• 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!

SINAD Measurements

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
Hi,
A few brands display the THD for the range 20 Hz to 20 kHz for their amplifiers. How can it be compared to the THD measured at 1 kHz ?

Also, for those products, why is the THD value under @1W lower than the THD value @50W? On Amir's charts, the curve goes down with the Watts before clipping.

In answer to the first question, THD at 1kHz is usually lower than for the full bandwidth. A device conservatively specified will spec the distortion over the full bandwidth, in the knowledge that that is the worse-case. Marginally specced products quote at 1kHz only, and you can only wonder what they're hiding elsewhere.

As to the second question, Amir's measurements aren't pure THD, but THD+N, that is, including noise. As the power (and hence voltage) level falls, noise becomes a larger proportion of the total, so the THD+N figure gets worse. With many (analogue) products, pure THD, i.e. excluding noise, should fall as power levels fall, or at least not increase. If it does increase, that points to low level issues like crossover distortion.

S.
 

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
3,687
Likes
4,068
In answer to the first question, THD at 1kHz is usually lower than for the full bandwidth. A device conservatively specified will spec the distortion over the full bandwidth, in the knowledge that that is the worse-case. Marginally specced products quote at 1kHz only, and you can only wonder what they're hiding elsewhere.

As to the second question, Amir's measurements aren't pure THD, but THD+N, that is, including noise. As the power (and hence voltage) level falls, noise becomes a larger proportion of the total, so the THD+N figure gets worse. With many (analogue) products, pure THD, i.e. excluding noise, should fall as power levels fall, or at least not increase. If it does increase, that points to low level issues like crossover distortion.

S.

Thanks a lot.
What would you think of an amplifier which has 1% THD between 20 Hz and 20 kHz ? 0.1 % ? 0.01 ?
 
OP
March Audio

March Audio

Master Contributor
Audio Company
Joined
Mar 1, 2016
Messages
6,378
Likes
9,317
Location
Albany Western Australia
For thd measurements you must know the bandwidth over which it is measured.

For a 1kHz fundamental if the measurement bw is 20kHz you could include 19 harmonics. However if your fundamental is 6 kHz you will include 2 harmonics, at 12 kHz no harmonics.
 

sergeauckland

Major Contributor
Forum Donor
Joined
Mar 16, 2016
Messages
3,440
Likes
9,100
Location
Suffolk UK
Thanks a lot.
What would you think of an amplifier which has 1% THD between 20 Hz and 20 kHz ? 0.1 % ? 0.01 ?

If that's a tube amplifier, it's OK, for a Solid State amplifier very poor. As to what the 1kHz figure would be, it's impossible to tell. For a tube amplifier, possibly 0.1%, but that's a guess and it could be a lot worse. For a SS amplifier, however, output power is often quoted at 1% THD, as that's just into clipping. Come down a little on power so out of clipping and distortion should drop a lot lower, 0.01% isn't unusual these days.

When I measure amplifier distortion, I always measure just below clipping, and as long as that's low, and the power output close to (or exceeds) the rated power, then I'm happy the amp is performing adequately.

My THD analyser uses a 100kHz bandwidth so I can capture enough harmonics for the measurements to be meaningful.

S.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Hi,
A few brands display the THD for the range 20 Hz to 20 kHz for their amplifiers. How can it be compared to the THD measured at 1 kHz ?
Since their range includes 1 Khz, then our 1 Khz results can be compared to their number. I also run THD+N versus frequency although its bandwidth is much higher than the manufacturers tend to use (so shows more pessimistic results).
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Also, for those products, why is the THD value under @1W lower than the THD value @50W? On Amir's charts, the curve goes down with the Watts before clipping.
They may be using filtering for the 1 Watt and/or letting the max power go well into clipping. I stop the moment distortion starts to rise in my power graphs.
 

DonH56

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

jasonhanjk

Active Member
Joined
Mar 24, 2019
Messages
252
Likes
180
Headphone amplifier S/N should be referenced to their specified rated output level, not their maximum output achieved.

Agree to this statement. However I like to add that nobody in their right mind would want to listen to their headphones at 1W power, even 100mW will ensure deafness. Headphone amp should measure from 1mW to the max of 10mW; or power level predetermine by the target headphone's target SPL level.
 

Juwieh

Member
Forum Donor
Joined
Sep 22, 2019
Messages
22
Likes
22
Location
Germany
Since their range includes 1 Khz, then our 1 Khz results can be compared to their number. I also run THD+N versus frequency although its bandwidth is much higher than the manufacturers tend to use (so shows more pessimistic results).
I understand that the single SINAD figure you publish in the bar charts scross all DACs are simply the 1 kHz SINADs. Is that correct?

