• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Class D principle, audible "higher frequency noise"

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
Well, a SNR at 80 dB is terrible. Most dacs are way over a 100 dB, so that points to the amp. As far as resolution less than 14 bit

Don't conflate SNR (signal to noise ratio) and resolution in bits- they are unrelated.

An amplifier with audible hum can still achieve a S/N well over 120dB with enough power on tap.
 

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
Don't conflate SNR (signal to noise ratio) and resolution in bits- they are unrelated.

An amplifier with audible hum can still achieve a S/N well over 120dB with enough power on tap.
This is why I have always wanted to see a risidual noise measurement in uV in @amirm reviews. It provides an absolute indication of how noisy an amp is. Can be very relevant especially with more sensitive speakers.
 

tmtomh

Major Contributor
Forum Donor
Joined
Aug 14, 2018
Messages
2,635
Likes
7,486
This is why I have always wanted to see a risidual noise measurement in uV in @amirm reviews. It provides an absolute indication of how noisy an amp is. Can be very relevant especially with more sensitive speakers.

Yes , that would be great. So many otherwise quiet amps have it at audible levels.
 
OP
D

donsna

Member
Joined
Jul 18, 2020
Messages
11
Likes
3
I have received/read in another forum following statement:
"If you listen close to the tweeter (within a feet), all variants of Class-D amps have the so called switching noise. That is normal and does not affect the listening experience. The noise level is dependent on the speaker's efficiency and tweeter."
The statement was posted by someone who is not really independent (strongly connected to a company selling good amps, but producing this kind of noise).

According to what I have experienced and learned so far, these sentences are not correct, they would be correct if they were modified e.g. as follows:
"If you listen close to the tweeter (within a feet), several variants of Class-D amps have the so called switching noise. That is normal for some manufacturers and if it does affect the listening experience, everybody has to judge for themselves. The noise level is dependent on the speaker's efficiency and tweeter. But: There are also absolutely quiet class D implementations out there..."

My Lyngdorf e.g. is absolutely quiet, so I can confirm that there are quiet class D implementations out there (quiet means in this case: no audible noise, I am not talking about noise-measurements), and several members confirmed and reported this here also: sure there are some class D implementations producing this noise, but others don't!
I just like to correct things, because there are several people out there searching for useful information in forums... and I don't like if someone with strong relation to a manufacturer tells them "fairy tales", so at least these people can find correct information here.
But of course someone may correct me, in the case that I am wrong.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
Same FUD-spreading disinformation you got before. Quit reading that other forum.
 

Killingbeans

Major Contributor
Joined
Oct 23, 2018
Messages
4,088
Likes
7,544
Location
Bjerringbro, Denmark.
According to what I have experienced and learned so far, these sentences are not correct, they would be correct if they were modified e.g. as follows:
"If you listen close to the tweeter (within a feet), several variants of Class-D amps have the so called switching noise. That is normal for some manufacturers and if it does affect the listening experience, everybody has to judge for themselves. The noise level is dependent on the speaker's efficiency and tweeter. But: There are also absolutely quiet class D implementations out there..."

It has nothing to do with the switching noise. It usually sits around 500KHz or higher. Even if a tweeter could output that, there's no way your ears would be able to pick it up. All amps, no matter class, has residual noise on the output, and sometimes it's audible. It's not a Class-D specific phenomena.
 
OP
D

donsna

Member
Joined
Jul 18, 2020
Messages
11
Likes
3
It has nothing to do with the switching noise. It usually sits around 500KHz or higher. Even if a tweeter could output that, there's no way your ears would be able to pick it up. All amps, no matter class, has residual noise on the output, and sometimes it's audible. It's not a Class-D specific phenomena.
Yes, you are right. And the word "switching" within the phrase "... the so called switching noise...", is an important detail within this context.
And as far as I know (at least up to now, I am learning right now every day a bit more about audio-technique ;-) ), the word "switching noise" is common within the context of "switching regulators" (and not when talking about noise audible at the tweeter).
There can be found quite good explanations in the internet explaining the "switching noise", explaining the fact that you mentioned:
The "switching noise" introduced at the output would have the same frequency as the switcher (and some harmonics of that frequency, means even higher frequencies).
So you are right, it's not correct to call an audible noise at the tweeter "switching noise" in the context of class D amplification, because the switching-frequencies are much higher.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
it's not correct to call an audible noise at the tweeter "switching noise"

Just call it residual noise. Specify a bandwidth, weighting and other conditions (shorted input/volume max etc) and specify/measure it in micro Volts (uV). That's the way it was done forever.
 

Panelhead

Senior Member
Forum Donor
Joined
Apr 3, 2018
Messages
348
Likes
137
Just call it residual noise. Specify a bandwidth, weighting and other conditions (shorted input/volume max etc) and specify/measure it in micro Volts (uV). That's the way it was done forever.


