• 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!

How does ultrasonic noise affect the sound quality?

asr6576

Member
Joined
Feb 18, 2020
Messages
8
Likes
5
Except for the NOS mode, the DACs always apply a digital low-pass filter to clean the sound, especially for the DSD playback. Does that mean ultrasonic noise is audible?
Rob Watts from Chord claims that RF noise makes sound warm and bright. Is it the truth or another hype?
My personal experience is that a noisy amplifier does sound warm and bright, but such noise is not solely ultrasonic noise.
I see amirm posts ultrasonic performance for power amplifiers but not for DACs or headphone amplifiers. I'd like to know the reason.
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
"Warm and bright"? Hmmm... In any event RF noise should only affect the sound if something in the signal chain is not properly rejecting it. If it is making significant audible changes I'd guess it is being mixed or rectified someplace and converted down to the audio (audible) range.

IMO the main reason for testing the ultrasonic performance of an amplifier is to makes sure it does not act up when presented with such an input, either from a wideband analog source, higher sampling rate DAC with high filter cutoff, or insufficiently suppressed DAC images.

FWIWFM - Don
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,281
Location
Oxford, England
"Warm and bright"? Hmmm... In any event RF noise should only affect the sound if something in the signal chain is not properly rejecting it. If it is making significant audible changes I'd guess it is being mixed or rectified someplace and converted down to the audio (audible) range.

IMO the main reason for testing the ultrasonic performance of an amplifier is to makes sure it does not act up when presented with such an input, either from a wideband analog source, higher sampling rate DAC with high filter cutoff, or insufficiently suppressed DAC images.

FWIWFM - Don


Isn't there a chance that noise-shapping (i.e. DSD) and wide-bandwidth amplifiers (i.e. Spectral) and hard-dome tweeters with 10 or 20dB peaks at 25 or 30kHz (i.e. BnW CM5) might act up? (extreme examples)
 

DonH56

Master Contributor
Technical Expert
Forum Donor
Joined
Mar 15, 2016
Messages
7,834
Likes
16,496
Location
Monument, CO
Isn't there a chance that noise-shapping (i.e. DSD) and wide-bandwidth amplifiers (i.e. Spectral) and hard-dome tweeters with 10 or 20dB peaks at 25 or 30kHz (i.e. BnW CM5) might act up? (extreme examples)

Not that I can see. Noise shaping pushes the noise band well above that, and then it is filtered, so little energy reaches the output above the audio band. Noise shaping produces a rising noise response, true, but there is still a low-pass filter that rolls off the noise. DSD also requires an anti-image filter, just at a higher frequency, but again that higher frequency is well below the noise hump. Ultrasonic energy should be in the mud for any competently-designed DAC, and Amir's wideband measurements seem to agree.

I'd be more worried about the images from non-filtered "NOS" DACs that produce wideband tones and noise. Too much marketing, too much ignoring of fundamental engineering, in my mind.
 
Last edited:
OP
A

asr6576

Member
Joined
Feb 18, 2020
Messages
8
Likes
5
Not that I can see. Noise shaping pushes the noise band well above that, and then it is filtered, so little energy reaches the output above the audio band. Noise shaping produces a rising noise response, true, but there is still a low-pass filter that rolls off the noise. DSD also requires an anti-image filter, just at a higher frequency, but again that higher frequency is well below the noise hump. Ultrasonic energy should be in the mud for any competently-designed DAC, and Amir's wideband measurements seem to agree.

I'd be more worried about the images from non-filtered "NOS" DACs that produce wideband tones and noise. Too much marketing, too much ignoring of fundamental engineering, in my mind.

