Mumbo-Jumbo and power supply caps

[Wavebourn]

Speaking of mumbo-jumbo...

Yesterday I found unexplainable fact. I was working on a tweaking of a power amp prototype. The amp was perfectly tuned and aligned. Better selection of tubes, better regimes, better distortion profile... But the sound was less real!
I could not understand why. FFT plots, especially in dynamics, shows it is better. But it sounded worse!

Previous versions had 800V B+ shunted by 2 uF film cap. The main capacitor is 235 uF electrolytic (a couple of 470 uF/450V low ESR electrolytics), so I decided to save on mumbo-jumbo this time. All other stages, and screen grids of output tubes, are powered from voltage stabilizers...

I soldered a couple of 4 uF/400V film capacitors in parallel with each 470 uF electrolytic, and the miracle happened!
Sound become alive. Binaural imaging become more precise. I tried to find any difference by measuring, but could not find it!

Since it affected imaging, and even oscilloscope on 800V B+ shows no differences, and no differences on FFT plots on different power levels, I believe it is related to modulated phase shifts that are invisible with such tests. But can you imagine how small they should be in order to be audible?!
According to experiments, monaural perceptibility is more than 10 milliseconds, while binaural perceptibility is 10 micro(!)seconds.
I can't find better explanation, because amplitude-related measurements show nothing

[Zen Mod]

next time - arrange transport to other measuring lab - with this gentleman :

YouTube - Blackadder: Back & Forth - PART 1

[SGregory]

Wavebourn,
Although my amp is nothing compared to yours, I had a similar experience with the OPUS. In that I used those huge eletrolytics, bypassed by 100uF motor run, with a 1uF bypass at the point of consumption. Take the motor run and bypass out and the amp is dead/dull sounding. On the other hand without the electrolytics, I find it to be analytical and thin sounding.

[Wavebourn]

One more finding: no difference on mono.
Also, no difference, is a shunting cap 2 or 10 uF. Effect the same.
Conclusion: phase intermodulation caused by some non-linearities in electrolytics!

[Miles Prower]

Details, schemos please. Was this a SE or PP design?

[kevinkr]

Hi Wavebourn,
I have had similar experiences and like you was unable to relate it to anything that I could measure at the time. Glad to hear that I am not the only one to hear this, still would be great if there was some plausible explanation. The effect was not subtle in my recollection - these days I usually use film caps in my supplies, and like you I regulate almost everything.

[Wavebourn]

Originally Posted by Miles Prower

Details, schemos please. Was this a SE or PP design?

It is Pyramid-VII, schemos were here several times. And it does not matter, actually.

800V powers anodes of GU-50 tubes only, with 10K P-P transformers. Both channels from the same PS. The rest, including screen grids, is powered from voltage stabilizers (+400V for phase splitter, and +270V for screen grids).

[mjf]

.......i know that phenomenon. i always thought it depends on/changes the output impedance of the psu.
usually it sounds more bright with paralelled plastic caps.........(with all types of amps?).

[Wavebourn]

Originally Posted by mjf

.......i know that phenomenon. i always thought it depends on/changes the output impedance of the psu.
usually it sounds more bright with paralelled plastic caps.........(with all types of amps?).

Sure, but if it changes impedance value only it would be at least measurable!

I still believe, it is result of phase inter-modulations caused by electrolytics that can't be measured in static amplitude-related tests.

[Zen Mod]

lets say that we can explain that with intrinsic inductance of electrolytic cap , but - we can't measure that precise enough .

explanation is simple , but proof is completely other thing - at least with measuring equipment I have

that's why I'm putting 50-70uF motor-runs even in solid state A class amps PSU

if memory serves me well - newest example of motor-run caps usage (as final bypass ? ) in commercial amp is EL34 PP , made by Audiomat ;

edit - I just found image on net ( this one not exactly EL34 , which I saw in flesh ) :