The advantage of regulated power supplies on amplifiers?

[routlaw]

I remember Keith going into some detail awhile back discussing the fact that some of the new Emotiva Amps have regulated power supplies which apparently is difficult to achieve with conventional linear power supplies but much easier with SMPS. 

I'm intrigued with this concept and what it means but more importantly how it affects the sound and performance of an amplifier assuming it really does make a difference. Reading the literature on the Audio By Van Alstine website they also have bragging rights about many if not all of their amps having regulated power supplies, but they do this with a linear power supply rather than a SMPS. This company seems to have an almost cult following for many of their products, especially the amps. They don't publish much in the way of specs though which I find very frustrating.

Interestingly enough their beast of amps (850 mono blocs and others) weigh in at about the same as the new line of Emotiva Amps but they are doing it with linear power supplies which normally account for at least half the conventional weight of most any amplifier. This seems a bit mysterious yet interesting to me. For those who are not familiar with the company and their products but interested in pursuing their website the link is below. Please understand though I am not trying to drum up business for them by any means. Besides there really isn't much info to go on there.

avahifi.com/collections/power-amplifiers


Thanks in advance.

[mgbpuff]

All 24, 12, 10, 5v power supplies are inherently regulated. The output device power supply 70v (tubes), 35v(solid state) or so need not be regulated - why? Because seldom does the musical output reach the maximum level of the power supply, which is called clipping. Clipping is undesirable anyway. Also because solid state amplifiers use feedback, they are self regulating and the basic power supply being regulated or not has little effect.

[DYohn]

The potential advantage of regulated power supplies is voltage consistency with varying load, meaning a standard transformer or switch mode P/S can "droop" under heavy load, and "spike" when load is released. This can happen rapidly during musical transients. This change in supply voltage can cause amplifier components to operate closer to their limits and produce more noise and other distortion than if they are consistently in their desired operating band. It is also why linear power supplies use large capacitor banks - to suppress these variations - generally a regulated supply can get away with using smaller capacitors.

The effectiveness of P/S regulation depends on the application, of course, but that's the theory.

[routlaw]

Oct 7, 2019 10:11:05 GMT -5 DYohn said:

The potential advantage of regulated power supplies is voltage consistency with varying load, meaning a standard transformer or switch mode P/S can "droop" under heavy load, and "spike" when load is released. This can happen rapidly during musical transients. This change in supply voltage can cause amplifier components to operate closer to their limits and produce more noise and other distortion than if they are consistently in their desired operating band. It is also why linear power supplies use large capacitor banks - to suppress these variations - generally a regulated supply can get away with using smaller capacitors.

The effectiveness of P/S regulation depends on the application, of course, but that's the theory.

Thanks. This makes some sense sort of.

But… what doesn't make sense in this theory, (or at least to me is), is very seldom will any Class A/B amp ever require more than a 10 watt bias at Class A, yet the XPA-1 Gen 2 mono blocs that I have are rated at 60 WPC once the switch is set to Class A. Whether its 10 watts or 60 Watts neither demand amounts to much energy draw in the big scheme of things. So how is it then that there could be enough voltage swings, drops or spikes, when so relatively little current and power are being used in the first place? 

Perhaps these are two entirely separate entities when discussing amplifier design. 

[KeithL]

This is actually a very complex subject - but let me try to hit some of the high points.

It's also worth noting, for anyone familiar with tube equipment, that the need for and benefits of regulated power supplies are very different in typical tube and typical solid state equipment.

The other thing to note is that, with a linear power supply, adding regulation drastically reduces efficiency.

Let's start with a definition...

A regulated power supply is simply a power supply that always delivers the same voltage, even when the input voltage or the load changes.

And, since noise is nothing more than a change in the output level, a regulated supply also delivers an output voltage with no noise.

(And, of course, some do this better than others.)

The fact is that most of the circuitry in modern solid state equipment is, by design, relatively insensitive to power supply regulation.

However, the circuitry that handles very low level signals, like preamps, and the input circuitry on power amps, is the most sensitive to it.

But, because that circuitry also uses very little power, providing it with regulated power isn't especially problematic.

Because only a small regulated power supply is required, the design is relatively simple, and the efficiency really isn't that important.

(Who really cares if a one watt power supply is only 50% efficient... it still won't burn much power or generate much heat.)

In contrast, the output section of your power amp runs on relatively high voltage, requires many amps of current at full power, and requires a large power supply.

