Moving on out
Which is what I'd do, but it then brings the problem of a suitably low-impedance feed to the device being powered. I wouldn't do it in a power amp, without full regulation, locally.
W.r.t. caps, many of the above discussion refer to hum and other obvious noise from ripple in the PSU, but in my experience it's the rythmic / dynamic integrity (i.e. the music) that suffers most from poor choice.
I've no experience of using inefficient classA amps with high / relatively constant quiescent drains, but in the case of low current supplies, and supplies for class AB amps, excess capacitance kills the sound. If you think it doesn't there's something wrong elsewhere making the difference inaudible.
In these cases the optimal capacitance seems related to the transformer electrical characteristics, but I accept it may well be different for class A operation at high current.
Andy.
Which is what I'd do, but it then brings the problem of a suitably low-impedance feed to the device being powered. I wouldn't do it in a power amp, without full regulation, locally.
W.r.t. caps, many of the above discussion refer to hum and other obvious noise from ripple in the PSU, but in my experience it's the rythmic / dynamic integrity (i.e. the music) that suffers most from poor choice.
I've no experience of using inefficient classA amps with high / relatively constant quiescent drains, but in the case of low current supplies, and supplies for class AB amps, excess capacitance kills the sound. If you think it doesn't there's something wrong elsewhere making the difference inaudible.
In these cases the optimal capacitance seems related to the transformer electrical characteristics, but I accept it may well be different for class A operation at high current.
Andy.
Power factor
An additional note - has anyone ever measured the harmonic content of their mains supplies?
The power factor issues in supplies with excess capacitance have very real effects upon the quality of the mains that feeds your system.
If you think it looks at all like 50 (or 60Hz) sine wave, then think again. Those rectifiers switching on for short periods, drawing tens of amps produce significant odd-order harmonic noise on the mains supply, all the way up to 5kHz or so, depending upon local mains impedance.
I've got some FFT plots if anyone is interested, taken at different times of day - anyone else noticed how much better their system sounds late at night?
Andy.
An additional note - has anyone ever measured the harmonic content of their mains supplies?
The power factor issues in supplies with excess capacitance have very real effects upon the quality of the mains that feeds your system.
If you think it looks at all like 50 (or 60Hz) sine wave, then think again. Those rectifiers switching on for short periods, drawing tens of amps produce significant odd-order harmonic noise on the mains supply, all the way up to 5kHz or so, depending upon local mains impedance.
I've got some FFT plots if anyone is interested, taken at different times of day - anyone else noticed how much better their system sounds late at night?
Andy.
Andy,
Interesting. Maybe that is why some systems sound better with the power amp fed from a separate outlet/group, not because the amp gets better, but because the junk from the amp doesn't get that easy to the CD/preamp etc.
I have already years ago decided not to use toroids for low-level equipment, because the wide bandwidth provides an easy entry for mains-borne junk.
Jan Didden
Interesting. Maybe that is why some systems sound better with the power amp fed from a separate outlet/group, not because the amp gets better, but because the junk from the amp doesn't get that easy to the CD/preamp etc.
I have already years ago decided not to use toroids for low-level equipment, because the wide bandwidth provides an easy entry for mains-borne junk.
Jan Didden
Toroids
Jan,
It's certainly true that toroids do have an inherently wide bandwidth due to inter-winding capacitance. Electrostatic shields are not a particularly effective way of reducing this.
I've not yet tried alternatives though, this is a line of investigation I will be pursuing later. I do believe though that mains impedance is absolutely crucial to sound however - lower it and things generally sound better.
Even using super-reg's, with massive line rejections, the differences in the raw DC supply, or the mains supply, in terms of impedance are clearly audible, and whilst lowering the bandwidth will help reduce mains-borne noise reaching equipment, it conversely raises impedance seen by the raw supply to an item, and hence could allow higher levels of harmonic content owing the the higher impedance presented by the secondary circuit.
