Floating shields

[woodyc]

I just read that closed thread called "impedance question" where people were suggesting running speaker cabs in series. I've made boxes for this in systems with pair cable using speakon and banana, but I would never do it with cliff jacks and phone cables. Because in amps with grounded outputs, one cable shield (and its phone plug) is going to have half the music rms on it. The exposed voltage creates a potential for shorting and in the case of bass or pa amps, a significant shock hazard.

I'm not a fan of the marshall mods that involve floating shields either. If you want more capacitance somewhere, use a capacitor.

I gather there isn't any code that restricts floating shields (if there were, amps like the mode four wouldn't exist) but it strikes me as a really bad practice. Maybe its just me, but I think of a shield as an extension of a chassis.

[Steve Conner]

Yup. When Marshall designed the MF350, they went for a bridged output stage, and 1/4" jack speaker connectors, which would therefore have half of the speaker voltage on their shells. MF350s seem to blow up a lot.

I also dislike the practice of putting a low value resistor in the speaker return line to derive a current feedback signal. Seems like an accident waiting to happen, if the speaker cable gets grounded somewhere else the feedback signal is lost.

I did once build a series adaptor, but I made it as a Y cord with captive plugs that had plastic shells.

[bob p]

I read the OP earlier today and at first I wasn't sure that he meant by "floating shields." If you build a series Y-cable the way that I described in Moonpie's thread, then you should use phone plugs with insulating shells or you'll end up with one phone plug that's got LIVE signal on it. Similarly, if you decide to build a junction box then you'd better think about using a non-conductive plastic box or insulated jacks. You'll get an interesting surprise if you don't.

I think that it goes without saying that it's a good idea to make sure that nobody goes and grabs your speaker cables and tugs on the connectors while you're playing -- regardless of whether or not they're insulated. Having someone yank out your cable could only yield a bad outcome -- for the yanker if you didn't use insulated wiring, and for your amp regardless of whether you used insulated wiring or not.

[woodyc]

...I also dislike the practice of putting a low value resistor in the speaker return line to derive a current feedback signal. Seems like an accident waiting to happen, if the speaker cable gets grounded somewhere else the feedback signal is lost...

I totally agree. Somebody should really complain to the lazy guys who design that stuff.

In most cases, it seems like all you'd have to do is ground the output jack, put the current sense resistor on the hi side, and add one diff amp to return the signal, i.e. in amps like this...

And probably there is a more clever simpler way?

Here's one where a fet complicates things a bit...

[Dave H]

Perhaps you could do it like this?

[woodyc]

Perhaps you could do it like this?...

That seems like a sound idea, but then you'd want to reference everything (including the input jack, line out, et cetera) to that supply common, right? Which means you couldn't plug anything grounded into the amp?

[Dave H]

That seems like a sound idea, but then you'd want to reference everything (including the input jack, line out, et cetera) to that supply common, right? Which means you couldn't plug anything grounded into the amp?

Nothing is referenced to the power amp’s supply common which is not grounded (It’s the negative feedback signal). The ground is where I have drawn the chassis ground symbols. Everything “including the input jack, line out, et cetera” is referenced to that point (ground).

[Steve Conner]

I guess that would work. One of the schematics posted earlier grounds the emitters of the output transistors, and drives the speaker from the centre tap of the power supply. But then they miss the chance to insert the current sense resistor in series with the emitters, they put it in the speaker return as usual. Well, assuming I read the schematic right. It all depends which of those ground symbols is connected to the chassis.

The only "gotcha" is that power transformers have a lot of interwinding capacitance and that is now in the signal path.

[R.G.]

Because in amps with grounded outputs, one cable shield (and its phone plug) is going to have half the music rms on it.

This is problem #1. Speaker cables in general don't need a shield, as the signal is large amplitude, low frequency (compared to what could couple in, anyway) and low impedance. Further, using a shield for return is problematical. Many solid state amps have that damped inductor in series with the output. That's there to prevent capacitive loading from making the amp unstable. Capacitance on the speaker output is a Bad Idea in general, and shielded cable makes it worse. Speaker cables should be twinlead or twisted pair. One could put a shield over that as long as only the sending end was connected.

We got to here by the cheapo practice of using 1/4" phone cables for sending power (audio power) signals to speakers instead of something more robust. They're primarily line-level signal connectors. This almost demands the misuse of shielded cables with 1/4" phone ends to connect speakers.

Dang, talk about slippery slopes!

I gather there isn't any code that restricts floating shields (if there were, amps like the mode four wouldn't exist) but it strikes me as a really bad practice.

Safety codes say that any exposed electrical conductor must not have more than ... um, used to be 42V peak AC or DC, and no more energy/current possibility than ... um, can't remember, but the idea is to not have high currents heat up things like jewelry. I think the current limit is about 1A for voltages less than the 42V potential. I'd have to go back and look.

Of course, the maker would say that they never intended, and probably somewhere in the users' manual say not to use a coaxial cable.

RF is a different beast. The shield HAS to carry the return currents. I don't know how the rules apply to that. Maybe a secondary safety shield.

I also dislike the practice of putting a low value resistor in the speaker return line to derive a current feedback signal. Seems like an accident waiting to happen, if the speaker cable gets grounded somewhere else the feedback signal is lost.

That would be a problem if that was the only feedback signal. Generally these amps have both voltage and current feedback. Shorting the current feedback converts it back to a voltage-feedback amp, which may change the gain, but doesn't make it smoke. Usually. Don't know about Marshall, which has a history of bleeding-edge designs.

I totally agree. Somebody should really complain to the lazy guys who design that stuff.

That would be the newly-minted MBA with an eye on the CEO office. I can almost promise the guys who designed the stuff didn't want to do it that way.

[Dave H]

The only "gotcha" is that power transformers have a lot of interwinding capacitance and that is now in the signal path.

I've never tried it in an audio amplifier (that's why I said "Perhaps") but it did work fine in a electromaget power supply.