filter-rf.pdf
What is the source of the SPDIF Signal?
Usually there is a Transformer to isolate the SPDIF in the Source unit from the Receiving unit. The 'Cold' of the Coax O/P shouldn't be 0. Volts or Chassis Potential.
The Signal for the Toslink almost certainly won't be able to 'drive' a Transformer so some sort of Buffer would be needed. A 74HCU04 will do the job with gates paralleled.
See attached circuit.
Cheers,
P.
Usually there is a Transformer to isolate the SPDIF in the Source unit from the Receiving unit. The 'Cold' of the Coax O/P shouldn't be 0. Volts or Chassis Potential.
The Signal for the Toslink almost certainly won't be able to 'drive' a Transformer so some sort of Buffer would be needed. A 74HCU04 will do the job with gates paralleled.
See attached circuit.
Cheers,
P.
One of these will do..................
http://cpc.farnell.com/w/c/electron...mers/prl/results?st=digital+audio+transformer
P.
http://cpc.farnell.com/w/c/electron...mers/prl/results?st=digital+audio+transformer
P.
- Status
- Not open for further replies.