The jury is out on this one for me, at present. Meanwhile all my PSU's have ridiculously large transformers, for the few mA I draw from them
This will be an interesting line of investigation when I get around to it.
Andy.
Jan,
It's certainly true that toroids do have an inherently wide bandwidth due to inter-winding capacitance. Electrostatic shields are not a particularly effective way of reducing this.
I've not yet tried alternatives though, this is a line of investigation I will be pursuing later. I do believe though that mains impedance is absolutely crucial to sound however - lower it and things generally sound better.
Even using super-reg's, with massive line rejections, the differences in the raw DC supply, or the mains supply, in terms of impedance are clearly audible, and whilst lowering the bandwidth will help reduce mains-borne noise reaching equipment, it conversely raises impedance seen by the raw supply to an item, and hence could allow higher levels of harmonic content owing the the higher impedance presented by the secondary circuit.
The jury is out on this one for me, at present. Meanwhile all my PSU's have ridiculously large transformers, for the few mA I draw from them
This will be an interesting line of investigation when I get around to it.
Andy.
I agreed with Andy for the following points;
)_Amp sound better at night, maybe also the fact that background noise surrounding the house is lower at night as well as what Andy stated.
)_My Class A amps drawing 4A Iq, sound changes with different PSU caps setup
with 30,000uf PSU caps open with dynamic mid-high, softer lows
with 60,000uf PSU caps middle of the road
with 90,000uf PSU caps highs lost air and dynamic suffers but spooky bass that make you think the woofer is going to jump out at you and land on your feet.
After listening to 30,000 and 60,000 settings for a while, I chosen 30,000uf because more CDs sound better with it than with 60,000uf caps. I do wish that I can use a switch so that for certain CDs I can use 60,000uf.
Andy, how is your optimal caps value formula coming along?
)_Amp sound better at night, maybe also the fact that background noise surrounding the house is lower at night as well as what Andy stated.
)_My Class A amps drawing 4A Iq, sound changes with different PSU caps setup
with 30,000uf PSU caps open with dynamic mid-high, softer lows
with 60,000uf PSU caps middle of the road
with 90,000uf PSU caps highs lost air and dynamic suffers but spooky bass that make you think the woofer is going to jump out at you and land on your feet.
After listening to 30,000 and 60,000 settings for a while, I chosen 30,000uf because more CDs sound better with it than with 60,000uf caps. I do wish that I can use a switch so that for certain CDs I can use 60,000uf.
Andy, how is your optimal caps value formula coming along?
Power supply FAQ ?
Hi!
I think it's time someone made a faq on this subject,
perhaps in the wiki section.
And set up a "general-rule" list on amount of filtering-caps
for a 50 / 100 / 250 / 500 W class-AB and class A amp.
And for line-stages/CD/...
And this will also have to relate to power of trafo...
Just an idea?
Arne K
Hi!
I think it's time someone made a faq on this subject,
perhaps in the wiki section.
And set up a "general-rule" list on amount of filtering-caps
for a 50 / 100 / 250 / 500 W class-AB and class A amp.
And for line-stages/CD/...
And this will also have to relate to power of trafo...
Just an idea?
Arne K
It's an interesting thing you experienced, because there
really isn't a good reason why additional capacitance would
screw up the high end as such. However, I have often
encountered situations where you change the bass, often
improving it, and find yourself with a different perception
of the top end. Or vice versa.
This occurs most often when screwing around with crossovers.
Occasionally you'll do something like change the phase of a
tweeter crossed up a 5K, and it will totally clean up the bottom
end.
Who knows what's the deal with this?
Not even the Shadow knows.......
really isn't a good reason why additional capacitance would
screw up the high end as such. However, I have often
encountered situations where you change the bass, often
improving it, and find yourself with a different perception
of the top end. Or vice versa.
This occurs most often when screwing around with crossovers.
