# How to interpret AVR measurement graphs?

#### -Matt-

##### Member
As this is my first post here, I'd like to thank Amir for hosting the site and for all the work that goes into equipment measurement. Debunking marketing claims and helping consumers to make informed choices is important work.

Whilst I understand the need for standardised measurements, a lot of nuance is missed if we only look at the headline dashboard figures obtained with 2V or 4V output.

If I typically listen to TV at -30dB (on a room corrected AVR) how do I find the SINAD that I should expect on an intermodulation distortion plot like this one for the Denon AVC-X4700h?

Should I project up from -30 to the curve - which yields a figure of approximately -80dB?

However, the scale on the horizontal axis is dBFS (dB full scale). I believe that reference level is supposed to be at -20dBFS. Does that mean that I should actually project up from -50dBFS and hence get a figure of just -60dB?

Additionally, much of the comment on this site is biased towards the use of seperate pre/pro and power amps (even in the case of AVRs). Many people will purchase an AVR intending to use the internal amplification and to listen at sane levels; however, it can be difficult to know how many watts will actually be needed.

Taking the 8 ohms amplifier performance as an example (also from the AVC-X4700h):

And using this online SPL calculator:
With sensitivity of 90 dB SPL, and two speakers near a wall, 15ft from the listening position.

If reference level is 85dB SPL and I listen at -30dB from reference then the target should be -55dB SPL. The calculator suggests that I only need 0.00165 W to reach this level. This seems ridiculously low (not even on the plot). Am I missing something here?

#### NTK

##### Major Contributor
Forum Donor
Welcome to ASR!
If I typically listen to TV at -30dB (on a room corrected AVR) how do I find the SINAD that I should expect on an intermodulation distortion plot like this one for the Denon AVC-X4700h?
When you see the curve transitions to a downward sloping straight line, with a slope of -1 (i.e. IMD+N goes up 10 dB when generator level drops 10 dB), the device is operating in the noise dominated region.

The SMPTE/DIN IMD distortion measures the distortions + noise using 2 sine tones, a 60 Hz tone and a 7 kHz tone at 1/4 amplitude of the 60 Hz tone. For a well behaving device, it is rough similar to THD+N at 60 Hz.

The way to interpret the curve is that the curve shows the ratio between the test signal and the noise + distortions. As music is not constant amplitude, thus the ratio is higher when the music is loud and lower when the music is soft.

Distortions in competently executed electronics is below audibility. Noise is the important part. What you want is the noise generated by the electronics to be below the background noise at your listening position. Say, for example, your max listening volume settin is -30 dBFS, and at your listening position this produces a max sound pressure level of 105 dB SPL. The IMD+N of the Denon X4700H at -30 dB is 78 dB (and it is all noise). Therefore, noise from the AVR at your listening position will be 105 - 78 = 27 dB SPL. If your room noise is higher than 27 dB, then the Denon will not be a problem.

If reference level is 85dB SPL and I listen at -30dB from reference then the target should be -55dB SPL. The calculator suggests that I only need 0.00165 W to reach this level. This seems ridiculously low (not even on the plot). Am I missing something here?
Unless it is for very soft background music, I highly doubt you'll be listening at an average level of 55 dB SPL.

#### peng

##### Major Contributor
Forum Donor
If reference level is 85dB SPL and I listen at -30dB from reference then the target should be -55dB SPL. The calculator suggests that I only need 0.00165 W to reach this level. This seems ridiculously low (not even on the plot). Am I missing something here?

You did not miss a thing, but 55 dB spl is likely too low for most people. Also, when you listen at 55 dB SPL, I would assume that's sort of average. Music has peaks that could range from just a few dB to more than 20 dB. Even your peak power need will still be low if you aim for 55 dB SPL, but it will increase 100 times for those who listen to 75 dB SPL. There are those who will say people don't realize how little power they actually need, then there also those who will say people don't realize their amps may clip more often than they think/or know, both can be true, it depends..

OP
M

#### -Matt-

##### Member
Thank you both for your replies; with your help, I might be getting there!
Peng, I also found the calculations that you posted on the AVC-X3700 review thread very helpful.