NOS means no digital filter but still could have an analog low-pass filter as I guess?
 

dmance

Member
Forum Donor
Joined
Sep 6, 2019
Messages
9
Likes
11
Location
Burlington, Ontario, Canada
My company (www.audiowise.ca) focuses on understanding the effects of RF noise on a DACs transparency. Of course, the in-band audio output from DACs does not show the clearly audible effects we hear. The increase in transparency, the detail, the nuance, the staging ...all that comes and goes as we make changes to input, power, cabling, etc. I've tried to look for changes in the in-band audio output with an APX-525 and could measure nothing. Rob Watts says the effects are well below the noise floor of such measurement equipment. Its all impossibly ridiculous that we can hear benefits but i know we can.
Regarding out-of-band-noise from a DAC ...well they are full of it ...all the way to 4Ghz. The primary outputs from my test DACs from Chord Electronics are full of energy down to around -60dB. I used a Signal Hound BB60 to measure it and custom scripts to analyse my various tests. I can detect when a galvanic input vs optical is used. I can detect when a noise PC vs a quiet PC is being used. More importantly, I can detect when a RF noise electronic device is nearby.
I use Out-of-Band noise as a marker for sound quality. I don't have a quantifiable definable metric for how SQ has been modified ...but its mere presence at the output means its made its way into the DAC all the way to the output. When the sound changes, it has to be harmonics or a bump in noise floor.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
NOS means no digital filter but still could have an analog low-pass filter as I guess?

It could, and when it would it should be a very steep one, and also shift in frequency depending on the sample rate.
Switched capacitor comes to mind. Haven't seen actual implementations though perhaps someone made one.
You could also make a very steep filter at the lowest sample rate (44.1kHz) but then the 'fun' point of NOS, which seems to be a perfect squarewave production (let's NOT call it reproduction) would be down the drain again.
NOS (Non-Over-Sampling) = BAD but thanks to our limited hearing range and poor capabilities of speakers to 'follow' those signals it still sounds 'good' much in a sense that vinyl also sounds good. Poor technical reproduction but good enough for the poor brain and bandlimited auditory system.
 

JeffS7444

Major Contributor
Forum Donor
Joined
Jul 21, 2019
Messages
2,347
Likes
3,507
I once owned a crappy tube amplifier which used wide-bandwidth 5755 input tubes. Due to error in PCB layout, there was no grid stopper resistor to limit it's bandwidth, and as a result, frequency response had a large resonant point above 100 kHz. As delivered, the amp's 10 kHz square wave output into a resistive load looked like this:
R0013881.jpg

Suffice to say, it sounded horrible (but the amp did receive at least one positive review from an obscure audiophile site!)
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,422
Likes
4,029
Location
Pacific Northwest
Except for the NOS mode, the DACs always apply a digital low-pass filter to clean the sound, especially for the DSD playback. Does that mean ultrasonic noise is audible? ...
Ultrasonic noise isn't audible, but it may trigger downstream components to produce audible distortion. That is, IMD may be introduced by the preamp, power amp, or speakers that are downstream.
For example, we can't hear 23 kHz or 22 kHz, but if both signals are present in the output, their difference 1 khz may be created by IMD in downstream components.
 

NakedRider

New Member
Forum Donor
Joined
Apr 7, 2020
Messages
4
Likes
3
I see amirm posts ultrasonic performance for power amplifiers but not for DACs or headphone amplifiers. I'd like to know the reason.

It is common to do this when measuring a class D amplifier (many power amps use this topology) due to the spikes you typically get near the switching frequency of the power stages (in the few hundreds of kHz usually). I do not think Amir measures this for audibility reasons.
 

scott wurcer

Major Contributor
Audio Luminary
Technical Expert
Joined
Apr 24, 2019
Messages
1,501
Likes
2,821
I once owned a crappy tube amplifier which used wide-bandwidth 5755 input tubes. Due to error in PCB layout, there was no grid stopper resistor to limit it's bandwidth, and as a result, frequency response had a large resonant point above 100 kHz. As delivered, the amp's 10 kHz square wave output into a resistive load looked like this:
View attachment 63049
Suffice to say, it sounded horrible (but the amp did receive at least one positive review from an obscure audiophile site!)

Unless no one noticed it's also oscillating.
 
Top Bottom