How big etc
promitheus,
I'm not very expert on transformers, but they way I understand it is that the effect is present in all cores, but because of the much higher magnetic efficiency in toroids vs. other types, the effect is much more pronounced. The inductance is the same everytime you connect, except for the brief moment where the magnetization gets in line with the line voltage, so to say. Maybe someone else can comment on this?
Jan Didden
promitheus,
I'm not very expert on transformers, but they way I understand it is that the effect is present in all cores, but because of the much higher magnetic efficiency in toroids vs. other types, the effect is much more pronounced. The inductance is the same everytime you connect, except for the brief moment where the magnetization gets in line with the line voltage, so to say. Maybe someone else can comment on this?
Jan Didden
As promised...
For those of you that would like a simple explanation of some basic power supply issues, I have posted the article I talked about last night. Please forgive the poor scanning, I did this very late last night. If a lot of members download it, I may have to move it to a different location. I am not sure my allowable bandwidth for the page's service provider. The article is ~1.7MB .pdf file. It can be found here:
Digital_supply.pdf
Happy reading,
For those of you that would like a simple explanation of some basic power supply issues, I have posted the article I talked about last night. Please forgive the poor scanning, I did this very late last night. If a lot of members download it, I may have to move it to a different location. I am not sure my allowable bandwidth for the page's service provider. The article is ~1.7MB .pdf file. It can be found here:
Digital_supply.pdf
Happy reading,
Re: As promised...
Todd,
This is usefull article, but note that it totally glosses over the issue of supply load impulse response. I think now everybody here realises that (almost) anyone can build a good (DC) supply, but getting low noise and residual clock pulses on supplies for things like DAC's is another kettle of fish.
Nevertheless, a good place to have a lot of issues together. I especially liked the section on layout, very instructive.
[Note also that fig 1.4 is somewhat misleading as to the waveform; picture 1.3 is better].
Jan Didden
stadams said:
Todd,
This is usefull article, but note that it totally glosses over the issue of supply load impulse response. I think now everybody here realises that (almost) anyone can build a good (DC) supply, but getting low noise and residual clock pulses on supplies for things like DAC's is another kettle of fish.
Nevertheless, a good place to have a lot of issues together. I especially liked the section on layout, very instructive.
[Note also that fig 1.4 is somewhat misleading as to the waveform; picture 1.3 is better].
Jan Didden
I posted the article hoping that it would be of value as a good starting point for DC power supply design.
I also found an article in the forum that was very helpful in eliminating some unwanted high frequency noise that occured due to the repetitive surge currents that you described in one of your previous posts. The falling edge of the current pulse would excite ringing in the power supply causing all kinds of low-level high frequecy "garbage" to color the DC supply. Here is the article:
Calculating Optimum Snubbers.
HH posted this article initially. Very good information.
Later,
I also found an article in the forum that was very helpful in eliminating some unwanted high frequency noise that occured due to the repetitive surge currents that you described in one of your previous posts. The falling edge of the current pulse would excite ringing in the power supply causing all kinds of low-level high frequecy "garbage" to color the DC supply. Here is the article:
Calculating Optimum Snubbers.
HH posted this article initially. Very good information.
Later,
As long as we are talking about deleterious effect of power supplies, we should not neglect that low ripple puts a greater peak load on the rectifier, by reducing the conduction angle. If you have a +20Vp-p rectified waveform and a 1V ripple, your conduction angle is quite small. Only a little squirt of current is passing through the rectifier twice per cycle. If the supply is drawing 1A, the peak current through the rectifier might be 50A. Read datasheets carefully.
If you want to reduce high frequency garbage in power supplies, don't neglect to use soft recovery types. Diodes with uncontrolled recovery characteristics will certainly exacerbate the falling-edge effect mentioned above. My favorite is IRF ultra-fast ultra-soft type in TO-220 package. Some others on this board have expressed a preference for IXYS types. Your taste may vary.