For evening TV viewing the AVR volume will typically read between -30 and -25 dB. I guess this is fairly quiet (my wife can fall asleep in the same room)! However, -25 dB relative to reference should equate to an average SPL of 60 dB which is apparently the level for normal conversation and therefore seems reasonable for a news readers voice. Presumably this still allows a peak SPL of 80 dB for explosions etc. (Perhaps the calibration on my AVR for reference level isn't all that accurate anyway). Of course, I like to listen to music and movies at much higher levels!

I had missed an important piece of information in the title in the main dashboard figure where it says "Vol = 82.5" (-2.5 dB relative to reference) and this is the volume setting needed to obtain 2V rms output from the preamplifier outputs. Is it the case that the volume control remains fixed at this position for all other measurements?

So, on the "Intermodulation Distortion" plot, when the generator level is at 0 dBFS do the constructive interference peaks of the 60Hz and 7kHz tones exactly use up every available bit of dynamic range in the digital input? I.e. There is no digital clipping of the generator signal and no remaining headroom?

Is the "SINAD vs Measured Level" plot also made with "Vol=82.5", by varying the generator level (and recording both SINAD and measured output level)? I had initially though that this was produced by sweeping the volume control.

#### peng

##### Major Contributor
Forum Donor
I had missed an important piece of information in the title in the main dashboard figure where it says "Vol = 82.5" (-2.5 dB relative to reference) and this is the volume setting needed to obtain 2V rms output from the preamplifier outputs. Is it the case that the volume control remains fixed at this position for all other measurements?

82.5 is just where he gets 2 V from the preamp output for the tests. He did not run Audyssey in his room with his speakers, so we don't know if he would get "reference level" in his mlp or not, most likely not. If you run Audyssey/auto setup, then after that, at volume 80 you should be at "reference level" from you main mic position but that's if you are watching a THX movie such as Star Wars.

So, on the "Intermodulation Distortion" plot, when the generator level is at 0 dBFS do the constructive interference peaks of the 60Hz and 7kHz tones exactly use up every available bit of dynamic range in the digital input? I.e. There is no digital clipping of the generator signal and no remaining headroom?

I'd leave that to NTK since he responded to your IMD question first.

Is the "SINAD vs Measured Level" plot also made with "Vol=82.5", by varying the generator level (and recording both SINAD and measured output level)? I had initially though that this was produced by sweeping the volume control.

If you mean the SINAD vs Pre out voltage then yes, unless stated otherwise (by him).

Last edited:

#### SKBubba

##### Senior Member
AVR volume will typically read between -30 and -25 dB. I guess this is fairly quiet (my wife can fall asleep in the same room)!

Yikes! My Mrs. is running out of the room yelling at me to turn that shit down at those levels. -25 is about my limit.

(That's for music, on my AVR, with built-in "reference" level, whatever that is.)

Last edited:

#### abdo123

##### Master Contributor
Forum Donor
Honestly i would ignore all AVR measurements (except for headless panther) because they're all within ~10 dB from each other anyway.

Buy only based on power provided and features.

#### NTK

##### Major Contributor
Forum Donor
So, on the "Intermodulation Distortion" plot, when the generator level is at 0 dBFS do the constructive interference peaks of the 60Hz and 7kHz tones exactly use up every available bit of dynamic range in the digital input? I.e. There is no digital clipping of the generator signal and no remaining headroom?
That would be my guess, but I don't know for sure.

I just read up on how SMPTE/DIN IMD is measured, and found out what I wrote in post #2 was totally incorrect

According to this from Audio Precision, not all noise is included in the measurement of IMD.
SMPTE IMD is a technique for measuring IMD (intermodulation distortion) according to the SMPTE RP120-1983 standard. The DIN intermodulation distortion technique uses a similar method.
The stimulus is a strong low-frequency interfering signal (f1) combined with a weaker high frequency signal of interest (f2). f1 is usually 60 Hz and f2 is usually 7 kHz, at a ratio of f1_f2=4:1. The stimulus signal is the sum of the two sine waves. In a distorting DUT, this stimulus results in an AM (amplitude modulated) waveform, with f2 as the “carrier” and f1 as the modulation.
In analysis, f1 is removed, and the residual is bandpass filtered and then demodulated to reveal the AM modulation products. The rms level of the modulation products is measured and expressed as a ratio to the rms level of f2. The SMPTE IMD measurement includes noise within the passband, and is insensitive to FM (frequency modulation) distortion.
The APx500 implementation of SMPTE IMD provides the capability to vary the stimulus frequencies and to choose a 1:1 stimulus ratio.

