On Class D Amplifiers Measurements

[March Audio]

@ 920kHz -104dBV = 0.00000631V = 6.3uV = -110dB opposite 2V
Consider that goes through an input filter that lowers it even more.
at 3FS this is higher but the input filter (certainly the one MA is using) will have attenuated this.
What is your point ?

Indeed, we have already been through this with his other plots.

 

[restorer-john]

@ March Audio Is it possible that you can stop significantly editing your posts after the fact to present information or data that wasn't present when people reply to your posts? It's just not cool.

Add your new information as a new post.

 

[pma]

Seriously, I'm not calling your testing FUD, you know that! I applaud your efforts wholeheartedly.

John, this was a misunderstanding, I was commenting on March post in a reply to you.

I am not interested in fights primarily, I just want all aspects to be considered and not refused as unrealistic etc.

 

[March Audio]

OK, same P122 amp output with my DAC1 connected (single ended).

EDIT: Putting the Motu on the same page for comparison (balanced).
Excuse the different number of averages between the two.

 

[solderdude]

The input RC filter is to be tuned somewhere above 120kHz/-3dB, otherwise the phase reponse is affected in the audio band.

View attachment 40474

Of course, that's exactly my point.
To get a similar flat audio range (phase and amplitude) using digital (with steeper filters) the BW of a digital signal (amp) can be much smaller.

Of course when we look at class-D amps the phase response (as well as upper FR with some designs and deviating loads) may well not be that good.
On the other hand we can safely same the exact same thing of many tube and a lot of not so well designed SS amps is it not ?

Don't get me wrong... your concerns are valid and feel we should measure the effect of the input filter (and output spectrum).
I fully agree class-D may need additional measurements as the technique used differs from analog.

I also agree with MA to not turn it into a which hunt and measure the values only. Those that feel it reaches audible levels (depends on so many aspects) can then make a more educated decision.

This thread looks more like a pissing contest between folks with an iron in the fire. (not directed to you specifically)

 

[March Audio]

John, this was a misunderstanding, I was commenting on March post in a reply to you.

I am not interested in fights primarily, I just want all aspects to be considered and not refused as unrealistic etc.

It is unrealistic.

Pavel you need to perform these tests with real dacs connected to an amp in a normal way otherwise it is totally artificial. If you try hard enough you can of course create problems.

You cant just post a plot of a DAC and assume its going to create an issue. The same with external RF. Being a "possibility" is not the same as indicating a real problem

To make out this is any kind of real world issue is just spreading FUD.

 

[pma]

Yes I got that... but my question is how much attenuation of that signal happens by the input filter of the class-D module that receives these frequencies.
How much of this will actually fold back as audible / measurable content.
(I am just asking not trolling or wanting to know better)

Say you connected mentioned DAC and measured the actual analog output 0-20kHz with the mentioned device merely switched on and off what level of change in the audible band do you see ?

1) only investigation will tell about audibility in class D case.
2) I measured the DAC switched on and with -100dBFS signal.
My experience with CD players is that sound differences were mostly linked to their HF output spectrum and to immunity of the following audio components. Output spectrum tends to differ also depending on wiring scheme use, instrument class (I, II), possible isolation transformers used, component design, case design HF rules, this is all very complex. That's why so many discussions. If everything sounded same and was simple there would be no reason for audio fora and discussion on sound. I know that sometimes there is no limit in assumptions what affects the sound, but to find a solid boundary is very difficult. I can make you sure that that SINAD 1kHz/5W plots and comparative charts do not tell the story.

 

[March Audio]

1) only investigation will tell about audibility in class D case.
2) I measured the DAC switched on and with -100dBFS signal.
My experience with CD players is that sound differences were mostly linked to their HF output spectrum and to immunity of the following audio components. Output spectrum tends to differ also depending on wiring scheme use, instrument class (I, II), possible isolation transformers used, component design, case design HF rules, this is all very complex. That's why so many discussions. If everything sounded same and was simple there would be no reason for audio fora and discussion on sound. I know that sometimes there is no limit in assumptions what affects the sound, but to find a solid boundary is very difficult. I can make you sure that that SINAD 1kHz/5W plots and comparative charts do not tell the story.

What evidence do you have to back up your experience?

 

[maty]

Several comments back have commented on the ability of the human ear to hear or not beyond 20 kHz. Tannnoy published a white paper many years ago, 2000 year I think, about dual concentric speakers, supertweeter and... phase too.

BTW, my loved KEf Q100 5.25" coaxial, at -3 db: 49 Hz - 40,000 Hz

The band above 20 kHz is for many the equivalent of junk DNA. We now know that it is also important and its study should not be neglected.

- End off topic -

 

[pma]

It is unrealistic.

Pavel you need to perform these tests with real dacs connected to an amp in a normal way otherwise it is totally artificial. .

But I do . I am the proponent of measuring of complete audio chain, including all wiring, grounding scheme. Why would you think I do not? I ask for my system to work flawlessly with 2 x 10m cables from room to room. Depending on wiring and interconnection scheme, there are huge differences, bigger than between components alone.

