I do this to prevent noise from the transformer.
When a center-tapped transformer and single bridge supply see net DC or low
frequency AC through the center tap, any mismatch in the secondary windings
will tend to saturate the core, causing mechanical and radiated magnetic noise.
Thanks for the clarification Nelson. I've done the same now on my last 3 power amp builds.
I thought you also had some comments about phasing of the secondaries - but I may have gotten confused with someone else.
Last edited:
Yes, there are options in which all these problems can virtually disappear.
With Full Balanced Working, for example, we can generate an earth with no connection to the PSU or speakers, and most of the PSU noise is invisible through common mode cancellation.
A very elegant solution.
Would love to hear reports about SMPS both objective & subjective compared to traditional ones - I did not try them yet.
I guess if we can get digital recorded music to sound good we can also make SMPS's sound good . . .
Uhh huh , yeah ....
Attachments
Not very good with marginal fm station as source ....Would love to hear reports about SMPS both objective & subjective compared to traditional ones - I did not try them yet.
I guess if we can get digital recorded music to sound good we can also make SMPS's sound good . . .
I think it has the potential to be better - if full advantage is taken of the opportunities that are there - just connecting 2 channels with a differential signal etc might not be doing that, but designed balanced, from he ground up I think can be a different story.
Last edited:
I don't know either, but for some reason as I look ahead on my audio journey I see fully balanced as my final destination - I think I have a ( possibly irrational ) attraction to the elegance of the symmetry - everything about balanced just seems to make sense - with the possible exception of dominant 3rd HD but I'm hoping I won't mind that.
Agreed.
But then again, I have heard some very unlikely topologies to sound very good indeed. A German made power amp, Mega 1 by LAS (Linear Audio Systems), from way back in 1978 or so, is a good example. Looking at its schematic, you'd swear it was just one more mee-too, run-off-the-mill power amp, but once you hear it in action, you quickly change your mind.
On the other hand, I have heard a few fully complementary amps souding pretty dull and muddy, despite the fact that the platform as such is promising.
So, in end, you seem to always come to the same point - it ALL depends on how well it's applied.
SMPS
Every challenge with traditional line frequency PS's is present when designing a SMPS, and then some. The main difference is the scaling of values due to switching frequency. Although I am an advocate of KISS (the principle, not the band) and SMPS violate this relative to line-frequency supplies, current state of the art in SMPS are extremely reliable. As a radio engineer the main issue I have with SMPS is improperly shielded and filtered designs which conduct and/or radiate RF. As Pavel and others have discussed, RF hash can cause multiple issues in low-level audio circuitry. Correctly shielding and transverse-mode filtering SMPS outputs is an important aspect often overlooked in the race to lowest price point. Just ask any amateur radio operator who has had to battle RF hash all over the spectrum from cell-phone chargers, plasma TVs, etc...as I have had to do.
My experience is limited to designing SMPS for mobile audio amplifiers, so I do have some grounding in the subject. As far as a source of DC, SMPS can be as good or better in the same size class as a line-frequency supply. By this I am referring to impedance and reservoir capacity. With regards to transient response, within power limitations they can be much quicker with similar reservoir capacitors.
As per usual, the devil is in the attention to details in design and construction. I would posit that there is no basic reason why an SMPS could not perform and sound as good or better than a line-frequency supply at any given cost or size class for audio. For powering analog RF devices, the nod goes to traditional line frequency supplies.
(getting tired of putting flame suit on...)
Howie
Howard Hoyt - WA4PSC
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
Would love to hear reports about SMPS both objective & subjective compared to traditional ones - I did not try them yet.
I guess if we can get digital recorded music to sound good we can also make SMPS's sound good . . .
Every challenge with traditional line frequency PS's is present when designing a SMPS, and then some. The main difference is the scaling of values due to switching frequency. Although I am an advocate of KISS (the principle, not the band) and SMPS violate this relative to line-frequency supplies, current state of the art in SMPS are extremely reliable. As a radio engineer the main issue I have with SMPS is improperly shielded and filtered designs which conduct and/or radiate RF. As Pavel and others have discussed, RF hash can cause multiple issues in low-level audio circuitry. Correctly shielding and transverse-mode filtering SMPS outputs is an important aspect often overlooked in the race to lowest price point. Just ask any amateur radio operator who has had to battle RF hash all over the spectrum from cell-phone chargers, plasma TVs, etc...as I have had to do.
