hello,
I assemble a clone and have two capacitors on the board for each +50V -50V legs.
On original, 0,1uF 100V polyester film was used. I saw many "mods", with some using aluminium, others using polypropylene...
As it's follow 10000uF, I not sure that it's for energy storage. I don't see dynamics factors usualy associated with polypro, and such small value may have a real utility ?
If filtering was in mind, an up to date 47uF MLCC can be more usefull...
what's your feeling ?
I assemble a clone and have two capacitors on the board for each +50V -50V legs.
On original, 0,1uF 100V polyester film was used. I saw many "mods", with some using aluminium, others using polypropylene...
As it's follow 10000uF, I not sure that it's for energy storage. I don't see dynamics factors usualy associated with polypro, and such small value may have a real utility ?
If filtering was in mind, an up to date 47uF MLCC can be more usefull...
what's your feeling ?
MLCC are good for all HF decoupling because of its extremely low ESR and ESL. But I sincerely doubt if their effect are noticeable in audio stuff.
But it can take RF noise off the rail and prevent it disturbing the input section. I think that's the logic, although its much better to decouple at the point of use for RF.
Still soldering the caps with 1cm long the terminals in the PCB, plus, say, 2cm of PCB track length, gives me 3cm. Now, this 3cm corresponds to a quarter line (The worst case resonant line) to a frequency of 25GHz. Clearly, including the best of the best AF amplifier is largely insensitive to those frequencies.
0,1uF value seems so low to have influence on AF... RF filtering as Mark said seems credible, or to prevent some oscillation ?
Last edited:
Ceramic disc 0.1µF @ 100V standard through hole will do the job. You can use better expensive parts, but sincerely doubt you can listen any difference.
One has to count twice the lengths. Current goes in loops. Those 3cm are 6cm in reality.
We are not in vacuum, but on a pcb (most time).
That is, the electrical length is not 3nsec/m but 5nsec/m.
Let's round up those 6cm to 5cm. That is 1/20 of a m, 0.05* 5nsec = 250psec.
That is 4GHz. Quarter length equivalent is 1GHz, rather then 25GHz...
Then the full calculus is invalid because nobody cares about wavelengths, but the resonances between the stray inductances and the bypass capacitors. In that case we had fallen back to a few MHz..
And if a 50- 100MHz opamp sees resonance peaks in it's PS lines at some MHz.. It will happily respond there.
Oscillating circuits do not sound well.
Ciao, George
Ja ja ja, don't joke, George. The guy is speaking about a bipolar simple output. No way that the amp be sensitive above a MHz. Perhaps I made wrong calculae, but it is still valid.
Ah ok it's about a power amp..
I was thinking of opamp bypass structures.
(Ad797 - GBwP=110MHz;
OPA1612: 40MHz;
OPA637: 80MHz;
OPA627: 16MHz,
ADA4627: ~80MHz etc..)
Ciao, George
I was thinking of opamp bypass structures.
(Ad797 - GBwP=110MHz;
OPA1612: 40MHz;
OPA637: 80MHz;
OPA627: 16MHz,
ADA4627: ~80MHz etc..)
Ciao, George
Ok. A note: that is typically a place for not less than 100uF bypass caps - so as to 'absorb', slow down the ringing between that cap and the power lines entering the board (their stray inductance, which can be easily into the dozen uH).
Maybe a, composit structure of an Alu electrolythic // film cap some hundred nF...
Maybe a, composit structure of an Alu electrolythic // film cap some hundred nF...