Designing a linear regulated power supply big enough to run it would be complicated and expensive - and it would make a lot of extra heat.

(Adding linear regulation to the output stage of a big amp would typically add between 30% and 100% to the complexity, the  weight, the amount of heat generated, and the price tag.)

Luckily, that part of the amplifier isn't very sensitive to voltage variations or noise, so running it from an unregulated supply won't affect the performance much...

That's why you almost never see big amplifiers with linear regulated power supplies.

However, because of how they actually work internally, SMPS are inherently regulated.

With a typical linear amplifier, as the audio output waveform goes up and down, and draws varying amounts of current, this causes the voltage output of the unregulated power supply to go up and down a little bit.

And, even with all of the filtering provided by the power supply capacitors, some of the variation from the 60 Hz line frequency still finds its way through.

(And the amount of line frequency variation that leaks through varies depending on how much current you're drawing.)

And, if any when you cause the amp to clip, that causes a sudden jump in the current being drawn, which is even more likely to cause a small but sudden jump in the power supply output voltage.

And, even though the output stage is relatively insensitive to changes in its supply voltage, it is not totally insensitive to them.

As a result, these small changes can and do cause tiny amounts of distortion and noise, which makes it through to the audio output, and becomes part of the output signal.

And, by PREVENTING these changes in voltage  from occurring, a regulated power supply prevents this from happening.

To be quite honest, the benefits of a fully regulated power supply on a power amp are relatively subtle, and you might or might not notice them, depending on the design.

However, when we can deliver even a slight audible benefit, while at the same time reducing the weight by thirty or forty pounds, improving reliability and efficiency, and reducing cost, we consider that to be a huge overall gain.

Oct 7, 2019 9:35:56 GMT -5 routlaw said:

I remember Keith going into some detail awhile back discussing the fact that some of the new Emotiva Amps have regulated power supplies which apparently is difficult to achieve with conventional linear power supplies but much easier with SMPS. 

I'm intrigued with this concept and what it means but more importantly how it affects the sound and performance of an amplifier assuming it really does make a difference. Reading the literature on the Audio By Van Alstine website they also have bragging rights about many if not all of their amps having regulated power supplies, but they do this with a linear power supply rather than a SMPS. This company seems to have an almost cult following for many of their products, especially the amps. They don't publish much in the way of specs though which I find very frustrating.

Interestingly enough their beast of amps (850 mono blocs and others) weigh in at about the same as the new line of Emotiva Amps but they are doing it with linear power supplies which normally account for at least half the conventional weight of most any amplifier. This seems a bit mysterious yet interesting to me. For those who are not familiar with the company and their products but interested in pursuing their website the link is below. Please understand though I am not trying to drum up business for them by any means. Besides there really isn't much info to go on there.

avahifi.com/collections/power-amplifiers


Thanks in advance.

[KeithL]

That is not at all true.

Low voltage power supplies are not at all "inherently regulated"... and most early low cost ones were not.

However, a lot of early digital integrated circuits absolutely required a regulated 5v supply in order to work properly.

And it requires very little effort or cost to add regulation to a small low-cost power supply.

And small SMPS, which are inherently regulated, have almost entirely replaced the linear variety (which may be regulated or not).

As a result, like wooden matches, while unregulated linear power supplies are in fact still available, you don't see them much any more.

Oct 7, 2019 9:56:19 GMT -5 mgbpuff said:

All 24, 12, 10, 5v power supplies are inherently regulated. The output device power supply 70v or so need not be regulated - why? Because seldom does the musical output reach the maximum level of the power supply, which is called clipping. Clipping is undesirable anyway.

[DYohn]

Oct 7, 2019 10:30:49 GMT -5 routlaw said:

Oct 7, 2019 10:11:05 GMT -5 DYohn said:

The potential advantage of regulated power supplies is voltage consistency with varying load, meaning a standard transformer or switch mode P/S can "droop" under heavy load, and "spike" when load is released. This can happen rapidly during musical transients. This change in supply voltage can cause amplifier components to operate closer to their limits and produce more noise and other distortion than if they are consistently in their desired operating band. It is also why linear power supplies use large capacitor banks - to suppress these variations - generally a regulated supply can get away with using smaller capacitors.

The effectiveness of P/S regulation depends on the application, of course, but that's the theory.

Thanks. This makes some sense sort of.

