• 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!

Question about measuring at end of an audio chain

ShiZo

Addicted to Fun and Learning
Forum Donor
Joined
Sep 7, 2018
Messages
835
Likes
555
Sorry, I'm uneducated on the topic and might not have ask this correctly. But basically I'm wondering if a system is help back by the sinad, snr ect of the system's weakest link or if there is some sort of sum instead?

Say if my dac has an SNR of 120, and my amp 110, what is the end result?
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
Sorry, I'm uneducated on the topic and might not have ask this correctly. But basically I'm wondering if a system is help back by the sinad, snr ect of the system's weakest link or if there is some sort of sum instead?

Say if my dac has an SNR of 120, and my amp 110, what is the end result?

The S/N ratio of the DAC is measured from the residual noise in uV, with respect to the maximum output in Volts. The amplifier is measured the same way, but with a shorted input with respect to full rated output. The amplifier will of course be connected to the D/A converter and amplify both the residual noise of the D/A converter and the musical content by the inherent gain and add its own residual noise.

At any arbitrary normal volume position considerably less than full power, the S/N ratio of the amplifier will be much less than 110dB, perhaps 70-80dB.

So there is no single answer to your question, except that the overall S/N ratio will be less than the amp, and often considerably less.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Adding to what @restorer-john said, dB is a logarithmic unit, so if the DAC noise levels are low they could be inconsequential anyway even when amplified.

For example to answer your question, the addition of two signals, one is -110 and one is -120 is -109.58 dB.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
Adding to what @restorer-john said, dB is a logarithmic unit, so if the DAC noise levels are low they could be inconsequential anyway even when amplified.

Not exactly correct.

A theoretical 120dB S/N DAC will have a 4uV residual with respect to its 4V output (assuming bal XLR).

A typical amplifier has around 29dB of gain. That is a voltage gain of 28 times.

You now have 112uV (~0.1mV) from amplifying just the residual noise from the connected D/A converter. And the residual of the amplifier itself.

A really good power amplifier with say 150WPC may have a S/N of 120dB plus, which will mean a residual of around 35uV, so the D/A's contribution is well over three times the amplifier's and the two will combine to become audible when the volume is high. That example would result in a S/N of ~107dB, a degradation of 13dB from a 120dB rated amplifier.

And of course, once we go down in level, we are into the 70s and 80s for sure.
 

abdo123

Master Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
7,425
Likes
7,941
Location
Brussels, Belgium
Not exactly correct.

A theoretical 120dB S/N DAC will have a 4uV residual with respect to its 4V output (assuming bal XLR).

A typical amplifier has around 29dB of gain. That is a voltage gain of 28 times.

You now have 112uV (~0.1mV) from amplifying just the residual noise from the connected D/A converter. And the residual of the amplifier itself.

A really good power amplifier with say 150WPC may have a S/N of 120dB plus, which will mean a residual of around 35uV, so the D/A's contribution is well over three times the amplifier's and the two will combine to become audible when the volume is high. That example would result in a S/N of ~107dB, a degradation of 13dB from a 120dB rated amplifier.

And of course, once we go down in level, we are into the 70s and 80s for sure.

correct me if i'm wrong but aren't the amplifiers measured here are measured with the analog output of the APx555? (which is around 120 SINAD).

so that 112 uV of noise is already added on top anyway.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
correct me if i'm wrong but aren't the amplifiers measured here are measured with the analog output of the APx555? (which is around 120 SINAD).

OP is asking about S/N. Signal to noise in amplifiers is measured by shorting the inputs to an amplifier, measuring the residual and then referencing that to the full output swing. The AP's analog output will short (relay) on a S/N test, so no residual.

AP's notes:
In a conventional SNR measurement a device is first stimulated with a signal at full scale with its volume control set to maximum, if present. A level measurement is made to establish the maximum output reference. The stimulus tone is then removed and the inputs of the device are either terminated or shorted. The residual noise is then measured. The SNR value is then the ratio of the full scale output level of the device to the residual noise level of a device.

https://www.ap.com/technical-library/signal-to-noise-ratio-snr-dynamic-range-and-noise/

The contribution of the AP's own residual noise multiplied by the gain of the amplifier in question if not shorting the inputs, would be taken into account on a dynamic range test via the analog gen. Dynamic range tests are also used with D/A converters (except with a digital signal) as S/N tests will likely return erroneous numbers due to mute on successive zeroes circuitry in most D/As.
 
Last edited:

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
But basically I'm wondering if a system is help back by the sinad, snr ect of the system's weakest link or if there is some sort of sum instead?

When you hear no hiss when nothing is playing and you have the volume control set to the position then you don't have to worry about the system noise floor.
Distortion at 1kHz at that specific set level is not really that relevant. The DAC might have that SINAD but the amp will most likely not nearly be at full power yet so distortion at 1kHz won't reach that of the power amp near full power.

Assuming since you seem to be a SINAD guy and thus distortion will be low it won't be a problem. Your speakers will 'f the 2nd harm distortion up to much higher values. Here's the thing though. The louder you play the more distortion you get from speakers but the less sensitive your hearing becomes for 2nd harm. (masking).
At normal SPL the system distortion will be dwarfed by that of the speakers and depends on the distortion of the speakers itself.

