Bob Cordell Interview: Power Supplies

Power supply quality

Here's a new topic to consider: power supply quality.

We constantly hear about people installing Black Gate capacitors, or Shottky diodes, or better bypasses in power supplies and getting startling sonic improvements. Often, we hear about better bass. I believe in putting good power supplies in power amplifiers, but often I've wondered what the mechanism is for improvement in sonics.

Sure, the power supply sags during a bass transient. Sure, we want to keep EMI off the supply lines. But if the rails don't sag to the point where the output stage doesn't clip or run out of headroom, why should there be a difference? After all, source-follower and emitter follower output stages have quite good power supply rejection. If we can take the amp up to nearly full power into a 4-ohm load without seeing any 120 Hz artifacts on the output, why should there be a sonic effect?

But there is a lot of anecdotal evidence that there is. Is this just a mystery of life in audio, or is there a mechanism to explain this?

Maybe the quality of the supply and its variations and stuff on it matter to the earlier stages in the amplifier, like the input stage and the VAS. But these stages require much less current, and it should be much easier to filter them or even regulate them. So if we just supply the earlier stages with exceptionally clean and stable rails, will we no longer see a big benefit from improving the main rails beyond stock? Maybe otherwise-very-high-quality amplifiers just don't pay enough attention to the rails for the input and driver stages?

I don't know. What do you guys think?

Bob

AKSA said:

Bob,

I've found there are gains to be had with superior power supply caps. I use the Nichicon Gold Tune, which are made to order. Nice caps with excellent bass.

However, topology must play a part. These caps have two purposes; one to smooth the power pulses coming off the rectifier (and these are pretty savage), and the other to accommodate speaker earth return current. If both these currents coexist in the same cap, then intermodulation results, so I like to use two caps per rail, separate them with a small resistor (typically 0R22, or even a tiny inductor around half a millihenry) so that the two currents are separated.

Cheers,

Hugh

Click to expand...

Thanks, Hugh

I agree completely. I also subscribe to the idea of inserting a little bit of impedance between two rail capacitors, as you mentioned. Much more bang for the filtering buck. Sometimes spending a lot of extra money on heroically low ESR is not the best approach.

Take a guy who has decided to provide 20,000 uF per rail. He can do it one of three ways, at fairly similar cost. He can put in one big 20,000 uF cap; or he can parallel a pair of smaller 10,000 uF caps; or, he can put in two 10,000 uF caps in parallel, but separated by a 0.22 ohm resistor. If you do the ESR math, the last approach is about 10 times better at filtering the nasty stuff, although total energy storage and p-p 120 Hz ripple are similar in all three cases. Of course, it is even better if the second cap is paralleled by some good-sized film caps, and maybe located very close to the output transistors, but this will increase cost a bit.

I also believe it is very important to be intentional about where the big currents flow, as they are nasty.

But I still suspect that a lot of designers overlook the bang for the buck available in making the lower-current supplies to the input and VAS stages more pristine.

If someone does an upgrade and doubles the size of the main caps, and hears better bass under normal listening conditions (i.e., not coming close to clipping), I'm guessing the effect he is hearing is better operation of the input and VAS stages because the supply they receive is moving around at low and subsonic frequencies only half as much, suggesting that they could have done just as well, if not better, by paying attention to the filtering (or regulation) of those lower-current sugnals.

Of course, the ordinary upgrader often doesn't have the option or desire to do much more than add/change existing capacitance. If that was an expensive amplifier, I would think that one should question why the designer didn't make a better power supply filtering arrangement in the first place.

Anyway, I am certainly an advocate of good power supplies, but I'm always curious about what works best, or is most efficient at making an improvement for a given $ spent.

Bob

Tony said:

Hi Bob,

thanks for your previous responses to my questions, you sure nailed them...now i am glad you opended up the topic about power supplies, i have had this suspicion that commercial amps power supplies are undesized. the manufacturer hoped that the end users will not play it at full volume all of the time.

just like car makers with their cars that are capable of running at 160km/hours, many of them hope the owners run them at speed limits of say 60 to 80 km.hr...

so for starters, let us say i have this power amp rated for 100watts per channel both channels running, what size power transformer would you specify?

Click to expand...

Your points are right on the money. Unfortunately, the answer to how many VA a power transformer should have for a given size power amplifier is not an easy one to answer, at least for me.

I could just say the bigger the better, but that would be a cop out. We all live in a world where the $ are constrained, even the Halcros and the Boulders. Do you spend the next $ on a bigger filter capacitor, or more output transistors, or a bigger heat sink, or a thicker, more polished front panel? This is what's called engineering (or a witche's brew between that and marketing).

Also, VA doesn't always mean the same thing from transformer maker to transformer maker or type to type. The biggest thing we think of is power supply stiffness, but we also must consider transformer heating, EMI radiation, and physical noise. Even stiffer is better is not always the choice, since this affects the difference between continuous power and peak power, i.e., dynamic headroom.

