The intention was purely to show that even cheap, simple and unregulated SMPS like the one used can overcame classic linear PSU and that was clearly proved, not just because of PSRR performance, that at least. SMPS simply brings the amp to higher level of sound quality without any special measures taken. That is also good starting point to go to much more capable SMPS selection and take the most suitable one for VSSA. As Esperado said: "Why bother?" (with heavy and expensive linear PSU).
This puzzles me.
Either the linear ones are not well oversized or the SMPS is very well regulated (which is mostly the case)... I dont have experience with switching PSUs in amplifiers, but I know of the crap that bad ones throw on the other components and I'm not so fond of them.
hmm...
Dear Shaan, you're not wrong, still "one transistor too much", this time between GND connection.
Yes they're dramatically opposite, but correct.
VSSA is the amp so revealing, that a power supply type can be reliably evaluated through listening test with ease. Look, linear PSU in the first stereo VSSA is 350 VA with 40 mF capacitor's bank per channel, so all together 700 VA and 80 mF. Enough power and current capability in absolute terms but lacking in "speed", that means poor in current transients capability in time. SMPS simply deliver more consistent current per time period, as if it never runs out of current. 100 kHz perfect square charging pulses to the capacitor bank is 1000x faster than 100 Hz semi sine pulses. Also the output impedance of SMPS is lower, helping to deliver higher ripple free current peaks than the linear one. All these little positive characteristics of SMPS combined together, brings sumarized advantage which is clearly audible through natural transients friendly amplifier like the VSSA is.
Some words about this. Yes, audiophiles pretend Linear PSU to be better over SMPS, and that watt i thought till... i had to replace urgent a supply in a monitoring amp while i had only one SMPS to help-me. I was surprised by the improvement. I made several tests after that, they all correlate. One friend, making restorations of old amplifiers had the same comments, replacing the original linear power supply in an old QUAD amp by a SMPS.
Lets try to understand why. First, the caps refresh 1000x more often/faster with a SMPS, and this is better for transients reproduction. Isolation with AC HF parasitic signals are better (less capacitance between primary/secondary trafos coils). Less HF components due to the diode's switching in linear (very hard to cancel). Ripple is outside of the audio bandwidth. In fact it is easier to provide very clean rails with a SMPS than a linear.
On the negative side, more pollution of AC lines, due to the switching frequency.
The hypex SMPS L.C. used is not regulated, but have a very symmetrical ripple, so the amp easily reject-it. The ripple is not an issue, as the amp is very fast, and will not produce IM because like if it was not able to follow-it.
Last, you can see, in your computer how it is possible to provide incredible stable voltages to CPUs at Giga Hz and huge currents (>50 Amps).
Now, it would be interesting if L.C can produce comparative plots of the rails, under charge, of SSA with SMPS, and Linear. I believe this will talk by itself.
[Edit]L.C published this in the same time i was writing my words, quite the same than his ones, strange, isn't it ?
Last edited:
SMPS cannot provide a constant voltage for a 'sustain' signal. My suspicion further is that it introduces a bit more third harmonic which makes the amp sound crisper.
A linear supply with fast rectifiers, big caps, series-pass darlington cap multiplier/regulator followed by multiple small caps should sound fast enough and sweeter compared to SMPS. In prolonged listening tests, well designed linear should win.
A linear supply with fast rectifiers, big caps, series-pass darlington cap multiplier/regulator followed by multiple small caps should sound fast enough and sweeter compared to SMPS. In prolonged listening tests, well designed linear should win.
Last edited:
The problem with smps is that ripple tends to be unlinear.
With a smps running at 500khz the main ripple is 500khz.
But the regulation Will try to compensate for the current draw.
It works but the ripple changes with the current draw.
So if you drive a load with a 1khz sine. It Will be present as a ripple that changes with 1khz of frequency. The AMP also has to compensate for that. This is not easy as psrr falls of with the frequence of the ripple .
With a smps running at 500khz the main ripple is 500khz.
But the regulation Will try to compensate for the current draw.
It works but the ripple changes with the current draw.
So if you drive a load with a 1khz sine. It Will be present as a ripple that changes with 1khz of frequency. The AMP also has to compensate for that. This is not easy as psrr falls of with the frequence of the ripple .
Hello
I built and tested over 100 or so amplifier over 23 years DIY. Here I had some serious clone to like AX, Aleph30, Hiraga (both) Symasym, Naim, ProFet so on.
All do these topology was my very first build and I did rebuild it after 20 years I can say the best sounding amp among I ever built! For that reason I still work on these topology.
I think LC took further these topology (I didn't tested yet) but if it sound as good or better than mine, easily I believe that statement the best amp he ever built or tested..
About the PS I would advise please test it very well because these amp is supper sensitive to the power supply but I can say sa same to evry (even passive) componets used on the PC board..
One transistor can change the sound of the amp in a way will sound like a total different amplifier.
I'm not sure the PS U test will be better job than the regular PS with good quality capacitors.
Please pay att. not to lose the deepnes or the dynamic of the sound.
Of course we all learn by test and try out new things.