If so, wouldn't it be much more appropriate to take the minimum SINAD (or maximum THD+N) from the THD-N vs frequency measurements, or an avarage across the frequency range? I mean, what's a high SINAD at 1 kHz good for if there's a poor SINAD @ e.g. 4 kHz?
 

anmpr1

Major Contributor
Forum Donor
Joined
Oct 11, 2018
Messages
3,722
Likes
6,405
I came across an interesting article on the audibility of distortion, 05/73 Stereo Review. It might have been the first of the "Carver Challenges", but with a different focus. We are familiar with Bob's exploits at Audio Critic (cloning the transfer function of the Mark Levinson ML-2) and Stereophile (cloning the transfer function of a Conrad Johnson tube amp). In this test Bob and the editors wanted to test audibility of distortion using a) test tones; b) music. Participants were Julian Hirsch, Craig Stark, Larry Klein, and Ralph Hodges. The test was not double blind, but a 'critical listening' session. I will list the 'findings' as they were presented.

1) the Phase Linear 400 was compared to 'an excellent 300 watt amplifier made by a competitor'. The brand is not mentioned, but from a photo, along with the recent power amp comparison SR did a couple issues earlier, it is clear the amp was a Crown DC-300. The Phase had been modified to include a switch that would 'inject' pre-measured amounts of crossover distortion into the output, which was fed into a monophonic (single) AR-LST speaker.

First was a note on level matching:

When we were finally able to get the output levels of the two power amplifiers matched, there was absolutely no audible difference when switching between them while listening to either white noise or music. During adjustments...it was demonstrated dramatically that minute differences in volume level (sound quantity) ...are interpreted by the ear as 'obvious' differences in sound quality. Everyone was startled by this effect...

2) fed by an HP generator delivering a 60Hz tone, and driven to 1 watt, the amplifier was stepped up to 0.05%; 0.07%; 0.1%; 0.15%; 0.2%, at which point distortion became audible. At 0.7% the distortion became a separate buzz on top of the 60Hz tone. Increasing power output to 4 watts increased hearing sensitivity of distortion to 0.15%.

3) mixed 60Hz and 7KHz tones required the distortion to be at 2% before it was audible.

4) mixed 60Hz, 3KHz, 7KHz tones. Distortion was heard at 4%.

5) with simple music, 6% distortion became audible on voice. Guitar and cymbals showed evidence of audible distortion at 12%.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
I understand that the single SINAD figure you publish in the bar charts scross all DACs are simply the 1 kHz SINADs. Is that correct?
Yes.

If so, wouldn't it be much more appropriate to take the minimum SINAD (or maximum THD+N) from the THD-N vs frequency measurements, or an avarage across the frequency range? I mean, what's a high SINAD at 1 kHz good for if there's a poor SINAD @ e.g. 4 kHz?
THD+N versus frequency gives you the full range. But it is not that simple. With 1 kHz tone, we can capture up to 20 harmonics to 20 kHz if the test bandwidth is 20 kHz. With a 10 Khz band, we only have one harmonic included in that range.

Now, we can expand the bandwidth of the test but then that starts to include ultrasonic information that is not audible. Indeed the THD+N versus Frequency does that and often requires me investigating what is going on in ultrasonic range to see if something is up there that is influence what seems to be audible band measurements.

Overall, I am not a fan of THD+N versus frequency.

A better bet is to look at the IMD test which is the combination of two tones: 60 Hz and 7 kHz. Because of the way that test works, we don't need to worry about ultrasonic content. And we have both low and high frequencies included.

I also have 32 tone test which pushes the above to max.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,521
Likes
37,050
I understand that the single SINAD figure you publish in the bar charts scross all DACs are simply the 1 kHz SINADs. Is that correct?

If so, wouldn't it be much more appropriate to take the minimum SINAD (or maximum THD+N) from the THD-N vs frequency measurements, or an avarage across the frequency range? I mean, what's a high SINAD at 1 kHz good for if there's a poor SINAD @ e.g. 4 kHz?
To add to what Amir has already replied, in testing of devices usually SINAD doesn't change or doesn't change much over the frequency band. While there are exceptions, they are just that....exceptions. Most gear I've tested with sweeps and dual tone IMD sweeps, varies very little from low to high frequency. Like 3 or 4 db variance. You aren't likely to test something at 1 khz with good results and get significantly worse results at 4 khz.
 

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
2,681
Likes
4,218
Location
Liège, Belgium
I also have 32 tone test which pushes the above to max.
@amirm
I always wonder if we could have a figure linked to that one.
We can compute the total power, then the 32 tones power and we can then have a ratio of some sort, in dB.
Is that realistic ?
(Of course, FFT window size should always be the same. You said you use 256k, didn't you?)

We could even add a weighting which takes frequency-dependent sensitivity into account.
And, ultimately, trend that vs level...

By the way, what's the phase of each of the 32 tones?
 
Last edited:

anmpr1

Major Contributor
Forum Donor
Joined
Oct 11, 2018
Messages
3,722
Likes
6,405
For those wishing to read the article in its entirety, attached is a PDF.
Another interesting 'finding' or perception was that on certain music, a small amount of audible crossover distortion provided a certain 'ambience' or 'richness' to the sound (0.35%), "presumably because the harmonic distortion generated by the crossover notch added 'a little harmony' (no pun)."
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
By the way, what's the phase of each of the 32 tones?

The used tones are not harmonics from each other so have no phase relation.
 
Top Bottom