Your Power Supply is TOO NOISY

[Ralf11]

https://benchmarkmedia.com/blogs/application_notes/152143111-audio-myth-switching-power-supplies-are-noisy?utm_source=Benchmark's+Application+Notes&utm_campaign=ebfa3d2c6f-Application_Note_2017_8-4-17&utm_medium=email&utm_term=0_7c8c792ee5-ebfa3d2c6f-198220265

[One and a half]

Power Supply comparisons

[Speed Racer]

I tend to think Benchmark wrote that article as a sales tool to justify what they did using a switching power supply. 

[Ralf11]

but are they right or wrong??

[Speed Racer]

7 minutes ago, Ralf11 said:

but are they right or wrong??

 

I don't know. But Vidar, an amp that costs 1/4 of the price of Benchmark's AHB2 with just a bit more power, has a SNR of ">125db, A-weighted, referenced to full output". The AHB2 touts 132db with no reference regarding output.They claim "No power amplifier is quieter than the AHB2."

 

The AHB2 is better, but shouldn't it be for the price? Is that a result of using a switching power supply? Is noise the only drawback of a switching power supply? What does a switching power supply throw out onto the AC mains?

[fas42]

Had a quick look at what they say - a mixture of of dodgy stuff, and genuinely meaningful. Going on about hum and buzz is silly; equipment has to be pretty terrible, these days, for such to be a problem. But helping to prevent sagging of the voltage rails is very important - and well worth pursuing.

 

Like just about everything in the audio game, the concept is barely the start of getting things right - it's the implementation that's all-important; the end result can be brilliant, or appalling, depending on how well all the design considerations were followed through on ...

 

 

 

 

[mansr]

My impression is that John Siau is a competent engineer. The linked post, like most of his, is technically accurate. He obviously promotes Benchmark products, but he's still worth listening to.

[esldude]

1 hour ago, Speed Racer said:

 

I don't know. But Vidar, an amp that costs 1/4 of the price of Benchmark's AHB2 with just a bit more power, has a SNR of ">125db, A-weighted, referenced to full output". The AHB2 touts 132db with no reference regarding output.They claim "No power amplifier is quieter than the AHB2."

 

The AHB2 is better, but shouldn't it be for the price? Is that a result of using a switching power supply? Is noise the only drawback of a switching power supply? What does a switching power supply throw out onto the AC mains?

From the latter part of the article on the AHB2:

 

 The A-weighted  SNR is 132 dB in stereo mode and 135 dB in mono mode. 

 

I think John Atkinson when testing for Stereophile said it was references to 100 watts output.  He measured something like 128 db saying it was getting close to the limits of his Audio Precision to measure.

 

And Benchmark said "to their knowledge" no amplifier had higher SNR.  I believe one of the Bruno Putzeys Mola Mola amps is said to be 128 db unweighted, which likely would be in the same ball park A-weighted as the Benchmark.  It costs several times more though it is several times more powerful as well.  Mola Mola  has an analog pre-amp including phono stage.  It uses analog circuitry, claims an SNR of about 128 db unweighted, and distortion not measurable though estimated at -150 db according to them. It uses a switching PS I do believe. 

 

Switching supplies are not evil noise polluting designs by nature.  Cheaps ones can be, and often are.

[Ralf11]

does this say anything about power supplies for DACs?

[fas42]

Power supplies for DACs  need to be super clean - and high frequency interference anywhere in the picture is death for optimum sound. Switching supplies could do it, but you would have to go to "extreme" lengths to tame all issues.

 

As an example of this, the well considered DEQX DSP correction component came originally with SMPS; a keen audio enthusiast in my area went all the way with this unit, but found the quality lacking - ripped out the switching supplies, and installed "audiophile" linear supplies, himself. Was much happier, and prodded the manufacturer, a local company, frequently about this - I've heard this setup several times, and at its best got many things right. Anyway, the current DEQX products have ditched the switching supplies, and have been much better received, apparently.

[opus101]

3 hours ago, Speed Racer said:

The AHB2 is better, but shouldn't it be for the price? Is that a result of using a switching power supply? Is noise the only drawback of a switching power supply? What does a switching power supply throw out onto the AC mains?

 

It puts out a whole slew of common-mode noise at ultrasonic frequencies. Which I rarely see mentioned in articles praising the merits of SMPSUs (which are undoubtedly real merits).

[opus101]

1 hour ago, Ralf11 said:

does this say anything about power supplies for DACs?

 

What it says to me is - if your DAC has a switching supply you'll do well to ensure it has a balanced output and your amp a balanced in. Use unbalanced and you're likely to run into SQ issues.