Greetings Gabor
Without taking sides, i too feel this is open to discussion on claims.
I realize this is a primarily constructors forum, however.
LC has done, as have others, a good job on the Compl-push-pull CM design iteration that has been going on for some 30+ years. As one who has developed the CMFB design concept and topolgy, I should know something about iterations.
The best use of the Current-Mode fb topology for audio amplifiers is in power amps where high current at high speed are needed most.
But dont forget what its all for.... making the music more real and thereby increasing the enjoyment of the music.
Enjoy,
-Richard Marsh
Last edited:
I have to make special thank to forum member supernet, who gracefully shared his presence, listening room and speakers, so we could made this test possible. If he's willing he could also say word or a two.
Well...many thanks for your effort and design too
As LC already mentioned, the difference between two VSSA was PSU and feedback capacitor. Classic PSU / FM cap vs SMPS / Nichicon FG cap. We tried logically to understood the difference regards the different caps and PSUs.
My clear opinion was that VSSA "classic" sounded a bit thin, vocals have a bit of goose effect, bass was more rounded, but with less punch and most interesting...music seems to be on slow side. The "smps" VSSA was other story...better 3D stage, more air in it, bass suddenly hit quick & hard and music started to get live.
From my previous amplifier experiences (around 15 years), I can clearly say that 2D sound with goose effect on vocals is quite similar for very low ESR capacitors like Panasonic FM or Rubycon ZL. These caps might be good for PSU, but I never use them anymore in feedback or other sensitive application. Regards the PSU...I think most clearly difference between smps and classic, is dynamic response and bass definition.
If I can made a similarity, I would choose comparison between two identical car engines. One have turbo charger, one is without it. Also...I think that theory do not always correlate with practice. I was also very skeptical about smps PSUs...until yesterday
One idea here would be to look at the supply rails while playing some music into a real load. You may find that there is a clear difference between the SMPS and the standard PSU.
On the linear PSU (some people call it a 'bulk' PSU = standard transformer, rectifiers and caps)), you will have 100Hz Triangular ripple, and on top of that, the half wave music harmonics. I am wondering if on the SMPS you have no 100Hz ripple reflected from the primary bulk capacitor (I am assuming the SMPS is regulated) and whether or not you have better filtering of the audio half wave gunk - you will need to put a scope on tho check it out though. If your PSRR is low, its possible that this may be affecting the sound (ripple eater can work wonders to clean the supply rail up BTW).
On the linear PSU (some people call it a 'bulk' PSU = standard transformer, rectifiers and caps)), you will have 100Hz Triangular ripple, and on top of that, the half wave music harmonics. I am wondering if on the SMPS you have no 100Hz ripple reflected from the primary bulk capacitor (I am assuming the SMPS is regulated) and whether or not you have better filtering of the audio half wave gunk - you will need to put a scope on tho check it out though. If your PSRR is low, its possible that this may be affecting the sound (ripple eater can work wonders to clean the supply rail up BTW).
SMPS400 is unregulated so there cannnot be any regulation influence to audio harmonics from SMPS itself, it simply emulates linear PSU, since both has completely the same 100 Hz ripple in shape as well as in absolute value. So closest possible match from both technologies suitable to be the first step to compare the effects on sound. Same ripple (transferred from primary caps bank) in SMPS is completely without any 100 kHz switching artefacts noticable. Quality in reproduction favourite to SMPS simpy goes to, as Esperado already said, better coupling through HF switching part and smaller transformer, resulting in better current response in all respects. Soon there will be also test with top regulated SMPS, so stay tuned.
AC filtering cap at input bias current injection point: 2,2 mF ELCO parallel with 10 uF MKT connected to GND.
Supernet was also very sceptical about SMPS, me as well, but the sound reproduction clearly showed which one the amp itself likes more to be connected to
Same kind of improvement that i got with my stabilization stage, despite a HUGE linear PSU before.Supernet was also very sceptical about SMPS, me as well, but the sound reproduction clearly showed which one the amp itself likes more to be connected to
Here, Amplificateur Quad 405 SMPS, a friend i was talking about made this comment about a Connex SMPS on a Quad 405:
In its original version, The Quad 405 is yet splendid. When powered by two SMPS blocs, it become yet better, and reach an other amplifier class: Peter walker must do rock'n roll in his grave.
All sceptical before, all convinced by our ears: Don't believe in mysterious evils, switch yourself.
At a certain level of quality, all improvements lies in the power supplies.
Now, L.C will try if adding the kind of simple regulation i use improve-it further, and if using one (double power rating) supply instead of two separate is not yet better, that i believe: more instant current, better 3D image, because the two channels behave the same. This, again, goes against all audiophiles claims for dual mono blocs, but it is logical.
Last edited:
Intersting comments about the SMPS - I am not fixed on either type - use what is best. But, my experience todate has only been with Linear PSU's. Maybe something to investigate in the future. I think Bruno Putzey's company in the Netherlands is doing
400W and 700W SMPS unreg for very resonable prices (bare board).
400W and 700W SMPS unreg for very resonable prices (bare board).