I will fall back to the THD+N measurement to calculate noise. Per Amir's measurements, full power (173 W into 4 Ω) SNR is 105 dB. Output voltage at full power is sqrt(4*173) = 26.3 Vrms. Therefore, noise voltage is 26.3*10^(-105/20) = 150 μVrms.

If the speaker sensitivity is 91 dB/2.63V, the noise from one speaker would be 91 - 20*log10(150e-6/2.83) = 91 - 86 = 5 dB SPL at 1 m distance. With 2 speakers (+3 dB) and boundary gain (+6 dB), noise at 1 m would be 5 + 3 + 9 = 17 dB SPL. The situation is actually better.

#### peng

##### Major Contributor
Forum Donor
That would be my guess, but I don't know for sure.

I just read up on how SMPTE/DIN IMD is measured, and found out what I wrote in post #2 was totally incorrect

According to this from Audio Precision, not all noise is included in the measurement of IMD.
SMPTE IMD is a technique for measuring IMD (intermodulation distortion) according to the SMPTE RP120-1983 standard. The DIN intermodulation distortion technique uses a similar method.
The stimulus is a strong low-frequency interfering signal (f1) combined with a weaker high frequency signal of interest (f2). f1 is usually 60 Hz and f2 is usually 7 kHz, at a ratio of f1_f2=4:1. The stimulus signal is the sum of the two sine waves. In a distorting DUT, this stimulus results in an AM (amplitude modulated) waveform, with f2 as the “carrier” and f1 as the modulation.
In analysis, f1 is removed, and the residual is bandpass filtered and then demodulated to reveal the AM modulation products. The rms level of the modulation products is measured and expressed as a ratio to the rms level of f2. The SMPTE IMD measurement includes noise within the passband, and is insensitive to FM (frequency modulation) distortion.
The APx500 implementation of SMPTE IMD provides the capability to vary the stimulus frequencies and to choose a 1:1 stimulus ratio.

I will fall back to the THD+N measurement to calculate noise. Per Amir's measurements, full power (173 W into 4 Ω) SNR is 105 dB. Output voltage at full power is sqrt(4*173) = 26.3 Vrms. Therefore, noise voltage is 26.3*10^(-105/20) = 150 μVrms.

If the speaker sensitivity is 91 dB/2.63V, the noise from one speaker would be 91 - 20*log10(150e-6/2.83) = 91 - 86 = 5 dB SPL at 1 m distance. With 2 speakers (+3 dB) and boundary gain (+6 dB), noise at 1 m would be 5 + 3 + 9 = 17 dB SPL. The situation is actually better.

Can you please explain what that +6 dB boundary gain it? Thank you

#### NTK

##### Major Contributor
Forum Donor
Can you please explain what that +6 dB boundary gain it? Thank you
The 6 dB boundary gain is a crude approximation. The rated SPL output of the speaker is for radiating into 360° open space (fullspace). If the speaker sits against a perfectly reflective front wall (radiating into half space), the wall reflection effectively doubles the output. There are many obvious simplifications and assumptions in this estimate, as no interferences from the reflected waves are considered, speakers are not omnidirectional at all frequencies, walls are not perfectly reflective, there are more than 1 boundaries in play, etc.

The assumption of 6 dB boundary gain is also used in the CTA-2034 estimation of in-room SPL.

#### peng

##### Major Contributor
Forum Donor
The 6 dB boundary gain is a crude approximation. The rated SPL output of the speaker is for radiating into 360° open space (fullspace). If the speaker sits against a perfectly reflective front wall (radiating into half space), the wall reflection effectively doubles the output. There are many obvious simplifications and assumptions in this estimate, as no interferences from the reflected waves are considered, speakers are not omnidirectional at all frequencies, walls are not perfectly reflective, there are more than 1 boundaries in play, etc.

