Question from inSync reader Jan D:

"I need to hook up the same pair of passive loudspeakers to two non-identical amplifiers, featuring a sort of bi-wiring. The equipment works fine when the speakers are hooked up to one amplifier at a time. The need for bi-wiring arises from the fact that I am unable to get rid of a ground loop when connecting PC sound system to hi-fi gear except when separating the TV component (Sony KX-27PS1, which consists of a monitor device and a tuner device; the monitor device contains one of the aforementioned amplifiers) from the rest of the hi-fi equipment consisting of the second amplifier, the PC sound system and two other audio devices. Of course, I am going to make sure I maintain the correct polarity for both amplifiers. However, is it absolutely OK to interconnect the speaker outputs of the amplifiers galvanically?"

Thanks for the question, Jan!

First of all, in you should not interconnect the speaker outputs of amplifiers - period! You will almost certainly damage components by doing so. The speaker outputs of power amplifiers should only be connected to speakers!

There is only one situation we can think of in which it is acceptable to connect multiple amplifiers to a single speaker cabinet: When using a crossover to split a signal into different frequency ranges, feeding those frequency ranges to separate amplifiers, then the outputs of those amplifiers to the individual drivers in the speaker cabinet. The speaker must be designed to function in this way; in general, hi-fi speakers are not so designed. (Referred to as bi-amping, tri-amping, etc.)

Bi-wiring is something else entirely: Connecting one amplifier to a single speaker using two speaker cables. There are those who hold that doing this improves the audio quality. Some hi-fi speakers and some studio monitors are designed for bi-wiring.

We can't see where bi-wiring can help alleviate a ground loop problem. Are you positive that what you are hearing is a ground loop, and not induced noise from the monitor or other gear? (Not that bi-wiring would solve an induced noise problem either...)