My experience is limited to designing SMPS for mobile audio amplifiers, so I do have some grounding in the subject. As far as a source of DC, SMPS can be as good or better in the same size class as a line-frequency supply. By this I am referring to impedance and reservoir capacity. With regards to transient response, within power limitations they can be much quicker with similar reservoir capacitors.
As per usual, the devil is in the attention to details in design and construction. I would posit that there is no basic reason why an SMPS could not perform and sound as good or better than a line-frequency supply at any given cost or size class for audio. For powering analog RF devices, the nod goes to traditional line frequency supplies.
(getting tired of putting flame suit on...)
Howie
Howard Hoyt - WA4PSC
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
SMPS
Do you have a graph showing the comparison with an excellent line-freq PS?
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
Do you have a graph showing the comparison with an excellent line-freq PS?
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
SMPS
Hi Sy!
I've never executed an off-line SMPS design. My understanding of PFC as applied to that design is it modulates the conduction angle synchronously with the incoming V phase to align the I phase. From the standpoint of spectral characteristics of the current-mode switching noise referred to the input, your comment makes sense...I would expect that once again, the devil would be in the details.
My experience with DC-input supplies shows it is easiest to filter a fixed-frequency continuous SMPS design, especially with synchronous rectification. It seems to me poorly executed a PFC design could cause erratic switching noise, especially when the supply transitions between continuous and discontinuous or hiccup modes. As I have experimented with designs I have found the low-load emissions to be the worst, with some of supply topologies (non-isolated with hiccup mode) creating wide-band RF hash. As a matter of design I would never use a SMPS controller in hiccup mode, despite the additional efficiency it achieves. This does however mandate a minimum load.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
Hi Sy!
I've never executed an off-line SMPS design. My understanding of PFC as applied to that design is it modulates the conduction angle synchronously with the incoming V phase to align the I phase. From the standpoint of spectral characteristics of the current-mode switching noise referred to the input, your comment makes sense...I would expect that once again, the devil would be in the details.
My experience with DC-input supplies shows it is easiest to filter a fixed-frequency continuous SMPS design, especially with synchronous rectification. It seems to me poorly executed a PFC design could cause erratic switching noise, especially when the supply transitions between continuous and discontinuous or hiccup modes. As I have experimented with designs I have found the low-load emissions to be the worst, with some of supply topologies (non-isolated with hiccup mode) creating wide-band RF hash. As a matter of design I would never use a SMPS controller in hiccup mode, despite the additional efficiency it achieves. This does however mandate a minimum load.
Howie
Howard Hoyt
CE - WXYC-FM 89.3
UNC Chapel Hill, NC
WXYC Chapel Hill, North Carolina - 89.3 FM
1st on the internet
Mine as well, but there are apparently some more sophisticated schemes that "spread out" the current pulse to achieve lower peak currents but with the same integrated energy, i.e., higher fundamental, lower harmonics. This is the distortion part of PFC, where the PF is reduced by 1/sqrt(1 + THD^2).
PFC Switch Mode Power Supply - Powersoft Audio for an example.
Linear vs. switcher
Here, for reference, are distortion plots of an ADC using three different supplies. The first is a Jameco linear supply kit. Its pretty crude and has more ripple than an regulates supply should. The second is a packaged Sola linear supply and finally a little Cosel switching supply. The opamps on the ADC board (AKD5394A) have no local regulation and only nominal local filtering. I was very careful with the grounding. The source has some hum in its output from leakage and supply ripple.
The conditions were not identical but similar and the plots will have different averages etc. so don't read the differences as due to the power supplies alone.
I could find no reason not to use the little switchers in this application even with such a stringent THD+N capability.
Here, for reference, are distortion plots of an ADC using three different supplies. The first is a Jameco linear supply kit. Its pretty crude and has more ripple than an regulates supply should. The second is a packaged Sola linear supply and finally a little Cosel switching supply. The opamps on the ADC board (AKD5394A) have no local regulation and only nominal local filtering. I was very careful with the grounding. The source has some hum in its output from leakage and supply ripple.
The conditions were not identical but similar and the plots will have different averages etc. so don't read the differences as due to the power supplies alone.
I could find no reason not to use the little switchers in this application even with such a stringent THD+N capability.
Attachments
- Status
- Not open for further replies.