If you want to reduce high frequency garbage in power supplies, don't neglect to use soft recovery types. Diodes with uncontrolled recovery characteristics will certainly exacerbate the falling-edge effect mentioned above. My favorite is IRF ultra-fast ultra-soft type in TO-220 package. Some others on this board have expressed a preference for IXYS types. Your taste may vary.
Back to the comment about the Sloane class AB versus
a Zen, you are comparing apples to oranges. The Zen
may need 10 times the capacitance to be quiet, but it
probably is drawing 10 times the current at idle. As a
result, it has 10 times the ripple on its supply. The
measurements of both amplifiers are being taken at idle,
and if you run them both up to near maximum power,
you will see more or less the same ripple noise on the
Zen, but maybe 10 times the ripple on an AB amp.
a Zen, you are comparing apples to oranges. The Zen
may need 10 times the capacitance to be quiet, but it
probably is drawing 10 times the current at idle. As a
result, it has 10 times the ripple on its supply. The
measurements of both amplifiers are being taken at idle,
and if you run them both up to near maximum power,
you will see more or less the same ripple noise on the
Zen, but maybe 10 times the ripple on an AB amp.
How big etc
Nelson,
Your first comment is interesting, in that it means that with a high idle current, you can increase capacitance without the negative effect of very narrow and very high current pulses. Because of the higher idle, the caps discharge faster, which means earlier start of the conduction cycle. Higher capacitance for a class-A (or ZEN) thus makes sense indeed.
You are also correct in the second part, that the AB at full power will have much higher ripple. But it remains to be seen whether this, at full power, really has a negative effect on sound quality (I know, you didn't say that, just a line of thought I'm following).
Cheers, Jan Didden
Nelson,
Your first comment is interesting, in that it means that with a high idle current, you can increase capacitance without the negative effect of very narrow and very high current pulses. Because of the higher idle, the caps discharge faster, which means earlier start of the conduction cycle. Higher capacitance for a class-A (or ZEN) thus makes sense indeed.
You are also correct in the second part, that the AB at full power will have much higher ripple. But it remains to be seen whether this, at full power, really has a negative effect on sound quality (I know, you didn't say that, just a line of thought I'm following).
Cheers, Jan Didden
Nelson is right, of course, as long as we only discuss the ripple
voltage on the rails. However, it still does not say much about
the effect of the ripple on the amplifier output. That requires that
we also take the amplifiers PSRR into account, and I suspect
that a Zen cannot compete with a traditional symmetric
differential-input amplifier in this respect. So in a sense we are
still comparing apples with pears, or at least very different
types of apples. So a traditional amp may still be able to reduce
the output ripple to the same amount as a Zen, but with less
capacitance in the PSU.
Note, this does not in any way imply that the Zen should be
an inferior amplifier. An amplifier should be designed with
a PSU that is sufficient to achieve the desired sonic result, and
sonically there is nothing wrong with a larger PSU. The PSU
is only a practical question of size and money.
I take for granted that Nelson intervenes if my assumption
about the PSRR of the Zen is wrong. Perhaps he uses some
audio wizardry to improve the PSRR, which is non-obvious
from the schematics.
voltage on the rails. However, it still does not say much about
the effect of the ripple on the amplifier output. That requires that
we also take the amplifiers PSRR into account, and I suspect
that a Zen cannot compete with a traditional symmetric
differential-input amplifier in this respect. So in a sense we are
still comparing apples with pears, or at least very different
types of apples. So a traditional amp may still be able to reduce
the output ripple to the same amount as a Zen, but with less
capacitance in the PSU.
Note, this does not in any way imply that the Zen should be
an inferior amplifier. An amplifier should be designed with
a PSU that is sufficient to achieve the desired sonic result, and
sonically there is nothing wrong with a larger PSU. The PSU
is only a practical question of size and money.
I take for granted that Nelson intervenes if my assumption
about the PSRR of the Zen is wrong. Perhaps he uses some
audio wizardry to improve the PSRR, which is non-obvious
from the schematics.