The assumption of 6 dB boundary gain is also used in the CTA-2034 estimation of in-room SPL.

View attachment 167617

I thought it has to do with room gain but didn't want to assume that's what you refer to. Does it really apply to noise though, unless the noise is made up of mainly the lower frequencies? Or you just want to assume the worst case scenario?

#### NTK

##### Major Contributor
Forum Donor
Yes, I am assuming worst case.

OP
M

#### -Matt-

##### Member
NTK, thanks for the link to Audio Precision - not sure why I didn't think to look there.

...Output voltage at full power is sqrt(4*173) = 26.3 Vrms. Therefore, noise voltage is 26.3*10^(-105/20) = 150 μVrms.

In the above, I wonder if it might be necessary to divide the output voltage by 4 to account for the 4:1 ratio between f1 and f2 and this:

"The rms level of the modulation products is measured and expressed as a ratio to the rms level of f2."

#### NTK

##### Major Contributor
Forum Donor
NTK, thanks for the link to Audio Precision - not sure why I didn't think to look there.

In the above, I wonder if it might be necessary to divide the output voltage by 4 to account for the 4:1 ratio between f1 and f2 and this:

"The rms level of the modulation products is measured and expressed as a ratio to the rms level of f2."
I used the THD+N graph for my noise calculations in post #8 instead of the IMD graph. SMPTE/DIN IMD measurements do not account for all the noise, only the portion that is "in the passband" of the IMD analyzer.

A detailed description of how SMPTE/DIN IMD is measured is given here:

OP
M

#### -Matt-

##### Member
I used the THD+N graph for my noise calculations in post #8 instead of the IMD graph.

Ah yes, of course

Thanks for the document, from a very brief skim through it looks fairly detailed but nice and clear.

#### Dj7675

##### Major Contributor
Forum Donor
Honestly i would ignore all AVR measurements (except for headless panther) because they're all within ~10 dB from each other anyway.

Buy only based on power provided and features.
I don't really agree with this...
* -10dB, if noise, could go from inaudible to an audible problem, in particular if using sensitive speakers or seating position is close to some speakers (such as atmos or surround speakers)
* This would ignore one of the most important measurements I think Amir does for AVR's. Without it, we would be in the dark as to amp sensitivity needed to match for best performance.

* How would we know if the amps in the AVRs are any good without looking at the measurements for the amp section (both power and quantity). Measurements can show very differing quality of power. For example the NAD T758 vs Denon X3700.

While I agree that people can get carried away on differences between products, I think saying to ignore all measures except total power and headless panthers is a mistake. For many that are listening at low volumes, in a noisy environments, or simply don't care, measurements really don't matter. However, if you want to be relatively certain that the device is transparent in your listening environment, the tests done by Amir are actually important. I have a pretty quite home theater (low noise floor) and sometimes like to listen pretty loudly. the last thing I want is for the electronics to get in the way. My completely arbitrary number is a SINAD of 100. I believe I can be pretty certain I could never tell the difference if it was any better. Is the actual number probably lower... probably so. But I can be more certain at a SINAD of 100 than 80. Since we have actual measurements thanks to Amir, why not factor this into a purchase decision?

Last edited:

#### peng

##### Major Contributor
Forum Donor
I don't really agree with this...
* -10dB, if noise, could go from inaudible to an audible problem, in particular if using sensitive speakers or seating position is close to some speakers (such as atmos or surround speakers)
* This would ignore one of the most important measurements I think Amir does for AVR's. Without it, we would be in the dark as to amp sensitivity needed to match for best performance.

View attachment 168222

* How would we know if the amps in the AVRs are any good without looking at the measurements for the amp section (both power and quantity). Measurements can show very differing quality of power. For example the NAD T758 vs Denon X3700.