The output noise is always there. The lack of an established standard for comparison prevents comparing these numbers. I see numbers specified for many recent amplifiers. Usually in double digit uV.
This is very quiet.
If the analyzer has enough resolution, there is measurable noise at every point in the system, thermal noise level.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO

PierreV

Major Contributor
Forum Donor
Joined
Nov 6, 2018
Messages
1,437
Likes
4,686

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
By linear extrapolation, we can deduce that in a few years we'll have a new class of integrated devices that combine a wifi router, a micro-wave oven and an amplifier. :cool:

Pretty sure that is non-linear extrapolation... :)
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
The output noise is always there. The lack of an established standard for comparison prevents comparing these numbers.

You don't need an "established standard" if the conditions are stated. That said, shorted input- gain wide open over a 20KHz bandwidth either weighted or not, is pretty much the defacto standard.

I measure noise with a shorted input (or 5.1K if DC coupled on the input), unweighted, over a 200KHz bandwidth on analog amplifiers, with or without a 400Hz filter (for hum and harmonics).
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
3,335
Likes
5,049
So this does beg the question: Why are a lot of class D amps exceptionally noisy? What about the nature of the architecture makes them prone to noise?
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
So this does beg the question: Why are a lot of class D amps exceptionally noisy? What about the nature of the architecture makes them prone to noise?

What amps do you have in mind when you say "lot"? IIRC, most of the ones Amir has measured exhibit very low noise.

The chart has a lot of class D amplifiers at eh far left (best SINAD) side... https://www.audiosciencereview.com/...iver-review.14946/#lg=attachment75333&slide=0

What we don't have AFAIK is a chart showing the actual residual noise at the outputs with inputs shorted, but you can infer reasonably low noise if SINAD is high.
 
Last edited:

VintageFlanker

Major Contributor
Forum Donor
Joined
Sep 20, 2018
Messages
4,939
Likes
19,689
Location
Paris
During the last weeks, I could test two different amplifiers from "Nuprime" (STA-9 and ST-10). Somehow I like them, but I had all in all three different noise issues.
To make things not to confusing, I want to focus here only on one of the three noises:
If one of the amplifiers is connected to the speakers and powered (means: in state "ON"), there is an audible noise in the higher frequency range.
Had the exact same issue with my previous STA-9s bridged. NOT with all my amps since (Class AB, Ncore, IcePower). Class D has nothing to do with that.
 

dfuller

Major Contributor
Joined
Apr 26, 2020
Messages
3,335
Likes
5,049
What amps do you have in mind when you say "lot"? IIRC, most of the ones Amir has measured exhibit very low noise.

The chart has a lot of class D amplifiers at eh far left (best SINAD) side... https://www.audiosciencereview.com/...iver-review.14946/#lg=attachment75333&slide=0

What we don't have AFAIK is a chart showing the actual residual noise at the outputs with inputs shorted, but you can infer reasonably low noise if SINAD is high.
Sorry, I should've specified - most of my experience is with class D chip amps in active speakers, almost all of which have much higher residual noise (no signal applied, balanced connections) than class AB or G amps in the same situation.
 

Killingbeans

Major Contributor
Joined
Oct 23, 2018
Messages
4,088
Likes
7,544
Location
Bjerringbro, Denmark.
So this does beg the question: Why are a lot of class D amps exceptionally noisy?

Sorry, I should've specified - most of my experience is with class D chip amps in active speakers, almost all of which have much higher residual noise (no signal applied, balanced connections) than class AB or G amps in the same situation.

My bet would be cost/profit margin. I can imagine that most (low cost?) active speakers use whatever of-the-shelf chips the designer can find that fulfill some completely different criteria than low noise. These chips are probably made for other uses, like cheap portable comsumer electronics, and while proper care for the implementation could most likely lower the noise, it simply wasn't deemed necessary.

What about the nature of the architecture makes them prone to noise?

Nothing really. It's all about implementation and cutting corners.
 
OP
D

donsna

Member
Joined
Jul 18, 2020
Messages
11
Likes
3
So this does beg the question: Why are a lot of class D amps exceptionally noisy? What about the nature of the architecture makes them prone to noise?
I don' think that class D amps are generally exceptionally noisy (like explained by some other members here).
The only "special" noise I know caused by the class D principle is the "switching noise", but this noise has the same frequency as the switcher or its harmonics (e.g. for the Nuprime STA-9 this should be e.g. 550kHz, 1100kHZ, ...), means: far outside the audible range. And anyhow, (at least theoretically) this noise should be filtered off before the signal is sent to the speaker drivers (at least I hope so ;)).
If we are talking about audible noise: I am using Lyngdorf TDAI 2170, I also had for some days a TDAI 3400, and for some month I used Dynaudio Focus XD active speakers. All of them class D, all of them absolutely quiet (quiet means "no audible noise" in this case), so there are Class D implementations out there which do not produce audible noise. But of course noise has many different types and possible root causes...
 
Top Bottom