But… what doesn't make sense in this theory, (or at least to me is), is very seldom will any Class A/B amp ever require more than a 10 watt bias at Class A, yet the XPA-1 Gen 2 mono blocs that I have are rated at 60 WPC once the switch is set to Class A. Whether its 10 watts or 60 Watts neither demand amounts to much energy draw in the big scheme of things. So how is it then that there could be enough voltage swings, drops or spikes, when so relatively little current and power are being used in the first place? 

Perhaps these are two entirely separate entities when discussing amplifier design. 

Those concepts are really not related, but understand that an amp operating in Class A requires more current than one operating in Class A/B, so in theory the impact of load fluctuation on the P/S could be greater for the same power output in a Class A amplifier.

[KeithL]

The operating bias level chosen for an amplfier has pretty much nothing to do with whether the power supply is regulated or not.

A higher bias current may often cause the output devices to run in a more linear portion of their operating range - which may result in lower distortion.

However, it also causes them to generate more heat, which results in a need for bigger heat sinks, and reduces component life.

It's also worth remembering that, in typical music listening, the average listening level is about 1/10 of the peak level.

So, for example, if you're listening to music at a peak level of 500 watts, your average listening level is probably about 50 watts.

Since a typical Class A/B amp is about 70% efficient, that means that, on average, you're generating about 25 watts of heat due to the inefficiency of the amplifier itself.

And, by running that XPA-1 Gen2 in "Class A mode" rather than "Class A/B mode" you are doubling or tripling the amount of heat it is producing.

(And, if you've ever owned one, you will have noticed that they do in fact get a lot warmer in Class A mode...)

The size of the capacitors required in a power supply depend on several things... the most important of which is operating frequency.

When you rectify the 60 Hz voltage coming from your wall outlet, it produces a voltage waveform that has peaks every 1/60 or 1/120 of a second (depending on how you rectify it).

The capacitors act as a sort of reservoir for current...

At each peak, the output voltage rises to that peak, and the capacitors are charged to that level...

Then, between peaks, the capacitors continue to supply current, as their output voltage slowly drops...

The capacitors must have enough capacity to continue to supply sufficient current until they are recharged by the next peak.

(Thus the output voltage from the capacitor bank rises and falls slightly in the classic sawtooth pattern.)

Because the circuitry they power is somewhat immune to variations in supply voltage a certain specific amount of variation is considered to be acceptable.

And, yes, in a linear regulated supply, smaller capacitors can be used, because the regulators should remove any variations that might not be eliminated by the capacitors.

A regulated SMPS actually requires much smaller capacitors for several distinct reasons....

First, because it's regulated, the circuitry in the SMPS actively controls the output voltage.

The line voltage is initially converted to DC, and filtered by capacitors which can be relatively small, because the level and amount of noise ripple in this reservoir isn't especially critical.

This reservoir is then used by the active regulators as "raw material" to produce the carefully regulated voltage that they deliver at their output.

This output is then again stored in and filtered by capacitors.

However, because the regulators operate at a very high frequency, their output capacitors are refilled MUCH more often, so they can also be relatively small.

It's also worth noting that, with a linear unregulated supply, any variation that makes it through will usually be in the form of audible 60 Hz or 120 Hz hum.

This is also true for a linear regulated power supply.

In contrast, any tiny amount of variation that makes it through the regulation in an SMPS will be ultrasonic, and so will be inaudible.

Oct 7, 2019 10:30:49 GMT -5 routlaw said:

Oct 7, 2019 10:11:05 GMT -5 DYohn said:

The potential advantage of regulated power supplies is voltage consistency with varying load, meaning a standard transformer or switch mode P/S can "droop" under heavy load, and "spike" when load is released. This can happen rapidly during musical transients. This change in supply voltage can cause amplifier components to operate closer to their limits and produce more noise and other distortion than if they are consistently in their desired operating band. It is also why linear power supplies use large capacitor banks - to suppress these variations - generally a regulated supply can get away with using smaller capacitors.

The effectiveness of P/S regulation depends on the application, of course, but that's the theory.

Thanks. This makes some sense sort of.

But… what doesn't make sense in this theory, (or at least to me is), is very seldom will any Class A/B amp ever require more than a 10 watt bias at Class A, yet the XPA-1 Gen 2 mono blocs that I have are rated at 60 WPC once the switch is set to Class A. Whether its 10 watts or 60 Watts neither demand amounts to much energy draw in the big scheme of things. So how is it then that there could be enough voltage swings, drops or spikes, when so relatively little current and power are being used in the first place? 

