Capacitors in output?

[Lisandro_P]

So, what're the real advantages/disadventages of putting a capacitor at the output of an amp? It might be it's late but, besides reducing bandwidth and killing DC offset, i can't seem to figure it..

[Geoff]

It is not a question of advantages/disadvantages but one of necessity. With a single power supply rail, the output dc voltage will be at approximately half the supply rail voltage and a capacitor is essential if the speaker is to survive. With dual power supply rails, the output dc voltage is zero and a capacitor must not be used as there will be no dc polarising voltage across it.

[Lisandro_P]

OK, but even with dual psu amps, isn't it a good idea to remove the dc offset from the output (even if it's a few milivolts)... i mean, i'd like to keep DC as away from the speakers as possible, so i wonder, why isn't a common practice to place caps in the output? It would help to contain dc imbalance if, say, an output transistor blows up too...

[dorkus]

Originally posted by Lisandro_P
OK, but even with dual psu amps, isn't it a good idea to remove the dc offset from the output (even if it's a few milivolts)... i mean, i'd like to keep DC as away from the speakers as possible, so i wonder, why isn't a common practice to place caps in the output? It would help to contain dc imbalance if, say, an output transistor blows up too...

putting a cap between your amp and speakers is a really bad idea. the theoretically-ideal capacitor, if large enough to keep the -3dB point below the audible range with the given load, would have no impact on sound, true. but real-world capacitors are not ideal, they are not lossless. in addition to having parasitics (ESR, inductance), they exhibit non-linear behavior, and thus don't sound "perfect." even the very best capacitors will veil the sound compared to a straight-wire bypass. so why use one when it's not needed? any properly-designed dual-supply amp (single-ended is a different matter) will either have a DC servo to eliminate offset or some DC-blocking at the input or feedback loop; this is more than sufficient to ensure DC protection for your speakers for all but the most paranoid. putting a capacitor at the output is wholly unnecessary and would degrade the sound... besides, you'd need a really huge one (at least 2000uF or so for a 4 ohm load) to ensure that no low-frequency cutoff would occur. such a large capacitor would have to be electrolytic, and in this application they probably would not sound so great.

[Edited by dorkus on 06-14-2001 at 10:46 AM]

[Lisandro_P]

I see.. thanks man
And, while we're on the subject, what's the maximum DC allowable through a speaker and what's its real effect on it?

[dorkus]

Originally posted by Lisandro_P
I see.. thanks man
And, while we're on the subject, what's the maximum DC allowable through a speaker and what's its real effect on it?

hmm hard to say, like you say, the less dc the better, but all dc really does is make your woofer cones pop out, in the case of a dynamic speaker. of course, woofers popping out is not a very good thing - your tweeters and higher-frequency drivers will be fine because they have high-pass filters which naturally block DC, but your woofers will definitely be affected. i would say that as long as the total DC power is under 5 milliwatts or so you'll be more than safe. 5mW at 8 ohm translates to 200mV, which is pretty liberal. if your DC offset at the speaker terminals is under 200mV, you have nothing to worry about at all. but i'm sorta pulling these numbers out of my a**, i'd say they're definitely on the safe side. anybody care to comment?

[dorkus]

hmm on second thought maybe 200mV is a bit much... maybe 100mV?

[Lisandro_P]

Originally posted by dorkus

hmm on second thought maybe 200mV is a bit much... maybe 100mV?

100mV seems to be a "standart" for maximum DC offset, but, dunno, 0,1V? isn't it a bit too much?

[dorkus]

well you know, 100mV (.1V) translates to only 1.25mW into 8 ohms, or 2.5mW into 4 ohms. not very much power, if ou think about it...

[Lisandro_P]

Yah, but... dunno, i'd just feel much mor safe with, say, 1-10mV... Ahhh well, time to fix some schematics!