So... when you don't hear noise at your listening position with nothing playing, late at night (don't attempt during the day) with the volume control at or slightly above the 'comfortable loud' position then you can stop worrying about SINAD, noise etc.

As restorer John already mentioned the final numbers depend on so many factors.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
When you hear no hiss when nothing is playing and you have the volume control set to the position then you don't have to worry about the system noise floor.

This I agree with, and is probably the most useful indicator.

Set your volume at normal, or even a high-ish typical listening level and, as long as your DAC hasn't been naughty and muted its output, if there's no hiss you can hear at your listening position, there's no problem. ;)
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
Sorry, I'm uneducated on the topic and might not have ask this correctly. But basically I'm wondering if a system is help back by the sinad, snr ect of the system's weakest link or if there is some sort of sum instead?

Say if my dac has an SNR of 120, and my amp 110, what is the end result?
The amp is the limiting factor here (in almost all cases). Here's what John Siau had to say about the improvement in SNR between the Benchmark DAC1 and DAC2, "the DAC2 has an SNR of 126 dB, while the DAC1 has an SNR of 116 dB. In a system with a properly configured gain structure, this 10 dB difference may not be audible. To hear the difference, the SNR of the power amplifier would need to exceed the SNR of the DAC1 by at least 6 dB."

To take @restorer-john's example of an amp producing 150W into 8ohms: at clipping it will be producing a signal level of 34.6V. If it has an SNR of 110dB, the noise produced by the amp will be 109uV. While the amp's stage gain is 29dB, the signal level coming in is considerably attenuated by the volume control before it hits the voltage amplification stage (almost all amps will go into clipping on a full-scale signal long before you hit the upper limit of the volume control), so the effective gain of the amp in this case is 20log (34.6/4) = 19.18dB and the 4uV noise produced by the DAC has only become 36.4uV. Thus the amp's noise ends up dominating the result.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,588
Likes
38,291
Location
Gold Coast, Queensland, Australia
While the amp's stage gain is 29dB

My example of 29dB was a "typical" 1.0-1.5V for full rated power amplifer. On an integrated, we are talking 150mV for full rated power on a line input being fed a typical 2.0 V source (D/A etc), so overall gain is way more and volume control is wound back significantly.

Depending on whether the front end is active or passive in that case, will mean a worse or better S/N. If using some of these low gain modern Class Ds or even the Benchmark in low gain, you have a different situation altogether.

S/N, levels and cascaded components is fun and games for sure. Nothing like measuring the chain at the end and at the beginning. :)
 
Last edited:

K man

Member
Joined
Jan 1, 2022
Messages
84
Likes
88
Location
Sweden
So how does it work for harmonic distortion when you lower the volume?

Since the residual noise is on a more or less steady level, S/N decreases when the volume decreases.

If you have a harmonic distortion at -90db, does that remain somewhat steady when you lower the volume, so eventually the residual noise (even when it's much lower than harmonic distortion at max volume) will take over and mask any harmonic distortion? Or does the level of harmonic distortion change with volume as well?
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
So how does it work for harmonic distortion when you lower the volume?

Since the residual noise is on a more or less steady level, S/N decreases when the volume decreases.

If you have a harmonic distortion at -90db, does that remain somewhat steady when you lower the volume, so eventually the residual noise (even when it's much lower than harmonic distortion at max volume) will take over and mask any harmonic distortion? Or does the level of harmonic distortion change with volume as well?
The answer is a little complex. If we simplify things and just talk about a pure analog amplifier then in general the distortion will remain about the same or rise slightly with increasing signal levels passing through the output stage. This is until the output stage reaches saturation (clipping) at which point the distortion rises markedly. Since power sweeps show THD+N, the noise typically dominates at lower volumes, then you reach a short plateau at which the lower relative contribution of noise is accompanied by slightly increased harmonic distortion, then you see the value shoot up as the amp clips.

Here's an example from the PA5 review:
Topping PA5 Measurements Power into 4 ohm Stereo Amplifier.png
 

K man

Member
Joined
Jan 1, 2022
Messages
84
Likes
88
Location
Sweden
If we simplify things and just talk about a pure analog amplifier then in general the distortion will remain about the same or rise slightly with increasing signal levels passing through the output stage. This is until the output stage reaches saturation (clipping) at which point the distortion rises markedly. Since power sweeps show THD+N, the noise typically dominates at lower volumes, then you reach a short plateau at which the lower relative contribution of noise is accompanied by slightly increased harmonic distortion, then you see the value shoot up as the amp clips.
So unless the harmonic distortion is high (as in tube amps) it really doesn't matter at all within normal listening levels (100uW to 1W), since background noise will mask it out completely? The harmonic distortion is effectively gone at those levels?
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
So unless the harmonic distortion is high (as in tube amps) it really doesn't matter at all within normal listening levels (100uW to 1W), since background noise will mask it out completely? The harmonic distortion is effectively gone at those levels?
Obviously there are a lot of factors that affect this, such as the dynamic range of the music you're listening to and the sensitivity of your speakers. But in general noise will be the dominant factor in music with a good dynamic range.

But for any competently-designed amp this will all be well below the audible threshold anyway. Even incompetent 'audiophile' designs need to try hard to break things in order to produce clearly audible distortion. It's certainly easy to find examples of tube amps where that's exactly what they did, but even in those cases you usually find that mains hum dominates the picture.
 
Top Bottom