Perhaps these are two entirely separate entities when discussing amplifier design. 

[KeithL]

It's a bit more complicated than that.

When you operate an amplifier like the XPA-1 in "Class A mode" you are actually operating it in "high bias A/B mode".

(All Class A/B amps operate in Class A part of the time; in that mode, the XPA-1 is operating in Class A more of the time.)

Therefore, because the average load on the power supply is higher in that mode, you can expect it to be more sensitive to additional load drawn from it.

However, with an amplifier that operates in "complete Class A mode" the reverse is actually true.

In that case, the power supply load, the power drawn, and the heat generated, all remain virtually constant.

A 200 watt "full Class A amp" burns 400 watts all the time - whether idling or running at full power.

The load on the power supply is constant - and very heavy - at all times.

(This is not absolutely true because the speaker load consumes some power.... but the percentage of load fluctuation between no load and full load remains quite small) 

Another thing worth noting is that, with an unregulated linear supply, the voltage variation and noise virtually  ALWAYS get progressively worse as the load increases.

Even with a regulated linear supply, while they should be minimal, both the voltage variation and noise generally get at least slightly worse as the load gets heavier.

In contrast, the performance of an SMPS may remain virtually the same, or may even improve in some ways, as the load increases.

Oct 7, 2019 11:15:05 GMT -5 DYohn said:

Oct 7, 2019 10:30:49 GMT -5 routlaw said:

Thanks. This makes some sense sort of.

But… what doesn't make sense in this theory, (or at least to me is), is very seldom will any Class A/B amp ever require more than a 10 watt bias at Class A, yet the XPA-1 Gen 2 mono blocs that I have are rated at 60 WPC once the switch is set to Class A. Whether its 10 watts or 60 Watts neither demand amounts to much energy draw in the big scheme of things. So how is it then that there could be enough voltage swings, drops or spikes, when so relatively little current and power are being used in the first place? 

Perhaps these are two entirely separate entities when discussing amplifier design. 

Those concepts are really not related, but understand that an amp operating in Class A requires more current than one operating in Class A/B, so in theory the impact of load fluctuation on the P/S could be greater for the same power output in a Class A amplifier.

[vcautokid]

Now this is an interesting topic. Nice. Regulation as you need it. Makes total sense. Totality as a philosophy a grand Panacea it doesn't always make. Or like I love to say. "Sometimes is never always."

[mgbpuff]

Oct 7, 2019 11:00:05 GMT -5 KeithL said:

That is not at all true.

Low voltage power supplies are not at all "inherently regulated"... and most early low cost ones were not.

However, a lot of early digital integrated circuits absolutely required a regulated 5v supply in order to work properly.

And it requires very little effort or cost to add regulation to a small low-cost power supply.

And small SMPS, which are inherently regulated, have almost entirely replaced the linear variety (which may be regulated or not).

As a result, like wooden matches, while unregulated linear power supplies are in fact still available, you don't see them much any more.

Oct 7, 2019 9:56:19 GMT -5 mgbpuff said:

All 24, 12, 10, 5v power supplies are inherently regulated. The output device power supply 70v or so need not be regulated - why? Because seldom does the musical output reach the maximum level of the power supply, which is called clipping. Clipping is undesirable anyway.

Wrong adjective, ‘inherently’; I really meant ‘almost universally’.

[routlaw]

Keith many thanks for all of the well thought out and concisely written explanations. This helps to explain things a lot, though I may never have more than a beginners true understanding of this subject.

[routlaw]

So after cogitating on this overnight, at least theoretically, these new amps with their regulated power supplies and associated SMPS should handily outperform the older amps such as the XPA-1 Get 2 with their heavier linear power supplies even while operating in Class A mode. Would everyone concur on this assessment. 

Reading what few end user reviews there are on them, it appears if nothing else they sound considerably different than other amps people have been used to, including but not limited to Emotiva amps.

[mgbpuff]

Oct 8, 2019 10:27:40 GMT -5 routlaw said:

So after cogitating on this overnight, at least theoretically, these new amps with their regulated power supplies and associated SMPS should handily outperform the older amps such as the XPA-1 Get 2 with their heavier linear power supplies even while operating in Class A mode. Would everyone concur on this assessment. 

