are output decoupling caps really necessary in symmetrically powered amps?

if the amp is using symmetric power supply, the output right at the opamp is idling at 0 V. Or, maybe, a couple tens of mV. So i don't see the reason (i suspect there might be).

i tried reading up on it, but i haven't got a satisfactory explanation.

to be honest, i didn't search the forum, mostly because i don't know how to create a query for such a specific question
If the op-amp stage is unity gain you don't strictly need coupling caps on the output. However, I always prefer to include them regardless. Also I would not leave coupling caps both on INPUT and OUTPUT of a volume control. POTs are not very forgiving of DC flowing through them, and scratchyness results and is worse without.

However input switching they are recommeded along with bleeder resistors each side to prevent switching thumps with different devices/DC offsets connected.
Also the feedback capacitor (normally to ground/0V in a noninverting configuration) should not be missed out as if it is, the input offset is then multiplied by the gain of the amp. on a power amp this can lead to significant constant DC offset into the speaker especially when using higher gains (x30 or more)
Last edited:
looking here,


i see that national's schematic does not use output decoupling caps. the pcb i got, from hifidiy, has some caps on the output, but i am not sure what they are for.

Also, not exactly related: i am thinking of inserting relays, one for each speaker, that connects the speakers after the amp has been powered ( 1 s delay, or something like that) . Is it enough? I think it is, i'm just looking for the OK from you guys... :)


Paid Member
2007-09-15 8:14 am
was that a question.

In a way :) Decoupling (the thread tiltle) refers to caps on and across supply rails. Coupling refers to caps that connect different stages together. They carry the signal between stages. The zobel network is the series C and R that is often fitted at a power amp output to maintain stability (and that seemed the "best fit" to how you phrased the original question).

In the schematic there appears to be an omission - there should be 20K resistor from both op-amp +inputs to ground, these along with the presence of Ci1 and Ci2 will keep output offset to minimum

There seem to be many omissions. It looks more like a basic diagram rather than a fully worked example.
Look at the TI datasheet....

P. 22 "Reference PCB schematic" shows the circuit I'd expect to use of such an IC. I'd still recommend that Rin1 and 2 (shown as 15K) should actually be = Rf (20K)

In response to your earlier question about output relays, if its just switch on and switch off thumps are you're worried about; they recommend a soft start circuit P.15.
If that's not good enough or you think you are going to use and abuse the gainclone probably better to get a seperate speaker protection PCB with built-in delay.
Last edited:
You don't need any capacitor between output of power op-amp and speaker in split rail supplies or bridge mode. This is because at zero signal in both these cases there should be insignificant DC volts (difference) across the speaker.

When a single rail supply is used and the speaker load is returned to ground, then half the DC supply rail exists at the amplifier output. This occurs even at zero signal to allow signal swing both upwards and downwards. In this instance, the resultant DC must be blocked by a large electrolytic capacitor of sufficient voltage, capacitance and current rating.

When a single supply is used and the speaker load is returned to another amplifier output, ("bridged" amplifiers) then the same "half the DC rail" exists at both the amplifier outputs but this means there should be zero DC (difference) across the speaker terminals. In this instance, there should be no DC left to block, and output capacitors are normally not necessary, same as "split rail" above.
Last edited:


Paid Member
2007-09-15 8:14 am
i am interested in the capacitors that are in series, between opamp output and speaker :)

Just to add a bit more to blu_glo's reply.

A cap between an "opamp" or chip amp output and speaker needs to be a "bi-polar" cap meaning it has no polarity. The reason is that an opamp output can be either positive or negative with respect to the zero or ground. The cap would need to be big in value, around 2000uf for an 8 ohm speaker and a roll offpoint of around 10Hz. That means two 4000uf caps wired back to back to make a bi-polar 2000uf one.

Caps used like this are a great safety feature for the speakers protecting them from any DC faults.