alternate feedback scheme?

[unclejed613]

has anybody ever thought about running sense lines to the speakers, much the same way is done with open frame power supplies? the sense line would be the amplifier feedback, and actually include the speaker wiring and it's losses in the feedback loop. of course, there would have to be some way to default the feedback in case one forgets to connect the sense line, and end up with open loop gain. has this been tried before, or is it more trouble than it's worth?

[teemuk]

I remember reading about something similar to this. The greatest reason why it caught no success was - if I remember right - the need for a "special" speaker cable (one with more than two conductors).

[Christer]

I have also had that idea, as I am sure many, many others have. I think the main problem is that the open loop frequency response of the amplifier now depends on the speaker cable. That could be very dangerous unless using a speaker cable with known characteristics and making sure the amp is stable with that cable. I also think, that almost regardless of what speaker cable we use, we should only take low frequency feedback from the speaker terminals, and have a high-frequency bypass from the amp output.

[teemuk]

Originally posted by teemuk
I remember reading about something similar to this.

Dug up some books... Ben Duncan's "High Performance Audio Power Amplifiers" pages 159-160 describe the principles of "four-wire" sensing: Basically the sensing and ground reference are extended to speaker terminals and speaker wiring is part of the NFB loop. This required (as name implies) a 4-wire speaker cable, where the feedback wires had to be shielded, twisted etc. since they ran alongside the speaker cables and such - usually for several meters. The alternative wiring method also caused lots of wiring errors. The text also tells that this idea was used at least by Deltec and UREI and it (naturally) seemed to work best in self-powered speakers.

[Tim__x]

You won't gain anything meaningful in my opinion. You'll improve output impedance, but not by an significant fraction of loudspeaker resistance. The stability issue can be overcome by making sure the ouput side of Cf goes directly to the output of the amplifier.

You'll find a lot more information about this idea with regard to power supplies if you search for "Kelvin Sensing".

[djk]

Was done over 30 years ago by Crown, Kenwood, Sansui, and others.

None of those do it now, draw your own conclusion.

[unclejed613]

i figured as much...... and using it for self-powered speakers is kind of redundant, since the speaker wires are usually very short anyway..... with power supplies, the required bandwidth only needs to be a few hundred hertz, since the regulator only has to deal with DC and the first few harmonics of the power line frequency...... sounds like it's more trouble than it's worth.....

[phase_accurate]

has anybody ever thought about running sense lines to the speakers, much the same way is done with open frame power supplies?

I know at least two companys who do (or at least did) it: Spectron (class-d HiFi) and Stage-Accompany (Dutch PA specialist).

Regards

Charles

[Graham Maynard]

The power amp should be in monobloc form beside the LS to cut out all of the LS cable induced problems anyway.

Cheers ........ Graham.