Impact of AC Distortion & Noise on Audio Equipment

[Tks]

Wish that thing had a headphone-out. Wanted to see Amir talk about how he experienced the best bass for a full second of 300v output.

 

[Xyrium]

Can do. I only have true sine UPS's though. They have come down in price so much that I don't see any reason to get the non-sine ones.

Sounds like another test is imminent...what type of sine wave are these low cost "true sine" UPSs providing? Just sayin'...

 

[Bob from Florida]

This is an ASR research project to see what impact AC noise and distortion has on performance of audio equipment. I have been wanting to write this article some three years ago and finally getting a chance to do the testing and release it.

As you will see, the heart of this study is a special instrument: the BK Precision 9801 Lab AC programmable generator:
View attachment 145395

With this machine I can generate a wide range of AC voltages, frequency and as you will see, distorted AC waveform. It is a massive box despite its smallish front. It goes way back and has a noisy fan to keep it cool. Heat is generated because this is basically a high-voltage audio amplifier. The audio source is a computer generated waveform but the rest is not much different than amplifier other than being above to generate hundreds of volts (which causes a lot more heat generated). The 9801 is rated at 300 volt-amps. There are higher power AC generators but they get massive in size and weight.

I also have a PS Audio P300 AC regenerator. It is even larger and heavier than the 9801 because it doesn't use a fan. It is much more limited though as you can only change the frequency. I will show its performance but to keep this project manageable, I am only going to show its spectrum.

General Test Protocol
I am going to show you the performance of three different audio products. Each will be powered by generic AC coming out of the wall, then by 9801 AC generator with and without distortion. Each one of these three modes has massively different AC spectrum of noise and distortion. If there is something to AC power making a difference in audio equipment, we will see it.

There are a lot of large dashboard shots here so you may want to review this article on a computer or a tablet. Maybe hard to read them using a phone.

PS Audio P300 and BK Precision 9801 AC Quality Measurements
Let's start as usual with analyzing the spectrum of noise and distortion of AC on my workbench. The source is a 20 amp feed going to a very stout power strip in my equipment rack that powers my workstation, Audio Precision Analyzer, and audio equipment being tested. A shared supply like this helps a bit to keep ground currents lower. Here is what AC looks like from my last review a couple of nights back:

View attachment 145396

We see our mains peak at 60 Hz which is delivering the bulk of the power. With it though we see harmonics of it and noise going all the way up to 90 kHz bandwidth of the measurements. Total distortion+noise is 1.9% and that is enough to visibly distort the waveform where the tops of it no longer look like a nice sine wave. Here is what it looks like with PS Audio P300:

View attachment 145397

Much nicer now. The sine wave looks like a sine wave now. SINAD as a measure of noise+distorion jumps massively from 34 dB to 61 dB. Second harmonic is down to -65 dB which is what mostly determines that figure.

Now let's examine the spectrum of BK Precision 9801:

View attachment 145398

Our sine wave is pure as before. SINAD is improved to 65 dB due to second harmonic dropping a few dBs. There is however mid-frequency noise that is higher than PS Audio P300. Overall, this shows that the P300 is doing its job to produce very clean power.

By the way, I am using a high voltage differential probe to sample the AC power. It divides the voltage by 100. That probe itself has some distortion. I measured that it and it is at or below 0.01%. So it is not making hardly any contribution to the numbers and waveforms you see above. I did not capture its output by it only has a tiny second harmonic spike and nothing else. So our instrumentation is much better than what we are measuring as it should be.

As I noted in the introduction, the BK Precision 9801 has a special feature to create distorted waveforms. What it can do is simulate what comes out of a light dimmer which chops off parts of the AC waveform. This causes the duty cycle to lower and with, produce lower power to the light bulb. I can adjust the amount of "dimming" by degree. Here is what the AC wavform looks like when I set that to 35 degrees (just a random number I picked):

View attachment 145399

Let's agree this is the mother of all dirty AC feeds! Due to step response in the AC waveform, we create infinite harmonics which you can see as a spray in our FFT graph. SINAD has dropped to just 15 dB and THD+N has shot up to 18%. Surely if our audio equipment cares about quality of the AC power, it should produce far lower performance using this type of AC feed. Let's see if this is the case.

Audio Equipment AC Impact Measurements
Let's stay with state of the art audio device because anyone searching for AC mains tweaks is surely also investing in the best audio gear they can. And at any rate, I never see any company advertising that their AC filters/regens are for poor performing audio equipment. Let's start with generic AC coming out of the wall:

View attachment 145400

Superlative performance as we have seen. Distortion is down to -128 dB in the form of third harmonic. Now let' switch to running it on our 9801 lab AV regenerator:

View attachment 145401

No difference at all. We expected this from prior testing of AC cleaners and such. But now let's subject it to the "torture" test of dimmer simulation in 9801 with its 18% THD+N:

View attachment 145402

Not a difference. Zilch. Nada. Clearly the A90 is filtering the AC mains well enough that even a highly distorted AC waveform doesn't bother it.

