loudspeaker design for current source power amplifier


2002-02-04 7:18 pm
I'm looking for comments on developing a loudspeaker that would be driven by a current source power amplifier.

That is to say, in a "traditional" amplifier, the voltage output is proportional to the input signal, regardless of the load impedance.

I'm interested in the implications of driving a speaker with an amplifier where the _current_ output is proportional to the input signal, regardless of the load impedance.

Would the only issue be to create a crossover that balanced out the frequency response? What other issues are likely to arise? Is there something fundamental that makes it difficult or impossible to drive a speaker with a current source?

Any comments or references would be appreciated.

Interesting concept. I don't know much about loudspeaker design and I have no idea who came up with the ubiquitous 8 ohms. Transistors control currents. Normal amps control their output V by various feedback mechanisms. If instead, current proportional to input signal was fed to the speaker then no feedback would be needed at all and circuits might be a touch simpler. Would this sound better? I don't know. It might make the speaker design more challenging. If you could keep the speaker Z low then it would allow the use of low V high I transistors - which I believe tend to have higher frequency performance. Worth considering.


2002-02-04 7:18 pm
Hello Dan,

Thanks for the references. I read the articles and I have to admit I don't understand.

The author states: "With a pure current drive signal, the input determines the current that flows through a loudspeaker (or other) load. This is rarely used in practice, as it is unsuitable for driving speakers." He doesn't explain why this is so.

He then goes on to analyze a current source with internal series impedance. That part doesn't make sense to me either.

I would assume a real world current amp would be modeled by an internal parallel impedance Rparallel that ideally is infinite:

       |        |          |
     Iout   Rparallel    Rload
       |        |          |
       =        =          =

I understand when Rload increases, V increases (and Iload goes down). The increase in V might be viewed as "negative source series impedance", but it seems to be an inherent property of a current source. I don't understand how this concept is useful. I also don't understand why this would lead to instability.

It's not generally recommended and let me explain why.

A perfect current source got an infinite impedance. This impedance define the damping factor of the amplifier. At this setting speakers will start ringing caus the amp loose the control he have on the speaker. This is very audible and not interesting.

For reference i realy recommand an article on source impedance at www.sound.au.com
Allen said:
I'm looking for comments on developing a loudspeaker that would be driven by a current source power amplifier.

I had a bit of a conversation with a fellow in Washington State (he was introduced to me by Lynn Olsen) who builds PP Class A tube amps designed to drive speakers in such a manner (at least that is what i understood). Most of it was over my head, but i do rememeber him saying that the amps output impedance matched the speakers impedance.

I could dig up his phone number if you want -- last email i sent him bounced.

Current Source Amplifier

In my experience servicing recording studio active monitors (i.e. Genelec, Mackie, etc.) a traditional audio amplifier is modified by placing a less than 1 ohm resistor in series with the speaker's negative lead to ground. Next lift the grounded side of the existing feedback network voltage-divider and connect it where the newly installed (sub 1 ohm) resisitor and the speaker's negative lead are joined. In this manner the varying impedence of the speaker's voice coil is reflected in the voltage drop across the sub 1 ohm series resistor. With the feedback network sensing this voltage variation (due to "Z" changes) the former voltage source amplifier now produces a current output proportional to the changing voice coil impedence. Generally the new series resistor is approximately 1/10th the nominal value of the speaker's rated impedence with 0.47 to 0.68 ohms being typical. It is also important that this resistor be non-inductive (i.e. not wirewound) so as not to introduce any new inductive phase shifts. Ohmite makes a series of 20-watt non-inductive power resistors in a TO-220 package that sell for about $4 each from Mouser or Digi-Key Electronics.