Occasionally you'll do something like change the phase of a
tweeter crossed up a 5K, and it will totally clean up the bottom
end.
Who knows what's the deal with this?
Not even the Shadow knows.......
Nelson Pass said:It's an interesting thing you experienced, because there
really isn't a good reason why additional capacitance would
screw up the high end as such.
[snip]
Nelson,
Increasing the cap will increase drastically the higher-order harmonics on the supply line. Depending on the PSRR of the amp at higher freqs, this can be a perfectly logically reason for screwing up the hi end.
Jan Didden
PMA said:How big do the capacitors have to be?
It also strongly depends on the PSRR of the amp. You might need 100 000 uF, or 10 000 uF will do the same work with another amp
The bigger the caps, the more hi freq junk on the supply lines. On the other hand, bigger caps do not help with low PSRR, because the PSRR at the low end usually is OK in a reasonably well designed amp. Bigger caps *can* screw up the top end (see above), so it is doubtfull that the size of the caps can compensate for mediocre PSRR.
Jan Didden
I don't think that's it.
In real life, all harmonics on the supply rails go down
with more capacitance.
If more noise is picked up from the AC lines, that is not
PSRR, but a problem with magnetic pickup. This is
easily, if not cheaply, fixed by proper layout and grounding.
Besides this, it is easy enough to construct a pi or similar
filter which allows tons of capacitance on the circuit rails
without the line (power factor) noise going up, and you
get the same effect.
In real life, all harmonics on the supply rails go down
with more capacitance.
If more noise is picked up from the AC lines, that is not
PSRR, but a problem with magnetic pickup. This is
easily, if not cheaply, fixed by proper layout and grounding.
Besides this, it is easy enough to construct a pi or similar
filter which allows tons of capacitance on the circuit rails
without the line (power factor) noise going up, and you
get the same effect.
How big etc
Well yes, you are right, there are many factors involved.
I was specifically addressing the question of 20kuF or 100kuF, *all other things being equal*. In that case, the supply harmonics definitely go up with higher cap because of the narrowing of the conduction cycle leading in turn to very narrow, high current peaks that lead to high level high harmonics to the line. And neither the electrolytic (unless bypassed wth a good foil cap), nor the amp PSRR are very effective against the higher harmonics.
The pi-filters you alluded to will indeed fix this, because they will again open up the conduction cycle, flatten the current pulse and decreasing higher harmonic content.
But who am I telling this?
Jan Didden
Well yes, you are right, there are many factors involved.
I was specifically addressing the question of 20kuF or 100kuF, *all other things being equal*. In that case, the supply harmonics definitely go up with higher cap because of the narrowing of the conduction cycle leading in turn to very narrow, high current peaks that lead to high level high harmonics to the line. And neither the electrolytic (unless bypassed wth a good foil cap), nor the amp PSRR are very effective against the higher harmonics.
The pi-filters you alluded to will indeed fix this, because they will again open up the conduction cycle, flatten the current pulse and decreasing higher harmonic content.
But who am I telling this?
Jan Didden
How many I can really use? 8 or 4?
Before I place order to Zalytron for
14 Gauge Super Silicon Grain Orientated Steel Core Q Coil with 2.00mh, 0.101 DCR @ $7.52
I am planning to use them between C3 and C5; and C4 and C6 from the diagram attached (From Geoff'site) of my power supply.
Can I use 4 of them for 1 PSU for 1 mono block amp, or is this a waste for the inductors in the ground rails too?
Thanks,
Chris
Before I place order to Zalytron for
14 Gauge Super Silicon Grain Orientated Steel Core Q Coil with 2.00mh, 0.101 DCR @ $7.52
I am planning to use them between C3 and C5; and C4 and C6 from the diagram attached (From Geoff'site) of my power supply.
Can I use 4 of them for 1 PSU for 1 mono block amp, or is this a waste for the inductors in the ground rails too?
Thanks,
Chris
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.