So people say that maybe a "low quality" audio device cares. Hard to know what this device would be. I looked around and found this old California Audio Labs Sigma tube DAC I had bought in an auction as a donation to our local audiophile society. Let's first run it with generic AC again:

View attachment 145403

Looks like we found our low performing device! It is actually not bad for a tube device but certainly miles different than what we like to see around here.

Let's switch to BK Precision 9801 and see if performance improves:

View attachment 145404

Nothing of note. The tiny bit of variation is due to vagaries of this tube product.

I was just about to switch the 9801 to dimmer mode and the DAC went crazy with its noise floor jumping way up and then down. I played with a bit but could not get it to be stable. So I had to abandon it at this point.

I looked around and found a Loxjie P20 tube headphone amplifier I had bought last year but never tested. Here is how it does with generic AC:

View attachment 145406

We definitely have a distortion factory here. Let's feed this one clean power to see if it makes a difference:
View attachment 145407

Nothing. Everything from distortion to mains hum and general noise remains the same. But maybe the torture test of dimmer simulation does it:

View attachment 145408

Nope! Tiny variations are native to the device. There is no trace of this device caring one bit about quality of the AC mains input.

Conclusions
Up until now we have tested a number of devices that reduce AC noise/distortion only to find them do nothing for the output of audio products. In this little research project, we went the other way, producing very dirty AC feed. Yet three devices from very different origins show zero, and I mean zero, dependency on AC quality. You could argue that we got unlucky with three devices not being sensitive but hard to make a case that they simply did not care.

Of course the explanation is clear: all of these devices first convert mains power to DC and then use it. DC by definition means no variation so filtering is used to remove noise and distortion. Sure, some remains but the rest of the circuit also has immunity to power supply vagaries. By the time we look at at the output of the audio product, we are so, so far away removed from AC that its "fidelity" makes no difference.

With both engineering knowledge and objective measurements backing each other, the conclusion that you don't need to worry about quality of your AC mains is exceptionally strong. This is in the context of fidelity of course where vast majority of these audio tweaks are sold.

-----------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Amir - a couple of folks made an observation about the tube base device tested regarding AC filaments. The Loxjie P20 tube headphone amp is most likely using DC filaments with the tubes in place for likely "visual" reasons. Tube preamps commonly use DC filaments for hum suppression - particularly tube phono preamps due to high gain. Power amps tend to stick with AC filaments - simplicity being one reason and another - the belief that AC filaments "sound better"(I am not promoting that belief - simpler design yes, sounds better needs some proof). If you can get an amp with AC filaments - even better a DHT - Directly Heated Triode - amp with AC filaments, that might be where you could measure a difference. The kind of audiophile with an 8 watt 300B amp with sensitive speakers would be a likely candidate for a power conditioner that actually does something.

 

[pma]

Let me contribute with some facts. The thread name is
Impact of AC Distortion & Noise on Audio Equipment
and it exists and we can measure it, if we know what to do and we do not need a fancy equipment to make measurements. I know from thousands of measurements made on amplifiers that mains line load and frequency base load management (remote control from PWR distributor) do affect amplifier output, depending on its PSR.

An example of pretty usual consumer class AB amplifier measured at 25W/4ohm:

No extra mains load or control

The frequencies in red ellipses can also be detected as increased transformer mechanical noise.

I do not agree with conclusions in post #1.

 

[Labjr]

I have a set of 300B amps. I can hear differences in sound quality at different times of the day. I've always attributed it to noise on the power line.

 

[Helicopter]

Thanks Amir. This would make a great video, but might want 1-2 more devices tested such as something with torroidal linear supply.

The basic concept seems clear to me. (Paraphrasing @JohnYang1997 ) If the design deals with the 50/60Hz itself, then it is probably doing fine with the far weaker harmonics and noise attached to it.

Ripple, DC and ground loop hum are real problems with various solutions including a regenerator, but not the drive for many to buy such things.

 

[peng]

Before discovering ASR, I would often look for advice on other subjectivist forums. It seems that all the items of kit that have been shown by Amir not to make any difference are the same items of kit that would kick off massive ranty threads about whether they made a difference or not. Stuff that does make a difference - e.g. speakers - never seemed to spark much debate. I don’t go on those forums any more, but if I did, I’d now know that if people are arguing fiercely about whether or not something makes a difference, it doesn’t.