That depends on the acuity of your hearing, the quality of your listening room, the amount of ambient noise in and around your listening room, the loudspeakers.
It may depend a bit on whether you crank the volume loud and thereby draw more current from the supply. (Which relates to the amp topology to which you are supplying the power.) Are the voltage rails drooping as you do this?
It may depend also on whether you are able to dispassionately listen to one iteration vs. the other having paid for all the capacitors needed. Speaking for myself I think I'd have an AWFUL hard time saying that the addition 20 to 40 mF was a waste of time and money.
All in all... it is a definite "maybe".
Erik
MikeW said:
The particular amp you are talking about will determine the effect. If we are talking about a Zen then the difference between 20mF and 40mF is audible. If you talking about other amps there may not be any audible difference.
Remember: Just because X is good does not necessarily mean that 2X is better. If hum is a problem in an amp, then adding capacitance (or a choke in some cases) can help. If you don't hear hum, then just adding capacitance for it's own sake is not going to magically transform your system with new superlative qualities.
Point is there is no rule of thumb that can be applied in all cases. Designing a power supply is a mathematical process that is based on the particular requirements of your amp. Blindly increasing the size of the caps in an amp is not certain to improve performance and could cause it to fail.
I suspect you are being influenced by stiffening caps from the high power car audio world. In a car sound system, stiffening caps are installed to prevent supply voltage sags in the power wiring during transients. This should not be an issue in well designed AC powered amps.
Phil
The Zen is an amplifier with a built-in excuse. It's hard
to slam.
Back to the topic, the power factor thing is mostly a matter
of reliability when the capacitance is huge. This assumes
that the transformer is not radiating the pulses, which of
course it does.
For everyone designing no-holds-barred amplifiers, having
physical distance from the transformer to circuit is a must.
Several feet is just great, but as little as a foot makes a
big difference. Rotating the transformer can also work
wonders.
Balanced circuitry is the other basic defense against this
form of noise acquisition.
to slam.
Back to the topic, the power factor thing is mostly a matter
of reliability when the capacitance is huge. This assumes
that the transformer is not radiating the pulses, which of
course it does.
For everyone designing no-holds-barred amplifiers, having
physical distance from the transformer to circuit is a must.
Several feet is just great, but as little as a foot makes a
big difference. Rotating the transformer can also work
wonders.
Balanced circuitry is the other basic defense against this
form of noise acquisition.
How big
Nelson,
In my experience, moving the xformer away in cases as we are discussion often makes things worse. It really isn't so much the xformer that radiates, but rather the supply wiring where all those sharp current pulses race through. And moving the xformers away increases (I would think) wire lengths. It's like lengthening your antenna for better transmission...
Cheers, Jan Didden
Nelson,
In my experience, moving the xformer away in cases as we are discussion often makes things worse. It really isn't so much the xformer that radiates, but rather the supply wiring where all those sharp current pulses race through. And moving the xformers away increases (I would think) wire lengths. It's like lengthening your antenna for better transmission...
Cheers, Jan Didden
Re: How big
Which is why you want to make sure that the feed cable is as low inductance as possible, which means keeping close spacing between conductors and using something like a twisted quad geometry rather than a twisted pair. Or perhaps a stacked ribbon type cable.
Of course, if one's going to go through the trouble of putting the transformer in a separate chassis, it wouldn't be much more trouble to just move your rectifier(s) and primary reservoir capacitors in along with it.
se
janneman said:
Which is why you want to make sure that the feed cable is as low inductance as possible, which means keeping close spacing between conductors and using something like a twisted quad geometry rather than a twisted pair. Or perhaps a stacked ribbon type cable.
Of course, if one's going to go through the trouble of putting the transformer in a separate chassis, it wouldn't be much more trouble to just move your rectifier(s) and primary reservoir capacitors in along with it.
se
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.