Why not only 4700uF?

[smartx21]

This is my home amp, brazilian brand from early 80's.
Supplies are separate - two transformers, rectifiers, filters.
The question is: people nowadays suggest to use 10,000, 20,000... even 40,000uF total electrolytics value per rail.
This amp has only 4,700uF (per rail, per channel, total under 20,000uF).
Even though, the sound is quite fine, at any level, NQA.
Are my ears going to the grave? Do I need to measure distortion, clipping, to realize that it is an old fashioned design???
BTW, I would be grateful to read comments about the topology, operating class, etc.
Outputs are original Toshiba's 2SD1148/2SB863, double pair.
According to its manufacturer's history, this must be a japanese design (at least inspired on).
Regards,
Max.

[AndrewT]

It looks like the passive input filter is set to 11ms.
The lowest I would set the PSU to is 22ms.
For an 8ohm speaker that would require ~+-2800uF per channel.
4700uF is plenty.

The amplifier and the PSU are designed to give virtually no deep bass.

[Mooly]

Large value caps have huge downsides, but it seems not everyone realises this and just throws more and more uF at their designs.

1. An amp should be relatively imune to ripple on the rails. If it's audible as hum etc then the design is poor and the amp has a poor psrr.

2. Large caps take huge charging currents. Put simple the bridge rectifier has to supply all the energy required over a very short part of each cycle. The conduction angle is small, the current s huge. This can actually cause problems causing core saturation in the transformer as the magnetic field collapses momentarily. The transformer may run much hotter than it should too.

As always it's a compromise... it sounds impressive having large caps, the reality may not be quite what you intended.

[AEIOU]

Reservoir caps smooth the rectified AC. Depending upon how much ripple is acceptable would be the determining factor. You would really have to look at or measure the amount of ripple and even set up a dynamic situation, not just at idling to know for sure. More capacitance is usually a good thing, but at some point you get diminishing returns.
So try it, but if you don't notice any improvement then it probably wasn't even necessary.

[ostripper]

This amp is a level -shifted , balanced design like the APT-1 or many Harmon- Kardon designs. As far as 4700uF X2 , I had a sony 5 !!! channel HT receiver with discrete outputs that only had 6800uF X2 in the PS. This low capacitance will cause the PS to "sag" giving the output stage a current limit (SOA/safety factor). It can still sound good with sufficient amplifier PSRR ... the Sony did.

OS

[smartx21]

Well, mine sounds very good. And is a lot better than the DIY one I finished a few days ago, regarding power and punch. Maybe the double pair outputs and higher rail voltages do the difference. Also non-fake, matched transistors!