[esldude]

3 hours ago, opus101 said:

 

What it says to me is - if your DAC has a switching supply you'll do well to ensure it has a balanced output and your amp a balanced in. Use unbalanced and you're likely to run into SQ issues.

I am surprised balanced hasn't become de rigueur in serious audio gear long ago. 

[opus101]

I have a suspicion that the failure to adopt balanced has something to do with the huge margins available to dealers on unbalanced tweak-o-rama interconnects. Its simply not profitable enough for balanced to become the default IC.

[mansr]

6 hours ago, opus101 said:

It puts out a whole slew of common-mode noise at ultrasonic frequencies. Which I rarely see mentioned in articles praising the merits of SMPSUs (which are undoubtedly real merits).

It doesn't have to, although it's common.

[pkane2001]

 

10 hours ago, opus101 said:

 

It puts out a whole slew of common-mode noise at ultrasonic frequencies. Which I rarely see mentioned in articles praising the merits of SMPSUs (which are undoubtedly real merits).

 

This was posted a few months ago. The one SMPS that I personally use (iFi Power) seems to do well in this comparison (the claim is that it has active noise suppression built in):

 

 

[opus101]

That Elberoth post looks to be concerning a powerline noise meter which measures the normal-mode noise (i.e. between line and neutral), not the common-mode noise. The latter is a much more difficult measurement to make as its context dependent. The next best thing I guess is a normal mode measurement as there's bound to be some degree of correlation between the two.

[pkane2001]

8 minutes ago, opus101 said:

That Elberoth post looks to be concerning a powerline noise meter which measures the normal-mode noise (i.e. between line and neutral), not the common-mode noise. The latter is a much more difficult measurement to make as its context dependent. The next best thing I guess is a normal mode measurement as there's bound to be some degree of correlation between the two.

 

Agreed, it would be a better test to measure noise on the other side of the SMPS. But, this does give a clue of just how much noise is being generated by these supplies, and some of this noise is 'context dependent' as you say. You can see it in the measurements under different load conditions.

[Kelly]

Hasn't Linn moved to SMPS or their fancy streamer/dacs? I think NAD has moved into them for their Master Series stuff. The Kii Three active speaker also has an SMPS. I haven't heard anyone complaining that any of these products are "noisy." 

[jtwrace]

I think he's right.  Bruno Putzeys (Hypex and Kii) and Tom Christiansen (Neurochrome) would probably agree.

 

https://www.neurochrome.com/

[firedog]

You can make a low noise SMPS if you have the knowledge and want to spend the money. It ain't cheap. 

Linn can do it b/c they sell expensive equipment. I assume Hypex and Kii can do it because they sell in large volumes relative to the audiophile world. So not difficult to make up the investment/development/parts cost. 

 

As far as noisy, I don't think the issue would be that you are hearing noise - it would be is the PS making some type or level of noise that makes your setup less revealing, sound not as good, etc.?

 

I don't think you could really determine that unless you somehow found a way to power these units with a quality non-SMPS supply and compared. 

[opus101]

6 hours ago, firedog said:

As far as noisy, I don't think the issue would be that you are hearing noise - it would be is the PS making some type or level of noise that makes your setup less revealing, sound not as good, etc.?

 

 

Precisely. The subjective effects of ultrasonic noise ingress are often harshness and annoyance (in large quantities) - at moderate quantities sibilance and loss of dynamics and spatial cues.

[jabbr]

On 8/4/2017 at 7:08 PM, Ralf11 said:

but are they right or wrong??

The area of improvement in switching power supplies is active. Often a good regulator & prefilter is able to provide a low noise output e.g the iFi iPower but generally agree that you get what you pay for.

[gmgraves]

On 8/4/2017 at 4:25 PM, Speed Racer said:

 

I don't know. But Vidar, an amp that costs 1/4 of the price of Benchmark's AHB2 with just a bit more power, has a SNR of ">125db, A-weighted, referenced to full output". The AHB2 touts 132db with no reference regarding output.They claim "No power amplifier is quieter than the AHB2."

 

The AHB2 is better, but shouldn't it be for the price? Is that a result of using a switching power supply? Is noise the only drawback of a switching power supply? What does a switching power supply throw out onto the AC mains?

 

Since both are below the threshold of human hearing, I would think that the difference between -125 and -132 dB, irrespective of what weighting would be purely academic. BTW, "A" weighting is fairly flat from 1KHz to 10KHz and falls-off above that making it most suitable for measuring S/N in the audible band. 

[esldude]

A weighting curve.  A combination of more or less mimicking human hearing response and being fairly simple to build into meters way back when.

 

 

Fletcher Munson curve.  If flipped upside down not so different than A weighting.