Audio: Shielding wire-runs within a amplifier chassis

[FenderBender]

Hey everyone,

About two years ago I built an amplifier from a few kit boards from eBay. It's about 60W x3 (L,R, and sub). I'm using LM3886 chip amps. They are a class AB design. I'm redoing a section of it, and I decided I would rewire the signal cables because I've had a bit of noise in the output and I'm guessing it might be from shoddy cable selection and wiring.

Anyway, the question I have pertains to the proper shielding of low level audio signals within an amplifier's chassis. I've asked some people who do audio, but they all have different opinions.

I have 2 RCA inputs for the left and right channel. The chassis is made out of wood (I know I know, not very good for EMI rejection). I have shielded twisted pair cable (2 wires and copper shield).

There seems to be a variety of opinions on how to prevent ground loops but also how to prevent the shield from acting as an antenna. To connect the shield at both ends or at one end? To connect to protective earth, or to signal ground? Many questions.

Any insight would be greatly appreciated.

Thanks.

[david77]

From my own experience it is usually best to connect the shield to signal ground on one side. I usually try to keep signal ground and PE completely seperate, connecting PE and signal ground can lead to very bad noise intoduction.

[FenderBender]

Right. Thanks.

Now have you ever heard of connecting the shield at one side only, but to PE? (Not connected to signal ground)

[david77]

Somebody else will have more insight into this, but in my experience this does not work. I guess it will work if signal ground and PE are connected but as I said that's never worked satisfactory for me. A shield not connected to signal ground does not really shield anything and PE connected to signal ground is a source for noise. I'd recommend not connecting them.

[FenderBender]

Thank you. That's actually what I had before. I ran it almost as coax, but with ground on one of the inner conductors and on the shield.

I was told this isn't the best option, but I said it worked for me.


Also, I have some pretty long wire runs in the amp. It comes from RCA to a DP4T switch which can toggle between sending the signal to the power amplifiers OR to an headphone amplifier. Headphone amp isn't an issue.

But, now in regard to running wire from this switch to the power amplifier boards, before I had kind of daisy chain. Essentially, I ran the L + R channel signal wires to the subwoofer board (which combines both L + R into one signal), and then off of the input of that same board, I sent two runs to the L + R channel. (Does that make sense??) This makes it a little cleaner and uses less wire. Instead of running 3 separate cables from the switch, I send one, and then split it off at the input of the sub board to the L + R channels. (Sorry if that's unclear).

However, again, I've heard that this daisy chaining approach is somehow bad. I'm not sure exactly why that might be. Any ideas?

Thanks so much.

[lewis]

There's some excellent insights here: http://www.douglas-self.com/ampins/groundloops/grndloop.htm and here: http://www.douglas-self.com/ampins/balanced/balanced.htm

In pro-audio gear, it is conventional to have the 0V and mains ground connected together via a 10-100R resistor in parallel with a 100nF capacitor. This provides a low impedance path to earth for any RF currents that decide to develop while providing some isolation at audio frequencies to prevent earth loop hum. It works very well... More info here: http://sound.westhost.com/earthing.htm

[FenderBender]

There's some excellent insights here: http://www.douglas-self.com/ampins/groundloops/grndloop.htm and here: http://www.douglas-self.com/ampins/balanced/balanced.htm

In pro-audio gear, it is conventional to have the 0V and mains ground connected together via a 10-100R resistor in parallel with a 100nF capacitor. This provides a low impedance path to earth for any RF currents that decide to develop while providing some isolation at audio frequencies to prevent earth loop hum. It works very well... More info here: http://sound.westhost.com/earthing.htm

Thanks. I do actually have something similar. I am using a CL-60 inrush limiter as a ground loop-breaker. This wasn't my idea but I've been snooping around audio forums for a few years, and apparently a putting a CL-60 (or equiv) is common practice. Not sure the actually characteristics of the CL-60, but people use it.

Those links are quite helpful.

[c4757p]

It will act as a "high" impedance (10 ohm) normally, but if there is a fault and the protective earth needs to be used as such, the impedance will drop rapidly because it is a negative coefficient thermistor. Not a bad idea, but I'm not sure about the reaction time.

[FenderBender]

Ah, how clever.

[c4757p]

I think I'd definitely do some tests first before adopting that solution. The protective earth can keep a chassis from becoming electrified, and I don't like the idea of a high voltage appearing on one for even a brief moment while the NTC heats up. I'd definitely experiment with it a bit - apply 170/340V spikes to it from different impedance sources and see how high the voltage actually gets. Perhaps I'd bypass it with a capacitor to handle some of the initial spike. One solution I've seen that I rather liked paralleled a 10-100 ohm resistor (don't remember the exact value), a 100nF capacitor, two diodes in series facing one way, and two more diodes in series facing the other way. The resistor is the normal connection, the capacitor bypasses it at high frequency, and the diodes quickly engage and sink any transients above two diode drops. You could easily replace the four diodes with one bidirectional low-voltage TVS diode, but at the price of 1N400x I bet you wouldn't be saving anything but a little space.

[NiHaoMike]

A pair of inverse power diodes works well for that. They won't conduct until the voltage exceeds 0.7V or so. Or use a power supply module that is double insulated so no ground or a soft ground is fine.

There's also differential and pseudo differential interconnects that inherently attenuate common mode noise and of course, optical if you want to get rid of all the problems. You'll have to upgrade to a digital amp to use optical, but that's the way to go unless you're planning to only use it with "legacy" sources.

[david77]

I think we're loosing track here a bit. As far as I understand he's got a simple multi channel analog amp with some sort of source switching/pre-amp in it and he's experiencing a bit of hum. That's no big deal.
Going fully digital doesn't really seem sensible, does it?
In my opinion it is also not necessary to use symmetrical wiring inside a simple amp. Hundreds of thousands of amps work without it and don't hum.

One question: Why would you connect PE and signal ground together? Are there any safety concerns if one doesn't? Consumer grade equipment usually doesn't, even if it's not double insulated and uses grounded (PE) chassis.
A lot of studio/professional gear allows the user to sever the connection between signal ground and PE from the outside. My understanding was that this option is provided to get rid of external ground loops and associated hum problems.

[FenderBender]

After some extended reading, I think I'll connect the shield to signal ground. No PE involved.

So I guess I'll just attach signal ground and shield at the RCA input then? Then connect them again after the switch?


How about the idea of connecting everything (L, R, + sub)  to the switch with 3 separate runs vs one run to the sub-pre board, and from that sub-pre board to the L + R channels. Not sure theoretically the difference.

Thanks all.