View attachment 168226View attachment 168227

While I agree that people can get carried away on differences between products, I think saying to ignore all measures except total power and headless panthers is a mistake. For many that are listening at low volumes, in a noisy environments, or simply don't care, measurements really don't matter. However, if you want to be relatively certain that the device is transparent in your listening environment, the tests done by Amir are actually important. I have a pretty quite home theater (low noise floor) and sometimes like to listen pretty loudly. the last thing I want is for the electronics to get in the way. My completely arbitrary number is a SINAD of 100. I believe I can be pretty certain I could never tell the difference if it was any better. Is the actual number probably lower... probably so. But I can be more certain at a SINAD of 100 than 80. Since we have actual measurements thanks to Amir, why not factor this into a purchase decision?

Very true what you are saying about the noise part. I always wonder why my AVR-X4400H is so much quieter than the Marantz AV8801 it replaced. Amir's measurements explained it. Not that I was hearing the hiss and hum when watching a movie or even music when I was using the Marantz, but knowing the noise was audible during a quiet passage with volume approaching 0, or even -10 would bother me.

With the ASR measurements now, I have the option to steer clear of audible distortions and/or noise if and when I am shopping for AVR or AVP again.

#### abdo123

##### Master Contributor
Forum Donor
I don't really agree with this...
* -10dB, if noise, could go from inaudible to an audible problem, in particular if using sensitive speakers or seating position is close to some speakers (such as atmos or surround speakers)
* This would ignore one of the most important measurements I think Amir does for AVR's. Without it, we would be in the dark as to amp sensitivity needed to match for best performance.

View attachment 168222

* How would we know if the amps in the AVRs are any good without looking at the measurements for the amp section (both power and quantity). Measurements can show very differing quality of power. For example the NAD T758 vs Denon X3700.

View attachment 168226View attachment 168227

While I agree that people can get carried away on differences between products, I think saying to ignore all measures except total power and headless panthers is a mistake. For many that are listening at low volumes, in a noisy environments, or simply don't care, measurements really don't matter. However, if you want to be relatively certain that the device is transparent in your listening environment, the tests done by Amir are actually important. I have a pretty quite home theater (low noise floor) and sometimes like to listen pretty loudly. the last thing I want is for the electronics to get in the way. My completely arbitrary number is a SINAD of 100. I believe I can be pretty certain I could never tell the difference if it was any better. Is the actual number probably lower... probably so. But I can be more certain at a SINAD of 100 than 80. Since we have actual measurements thanks to Amir, why not factor this into a purchase decision?

The NAD amplifier/receiver you used as example had the headless panther. In reality, the difference is very very minute between everything Amir measured.

Either way, the reciever with the better room correction will sound best, and the reciever with the most stable and easy to use operation system will provide a better use experience. Amir doesn't cover either of those two. By all means check if the reciever is broken or not in the review, but the moment it's not broken a lot of other variables are way more important.

#### Chromatischism

##### Major Contributor
Forum Donor
Therefore, noise from the AVR at your listening position will be 105 - 78 = 27 dB SPL. If your room noise is higher than 27 dB, then the Denon will not be a problem.
I have tested this with my room's noise floor of 45 dB. Even the smallest unit of output from the AVR, or -79, ("1" on the absolute scale) is just audible. So sounds don't need to be louder than the noise floor to be audible.

#### beagleman

##### Addicted to Fun and Learning
Very true what you are saying about the noise part. I always wonder why my AVR-X4400H is so much quieter than the Marantz AV8801 it replaced. Amir's measurements explained it. Not that I was hearing the hiss and hum when watching a movie or even music when I was using the Marantz, but knowing the noise was audible during a quiet passage with volume approaching 0, or even -10 would bother me.

With the ASR measurements now, I have the option to steer clear of audible distortions and/or noise if and when I am shopping for AVR or AVP again.
Your post is a bit confusing.........

Nowhere on this forum, have I heard anyone say they could actually "Hear" audible noise or distortion on any AVR, but merely the SINAD for the most part simply is mostly an engineering exercise in how well designed something is.

Even the worse measuring AVR, the noise levels would easily be drowned out by program material or the noise floor of the house or room.

Headphone at very high levels, much more a possibility, but the SINAD at 5 watts of even -70db would be FAR below the actual program content, which would be over 90db with most speakers.............

Replies
48
Views
3K
Replies
32
Views
3K
Replies
574
Views
94K
Replies
254
Views
42K
Replies
27
Views
2K