What rail voltages?

What transformer VA rating (per rail)?

What values of reservoir capacitance per rail?

What cap values are used for Cs?

If everything else is OK, you would also want Cs (local decoupling and bypass caps) to include 0.1 to 1 uF high-frequency bypass cap directly at each power pin, and also another 470 uF to 2200 uF (or more) for local decoupling, to be able to supply transient current demands with low-inductance (i.e. short length!) connections (preferably using three or more smaller paralleled electrolytic caps with values that sum to the desired total capacitance per rail, mounted as closely as possible the chip's power pins).

For your power supply, you should have a

**minimum** of roughly 14,000 uF of reservoir capacitance per rail. That is based on calculating the capacitance required to be able to supply the current to produce a 5-Amp-peak sine wave at 20 Hz without causing the rail to dip by more than 1 Volt, assuming that sufficient-current recharging pulses are being provided every 8.3 ms (60 Hz mains; add 15% or so for 50 Hz mains). Using 3X 4700 uF would work nicely, for that.

But I would probably want roughly at least 3X 14000 uF per rail, which should be able to support 5 Amps DC current into 8 Ohms, continuously, for rails rated at 100 Watts RMS (40 Volts peak), without letting the rails dip by more than 1 Volt, assuming 120 Hz recharging pulses with sufficient current are available. That should cover almost any "worst-case" signal, at max output power.

For the equations, see the post at

http://www.diyaudio.com/forums/chip-amps/152471-finalizing-tda2050-lm3886-2.html#post3363511 , which were explained and derived at

http://www.diyaudio.com/forums/power-supplies/216409-power-supply-resevoir-size-169.html#post3320547 .

Your transformer VA (Volt-Amps) rating should be AT LEAST 1.5X to 2X the rated maximum RMS output power, just to account for heatsink losses (i.e. the amp is not 100% efficient). But I would use at least 3X the max rated RMS Watts for the VA rating, unless I had significant headroom in the rail voltage.

Cheers,

Tom