 

[March Audio]

But I do . I am the proponent of measuring of complete audio chain, including all wiring, grounding scheme. Why would you think I do not? I ask for my system to work flawlessly with 2 x 10m cables from room to room. Depending on wiring and interconnection scheme, there are huge differences, bigger than between components alone.

View attachment 40482

View attachment 40483

Pavel these plots have absolutely nothing to do with the specific issue in hand and what we are discussing.

I think it because its exactly what you have done on this specific issue. You have extrapolated an unrealistic and unrepresentative technical bench test to imply its something that happens in real life.

 

[solderdude]

I can make you sure that that SINAD 1kHz/5W plots and comparative charts do not tell the story.

That's why one always should judge a device based on all available measurements and not just one aspect and that the right things need to be measured in the proper way.

Just routing an interlink cable or measurement cable or introducing a common mode current here and there can skew measurement results.

I am fully aware that users of audio equipment at home, under vastly different circumstances, sources and loads are often responsible for different impressions of gear and issues people have.
On the other hand the vast majority of 'subjective and non blind, incorrectly, or not controlled 'listening tests' and findings are completely erroneous because of the way it is tested.

Measurements are just that, performed under prescribed and verified conditions. These may differ from reality as well as due to product and application variances.

 

[March Audio]

Several comments back have commented on the ability of the human ear to hear or not beyond 20 kHz. Tannnoy published a white paper many years ago, 2000 year I think, about dual concentric speakers, supertweeter and... phase too.

BTW, my loved KEf Q100 5.25" coaxial, at -3 db: 49 Hz - 40,000 Hz

The band above 20 kHz is for many the equivalent of junk DNA. We now know that it is also important and its study should not be neglected.

- End off topic -

Matey, this is the same data that was referenced earlier in the thread. Note the caveats from the original paper. note that only 3 of the 16 electroencephalograms were provided for study and that activity was going on a long time after stimulus yet still attributed to ultrasonic stimulus. Has the data from that study been reproduced in other studies?

So no we dont "now know".

 

[March Audio]

That's why one always should judge a device based on all available measurements and not just one aspect and that the right things need to be measured in the proper way.

Just routing an interlink cable or measurement cable or introducing a common mode current here and there can skew measurement results.

I am fully aware that users of audio equipment at home, under vastly different circumstances, sources and loads are often responsible for different impressions of gear and issues people have.
On the other hand the vast majority of 'subjective and non blind, incorrectly, or not controlled 'listening tests' and findings are completely erroneous because of the way it is tested.

Measurements are just that, performed under prescribed and verified conditions. These may differ from reality as well as due to product and application variances.

We have 2 issues here.

1. Do DACs produce enough RF at the "wrong" frequencies to intermodulate with the switching frequency and create in band products and spuria.

2. Does external RF get into a system at high enough levels to do the same?

1 is easy to test for. Just plug a DAC in and look at the noise floor.

2 is not easy easy to test for unless you have an RF test chamber. Probably best we can do is be near a real AM broadcast transmitter. However if you measure the noise floor problems will be indicated. You are going to need to be very close to a tx, and one broadcasting at the right (wrong) frequency before this becomes an issue. In which case it will probably be a problem for any type of amp.

 

[Wombat]

General comment:

 

[restorer-john]

General comment:

Clearly there's some serious loss of detail in the top end with that clip. I'm sorry, it's not representative and must be dismissed as a valid comment in this thread....

 

[pma]

OK, unrealistic and FUD. There are thousands of players and DACs and they behave very differently above audio band. This is Trend DV-821A DVD player, just "ON". A good candidate for my next review.

Nice SMPS throughput

 

[March Audio]

OK, unrealistic and FUD. There are thousands of players and DACs and they behave very differently above audio band. This is Trend DV-821A DVD player, just "ON". A good candidate for my next review.

View attachment 40484
Nice SMPS throughput

View attachment 40485

Yep. Now plug it into an amp and demonstrate it causes problems.

So again you post a plot which contradicts your assertion that 10mV is a representative and realistic noise level. You have levels at the switching frequency of about -90dBV which is 31 uV.

Have to ask, that hardly seems a quality player. Whats the 60kHz and harmonics all about?

 

[miero]

Has anybody tried to play DSD? I guess following graph is without output filter, but nevertheless there will be much more HF content than with PCM.

Source: https://positive-feedback.com/Issue66/dsd.htm

 

[solderdude]

Do DACs produce enough RF at the "wrong" frequencies to intermodulate with the switching frequency and create in band products and spuria.

is easy to test for. Just plug a DAC in and look at the noise floor.

There are also DAC's that may produce US crap all the way up there when music is playing. Think of those non filtered NOS DAC's that have no filtering whatsover (because filters are baaaaaadd) that may produce crap all the way up there that may fold back.
Then again ... such individuals deserve to be be punished

Anyways... when a swicthing amp works on a high enough frequency and has proper input filtering on most situations I don't think it will lead to any practical issues.