People are strange.

Thanks Amir!

Very true but also easy to understand why.. It is not hard to convince most hifi enthusiasts who are not that technically inclined and don't have time to do their research/fact checks. Aside from speakers, we also don't see debates, fierce or not, on the quality of recording/mastering making much more difference than that between a $1000 class AB amp and a $3000 class A amp. Many who would rave about night and day difference heard even just by adding an integrated amp (with may be 1 dB more power output) to their AVRs; and all they listened to for comparison were their favorite typically amplified music recordings, their reasoning would seem totally illogical to some.

 

[vkvedam]

Cool, just reiterates. Anyway, people who really buy these won't believe in measurements and they only trust their ears apparently

 

[Grumpish]

I have a set of 300B amps. I can hear differences in sound quality at different times of the day. I've always attributed it to noise on the power line.

More likely to be variations in the power line voltage, since most tube amps use unregulated. power supplies.

 

[pma]

This is how the sudden change of mains line load is detected at the amplifier output noise voltage. I am just monitoring the real life situation.

 

[sonhouse59]

Wow ! Very good, thx.

 

[Labjr]

More likely to be variations in the power line voltage, since most tube amps use unregulated. power supplies.

I don't think the line voltage varies much. They also use tube rectifiers and old Mullard 5AR4's seem to sound best. So maybe they're sensitive to small changes.

 

[Ata]

Let me contribute with some facts. The thread name is
Impact of AC Distortion & Noise on Audio Equipment
and it exists and we can measure it, if we know what to do and we do not need a fancy equipment to make measurements. I know from thousands of measurements made on amplifiers that mains line load and frequency base load management (remote control from PWR distributor) do affect amplifier output, depending on its PSR.

An example of pretty usual consumer class AB amplifier measured at 25W/4ohm:

View attachment 145450
No extra mains load or control


View attachment 145451

View attachment 145452
The frequencies in red ellipses can also be detected as increased transformer mechanical noise.

I do not agree with conclusions in post #1.

I do not follow, what is your argument here? Are you claiming you can hear the circled ripples at -102dB when the SINAD is -82.4db? Where is the comparison with and without power conditioner, and what equipment is used?

 

[PeteL]

Let me contribute with some facts. The thread name is
Impact of AC Distortion & Noise on Audio Equipment
and it exists and we can measure it, if we know what to do and we do not need a fancy equipment to make measurements. I know from thousands of measurements made on amplifiers that mains line load and frequency base load management (remote control from PWR distributor) do affect amplifier output, depending on its PSR.

An example of pretty usual consumer class AB amplifier measured at 25W/4ohm:

View attachment 145450
No extra mains load or control


View attachment 145451

View attachment 145452
The frequencies in red ellipses can also be detected as increased transformer mechanical noise.

I do not agree with conclusions in post #1.

I am not sure what this mean, what did you do to go from 3rd graph, to second graph to first graph? And what does it say about the power itself being distorted or noisy, You can get figures like third graph in some amps with fully undistorted and noise free power no?

 

[nick-v]

Very interesting. Almost two decades ago I bought a used Sound Application CF-X power line conditioner on Audiogon. It retailed for over $5K, I picked it up used for under $1K. At that time I was a fairly new "audiophile" and I was buying up all sorts of snake oil "tweaks": silver plated copper speaker wires, pure silver interconnects, Bybee AC filters, vibrapods, all sorts of different power cords, etc...

Apparently the Sound Application power conditioners are big with mastering and recording engineers (and Audiophiles ). They definitely offer better protection for your gear than a standard "surge protector", but $5k+ MSRP is pretty big leap if they don't technically improve anything.

I'd be interested to see a test of run of the mill surge protectors and power strips to see if they do any harm? Where is the sweet spot of protecting your gear without buying into the "audiophile tweaks" segment? Also, my understanding is that some of the basic surge protectors and power conditioners limit current and shouldn't be used with power amplifiers?

Panamax and Furman make some nice conditioners under $1K. Or is something like an AC Infinity (or any other brand) power strip with onboard surge protection for under $100 the sweet spot?

 

[Ata]

Looking at some compact AV receiver options today I came across the following 8 most expensive items on "special" (https://klappav.com.au/collections/specials?sort_by=price-descending):

- 2 speaker cables;
- 2 power conditioners;
- a high-end projector;
- 3 top of the line B&W speakers.

Guess which of these are good value for money?

Someone send Amir an IsoTek EVO3 Genesis Ultimate Power Conditioner, please!