Reading what few end user reviews there are on them, it appears if nothing else they sound considerably different than other amps people have been used to, including but not limited to Emotiva amps.

I do not agree, nor do reviewers such as John Atkinson of Stereophile, who criticized the high end above 10khz as having high THD (1% at 20 khz). Is this due to the SMPS? Some think so!

[routlaw]

A google search did not turn up a Stereophile Review of these amps, can you provide that link? Thanks.

However I was able to find a number of other reviews of the DR 2 and Get 3 amps from other reviewers and have not found anything discouraging at all.

[MusicHead]

For those interested, a series of articles on EETimes on this subject by Doug Self, very well known in the audio circles as a power amplifier designer:

www.eetimes.com/document.asp?doc_id=1278195

[mgbpuff]

Oct 8, 2019 11:28:37 GMT -5 routlaw said:

A google search did not turn up a Stereophile Review of these amps, can you provide that link? Thanks.

However I was able to find a number of other reviews of the DR 2 and Get 3 amps from other reviewers and have not found anything discouraging at all.

www.stereophile.com/content/emotiva-xpa-gen3-two-channel-power-amplifier

[routlaw]

Thanks for the link, but I wasn't referring to the Gen 3 amps but rather the XPA DR1's or DR2 which I assume to be quite a bit different. No doubt the reason it didn't show up in a google search.

Busy week, so only had time to quickly peruse this article or review. I didn't come away with the feeling either JA or the reviewer were all that negative based upon one measurement. In the end they gave it a B rating no small feat for a $1000.00 amp, especially for this magazine. FWIW, I struggle to read reviews such as this, just give me the facts and cut to the chase is all that is needed. If I wanted to read a creative short story or novel I would do just that without the discussion of hifi products thrown in.

What few reviews I have found and read on the DR 2 and DR-1's have all been very positive with none of the negatives that Stereophile came up with on the Gen 3.

Final thoughts from JA quoted below:

"Emotiva's XPA Gen3 amplifier offers high powers with low levels of noise and distortion, at least at frequencies below 10kHz, at a very affordable price. But the amplifier's problems at the top of the audioband bother me. Predicting the subjective effect of this objective behavior is probably an exercise in futility, but I do wonder if it correlates with Herb's finding that the Emotiva sounded "hard and 100% masculine" through two of the speakers he tried with it.—John Atkinson"


Read more at www.stereophile.com/content/emotiva-xpa-gen3-two-channel-power-amplifier-measurements#oDhlTP9KUAsDtWS5.99

[davidl81]

Oct 9, 2019 11:29:15 GMT -5 routlaw said:

Thanks for the link, but I wasn't referring to the Gen 3 amps but rather the XPA DR1's or DR2 which I assume to be quite a bit different. No doubt the reason it didn't show up in a google search.

Busy week, so only had time to quickly peruse this article or review. I didn't come away with the feeling either JA or the reviewer were all that negative based upon one measurement. In the end they gave it a B rating no small feat for a $1000.00 amp, especially for this magazine. FWIW, I struggle to read reviews such as this, just give me the facts and cut to the chase is all that is needed. If I wanted to read a creative short story or novel I would do just that without the discussion of hifi products thrown in.

What few reviews I have found and read on the DR 2 and DR-1's have all been very positive with none of the negatives that Stereophile came up with on the Gen 3.

Final thoughts from JA quoted below:

"Emotiva's XPA Gen3 amplifier offers high powers with low levels of noise and distortion, at least at frequencies below 10kHz, at a very affordable price. But the amplifier's problems at the top of the audioband bother me. Predicting the subjective effect of this objective behavior is probably an exercise in futility, but I do wonder if it correlates with Herb's finding that the Emotiva sounded "hard and 100% masculine" through two of the speakers he tried with it.—John Atkinson"


Read more at www.stereophile.com/content/emotiva-xpa-gen3-two-channel-power-amplifier-measurements#oDhlTP9KUAsDtWS5.99

One way to find out... buy the Dr-1 or 2 from Emotiva and try it out for 30 days. FWIW I was happy with my DR-2 through my Marantz 8802, but going to the RMC-1 has really opened up the amp in 2 channel.

[routlaw]

davidl81, I don't doubt this at all and not the first time I've heard this regarding those Marantz HT receivers. They make nice gear but its what it is, an all in one black box, with performance compromises. Awesome you moved up to the RMC-1.