What is a DC power amp

[Bridges]

Do DC {direct Current} class AB power amps sound better then amps with coupling capacitors? In the late 70's they made a big deal out of these amps, what about today?

 

[mhardy6647]

Direct coupled. No low frequency roll off or phase shift without a coupling cap. Also, though, no speaker protection against direct current (without other provisions). A mixed blessing

 

[DVDdoug]

They don't necessarily sound better or worse. There are lots of ways to make a good amp. But the manufacturer will usually tout all of their design choices as a feature or benefit.

Most "modern" amps are DC coupled at the output. It saves the cost (and space) of output capacitors, at the expense of requiring dual (positive and negative) power supplies.

The "balanced" power supplies also make it easer to avoid/prevent pops when the amp is powered-on or off. I have an older amp with a single supply and output capacitors, and it turns-on quietly but when I turn it off I get a "thump" and I can see the woofers move. (It's not "terrible".)

But most amps do have a (small, inexpensive) input capacitor because you don't want any DC voltage on the input amplified and sent to the speaker.

 

[fpitas]

no speaker protection against direct current

True. The better amps monitor the DC output and disconnect the speaker during a fault.

 

[fpitas]

In the late 70's they made a big deal out of these amps

As DVDdoug implied, nobody (except hobbyists) make amps with the huge output cap anymore. Like quasi-complimentary, it has gone into the dust bin of time.

 

[Michael Fidler]

It's quite possible (with judicious splitting of negative feedback) to take feedback from after the coupling capacitor and limit the distortion that it causes (electrolytic capacitors can produce low-frequency distortion).

An advantage of the output capacitor is that you get DC offset protection for free as it naturally blocks DC. It also gives you the opportunity to use a single supply rail without bridging.

My view today is that it's better to do without it as it's a large, cumbersome, and potentially unreliable part.

 

[fpitas]

it's better to do without it as it's a large, cumbersome, and potentially unreliable part

Well there you go.

 

[Michael Fidler]

Well there you go.

Also, counter intuitively, the energy storage in the capacitor causes increased output stage dissipation as the turnover frequency of the capacitor and speaker impedance is reached...

 

[fpitas]

Also, counter intuitively, the energy storage in the capacitor causes increased output stage dissipation as the turnover frequency of the capacitor and speaker impedance is reached...

Good point. I remember "tuning" cheap sub drivers that way.

 

[Mnyb]

There are also things like a DC servo that can be used instead .

Many old designs are from a time when every discrete part had a substantial cost , nowadays a designer does not have to be that frugal.

 

[fpitas]

There are also things like a DC servo that can be used instead .

Many old designs are from a time when every discrete part had a substantial cost , nowadays a designer does not have to be that frugal.

Also, having seen it unfold (yeah, I'm old), everybody started with tube topologies, and slowly unique topologies were invented for SS that were much more appropriate.

 

[Michael Fidler]

There are also things like a DC servo that can be used instead .

Many old designs are from a time when every discrete part had a substantial cost , nowadays a designer does not have to be that frugal.

Exactly! As an example we've shifted away from using electrolytic capacitors in 'bootstraps' in power amplifier design to discrete transistor constant current sources in the voltage amplifier stage that precedes the output stage...

IME a DC servo isn't necessary as in a proper power amplifier design 50mV or less of DC offset is quite easily achieved. The main source of trouble once all other factors are taken care of is mismatch in the base currents of LTP inputs. Including the servo gives you just another thing to go wrong if you ask me...

DC servos can also turn into LF oscillators if the input capacitor to the amplifier at the LTP is an electrolytic. If this part dries out a bit the time constants in the circuit change and the thing can easily go from damped, to underdamped, to LF generator. Not nice!

 

[Michael Fidler]

Also, having seen it unfold (yeah, I'm old), everybody started with tube topologies, and slowly unique topologies were invented for SS that were much more appropriate.

Oh and another major point is that output caps are going to generate large amounts of even-order distortion if they're electrolytics. If the AC voltage across the cap exceeds 100mV (requiring 250,000uF of output capacitance for an 8 ohm load in a 50W amp) then you're going to get 10ppm distortion at 20Hz. Most output caps are around 2200uF so you can see where this is going...

Luckily frequency compensation isn't affecting the open-loop gain of the amplifier at this frequency so much, but even then it's an ugly thing and best avoided.

 

[fpitas]

I'd say no one misses the output cap, or quasi-complimentary for that matter, except a small group of determined hobbyists.

 

[Michael Fidler]

I'd say no one misses the output cap, or quasi-complimentary for that matter, except a small group of determined hobbyists.

Ah but how does is SOUND... I have it on good authority that the Qcomp output has more midrange insight and bass authority when using an authentic Plessey 470uF axial capacitor and original RCA 2N3055s

 

[fpitas]

Ah but how does is SOUND... I have it on good authority that the Qcomp output has more midrange insight and bass authority when using an authentic Plessey 470uF axial capacitor and original RCA 2N3055s

 

[AnalogSteph]

I'd say no one misses the output cap, or quasi-complimentary for that matter, except a small group of determined hobbyists.

I would still consider quasicomp if sheer ruggedness was a major concern. Besides, with a Baxandall diode in the driver stage, distortion performance can actually be improved quite a bit.

 

[fpitas]

I would still consider quasicomp if sheer ruggedness was a major concern. Besides, with a Baxandall diode in the driver stage, distortion performance can actually be improved quite a bit.

I think I still prefer to add output devices for ruggedness, unless fidelity is a secondary concern.

 

[Michael Fidler]

I would still consider quasicomp if sheer ruggedness was a major concern. Besides, with a Baxandall diode in the driver stage, distortion performance can actually be improved quite a bit.

Somewhat but IME it's still nowhere near as good as a CFP output stage which allows a good deal of paralleling of the output devices and much better thermal stability... It also lets you get the output closer to the rails which is never a bad thing.

 

[mhardy6647]

Well there you go.

Conversely, real audiophiles adore large, cumbersome, and potentially unreliable part.