 

[Bob from Florida]

I don't think the line voltage varies much. They also use tube rectifiers and old Mullard 5AR4's seem to sound best. So maybe they're sensitive to small changes.

If your 300B amps plate voltage is 500 volts DC and you are using a 5AR4 rectifier your power transformer HV secondary is around 385 VAC or about 3.2 times your mains voltage. If your mains shifts 5 volts the plate voltage will shift 5x3.2x1.3= 20.8 volts of shift - whichever direction. If your amp uses cathode bias this is somewhat compensated by the cathode resistor. Fixed bias amps don't compensate as well. The bias point changes a bit with the power line fluctuations and maybe you are hearing this. This is what happens with an unregulated power supply.

 

[thefsb]

Amir, your commitment to this kind of project is impressive!

I have a nitpick quibble in the first sentence of the conclusion.

Up until now we have tested a number of devices that reduce AC noise/distortion only to find them do nothing for the output of audio products. ...

Of course the explanation is clear: all of these devices first convert mains power to DC and then use it.

Devices that have DC power supplies and audio products in general are two different categories.

Iiuc, some audio products do not first convert mains power to DC. I recall when Linn introduced a power supply for the LP12 with a stable oscillator because the LP12 itself has a synchronous motor. I think synchronous motors are still used in some audio products. Idk, perhaps other audio devices don't use DC power supplies.

 

[tmtomh]

I might suggest that choosing a couple of ****** little tube amps that under the best of circumstances are going to be distortion generators is not the right exemplar of a device to test power supply issues. They are crud no matter what happens, and the bad power is simply unable to make them worse than they already are.

As I wrote recently, there is some case to be made that there is a class of devices that are sensitive. It isn't just a matter of being poor quality. It is is being intrinsically sensitive, so much so that they measure well with good power and degrade with bad. These tube amps don't meet this description.

First cab off the rank would be a traditional class-A amplifier with a conventional transformer and capacitor PS.
Then amplifiers that tout their low or no-feedback design.

In both cases I would expect them to perform reasonably (at least within the bounds of the limitations of their design) with good power, but as a result of poor power supply rejection intrinsic to their design, bad power may show up in the results.

I respect and agree with your desire for rigor and further testing - but while your comment here is clearly made in good faith, the effect of your comment is identical to FUD (spreading fear, uncertainty, and doubt). You have mentioned the poor-performing tube DACs but have completely ignored the fact that @amirm tested them after and in addition to the Topping DAC, which has excellent performance - low noise and low distortion.

So you can't have it both ways: if the tube DACs are too noisy and distorted for power "purifiers" to clean up, then the Topping should have been positively impacted by this test. If the Topping, by contrast, is too clean and robust to be affected by dirty AC power, then the tube DACs should have been.

What you're basically arguing here is that the tube DACs' power supplies are simultaneously (a) too noisy to be improved by power cleaning devices, but also at the same time (b) robust enough in their AC filtering that their measured performance is not worsened by the application of dirtier AC power.

With respect, that's completely illogical. Also, remember that SINAD for the tube DACs was around 81dB for one and around 66dB for the other, while the "torture test" dirty AC had SINAD of about 18dB with massive distortion. The claim that these DACs' power supplies and associated circuitry are too noisy to be iomproved by additional AC filtering, but somehow will not show a single iota of measurable degradation when subjected to an AC feed with 48 to 63dB higher noise and distortion, is not a persuasive claim (to say the least).

 

[Balle Clorin]

With the utmost respect Amir, it appears you are prosecuting a line of reasoning and laying out tests and investigations to lead yourself to a preconceived position. So, no, you are no remotely covering "the full gamut of AC filters/conditioners to regenerators"

I know 1050Hz aberrations on the mains waveform. I can hear them everyday (and night). And
a) they make it through to reproduced audio on many (not all), of my own amplifiers of all different pedigrees, brands and costs. They are all conventional transformer based Class A or A/B designs. And
b) how an inexpensive (relatively) filter can remove the effects of those line related anomalies completely.

So I am 100% confident, as are the thousands of people who buy similar filters, that distortion on the mains (which ripple control is) can be ameliorated by passive and frequency specific filtering. You cannot throw all mains filters in the snake oil basket.

For our state these guys make a specific filter:
https://www.gayrad.com.au/ripple_control_or_off_peak.html

View attachment 145432

With all respect Please , If you can hear it you can measure it and post it here. My oscilloscope below show the haze when the SS rectifier is conducting in a class A tube amp I have. Measured at loudspeaker terminals with loadspeaker connected. Yes I can hear some power supply noise. Hard to distinguish from any mains noise..Cannot hear any noise at all with my present SS amplifier, dead silent with my ear at the tweeter at max volume