My criteria tend to be as follows. First, if you run the amplifier at full power into 4 ohms, both channels, for an extended period, the transformer temperature shall not exceed 50C. Secondly, if the amplifier delivers 100 watts continuous at clipping into 8 ohms both channels driven, it should deliver at least 160 watts per channel into four ohms (a perfect supply, assuming no current limiting in the output stage, would give you 200 wpc into four ohms). Some would call these criteria overly conservative, others would call them just the opposite. Relating advertized VA to achieving these criteria is difficult, at least for me. Recognize that with the usual capacitor input filter, the peak currents are extremely high, so the transformer's ability to source these very high, brief peaks is what tends to affect stiffness and regulation the most.

Hope this helps. There may be others here that can provide more concrete rules of thumb.

Bob

peranders said:

In Europe it's usually 105 deg C core temp at 40 or 70 deg ambient temp.

Click to expand...

Thanks for this information, P-A.

I'm certainly not a transformer expert, but I would assume that these ratings are arrived at by loading the transformer with a resistive load, as opposed to a rectifier feeding a capacitor input filter. Am I right? If this is the case, the VA rating may not be a very good indicator of regulation and stiffness in a typical amplifier power supply application.

Another thing. If the VA rating is primarily an indicator of maximum core temperature, then different transformers of the same VA rating could have vastly different useable d.c. rectified current capabilities and regulation in the real world. For example, copper is now very expensive. A transformer manufacturer could probably arrive at the same thermal VA rating by using less copper and more iron, and maybe arranging that iron to dissipate heat better. The resulting transformer would have significantly more winding resistance and would not have as good regulation, even though it had the same VA.

So I guess what I'm saying is that VA doesn't do a very good job of telling the whole story.

Bob

Re: Re: Power supply quality

MikeB said:

Bob, i think that this issue is one of the most underestimated ones in a lot of amps.
Even the outputstage (EF/SF) is quite sensible to variations from the supply lines because of the highish outputcapacitance (C <-> B) of power devices. Especially for higher freqs psrr degrades quickly. Because of this, EMI and other RF stuff from the powersupply easily find its way to the output / into the feedbackloop. With more outputdevices this effect is even more increased.

For the earlier stages you can use some simple RC-filtering in the supplies, filtering out most of the garbage. I always found instant increase in sonic qualities by applying these filters to the early stages.

In ClassAB/B amps, the supply rails are severely wasted with harmonics from the halfwave currents flowing into the outputdevices. Especially when reproducing treble content, the harmonics of these halfwaves become very high freq and difficult to filter because of "bad" ESR from the elyt caps and inductance of traces and wiring.

An easy way reducing this problem seems to be placing a maybe 1000µF between the collectors of the upper and lower outputdevices with short traces. (smoothing the halfwave currents to mostly 2nd harmonic content) Still, as the fundamental of these half wave currents is double the frequency of the signal reproduced, it might be harmful letting these find back into the feedbackloop. This can also find its way back through inproper grounding.

Paying enough attention to these problems seems to be one of the keys to a good sounding power amp.

A difficult question: How much psrr is enough ? -90db ?

Mike

Click to expand...

I agree with everything you have said, Mike. Those Class-AB half-wave output transistor currents are especially toublesome in regard to their radiated magnetic field. Closing those current paths right close to the output stage, with good ESR out to high frequencies, and also with attention to any needed damping to prevent resonances (snubber networks) is very important. Of course, this is a much bigger deal than just sticking in a couple of expensive Black Gate capacitors in the main power supply. I guess the point is that good engineering more often overcomes blind brute force.

I have also used the X capacitor, most recently in my Super Gain Clone amplifier. Good supply treatment is especially important in IC power amplifiers because in those we don't usually have the option of putting extra filtering in for the input and VAS stages.

Bob

janneman said:

Note to moderators: Maybe the power supply discussion as initiated by Bob should be split off in a separate thread?

Jan Didden

Click to expand...

Jan,

I don't think the power supply discussion is worthy of splitting off. I think its useful to be able to move around within a larger topic as long as it doesn't get out of control. For example, I don't see this thread as an error-correction thread. Had that been the case, the recent useful discussions of SOA of output transistors would have been seen as inappropriate.

Bob

lumanauw said:

Hi, Gerhard,



How to do that (0 Zsource)? Using overspec power transformer + 10's or 100's uf cap bank + many parrarel rectifier diodes?



Gerhard, sometimes ago I tried an experiment. Comparing GainClone, with ordinary PS and with car batteries. Gainclone+car batteries have better audible sonics.
(But you won't notice this if you don't make instant A/B switch comparison)

What do you think causes it?

Click to expand...

I'm always intrigued when I see high-end designs run from batteries (not just a Gain Clone). Does this mean that it is not possible or economical for us to make a power supply as good as a battery? In fairness, to do as well as a battery, one might have to house the power supply in a separate box to completely avoid the effects of interference emissions from the PS.

As I mentioned earlier, I do think Gain Clones need a good, quiet, well-filtered power supply for them to yield their best performance, since the input and VAS circuits must live off of the same rails as the output transistors.

Bob