Understanding Audio Dynamic Range / SNR (Part 1)

[amirm]

Hello everyone. I am a bit under the weather today so didn't do any measurements but thought I do a video on dynamic range / signal to noise ratio and what it means at high level. In a future video I will dive much deeper into how much of it we need. So consider this the first installment.

------------
As always, questions, comments, recommendations, etc. are welcome.

Appreciate any donations using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

 

[abdo123]

It might be a good time to revive the room measurements for dummies threads and make them into a YouTube series.

 

[fcracer]

Amir, this is another great video! Thank you for the opportunity to learn from you; I think you’re doing a great service to the audio community and future audio fans!

In all the modern amplifiers I’ve used, I’ve found the only two differentiators that I can subjectively hear are dynamic range (hiss from efficient speakers) and damping factor. I have a question about dynamic range as it relates to when you combine electronics.

Please pardon the question if it’s a dumb one but I did search for answers and was not able to find them. Let’s say you have a DAC pre-amp with 120db of dynamic range, and you marry that to an amplifier with 100db of dynamic range, is there a way to estimate the system’s dynamic range?

I assume it would be the lesser of the components at 100db, but then I wonder what impact the interconnects have, and is there any effect on noise and thus dynamic range of having the two units sitting near each other? Are there other factors that impact dynamic range in this situation?

Your insight and knowledge are greatly appreciated. Thanks again!

 

[amirm]

Please pardon the question if it’s a dumb one but I did search for answers and was not able to find them. Let’s say you have a DAC pre-amp with 120db of dynamic range, and you marry that to an amplifier with 100db of dynamic range, is there a way to estimate the system’s dynamic range?

I will actually do a video and practical demo of it. My general rule is that the upstream device needs to be 10 dB better to have to have little effect (0.5 dB) on the downstream device's dynamic range. In your example, the effect would be negligible since the 120 dB device is so much better. There are calculators online for this by the way.

 

[okok]

SNR under what vrms
most important question

 

[B4ICU]

Some remarks:
1. Electronic NR (Noise Range) is very high today, and can reach values way over 120dB.
But our environment is way worst. Urban background noise (you can check it easily with an application and your cell phone), is about 50 dB in a good day. Late night or a nice quiet neighborhood. Taking that as the noise floor, would ask to amplify the volume to some more significant levels. Lets say
90dB plus. 90-50 dB would leave you with a more practical 40 dB dynamic range, rather than 130 dB of an AP measurement screen.
2. There is a level of how loud you can go, with the Amp's max power and speakers efficiency. You may have a police visit after a neighbour's complain. So, after all, there's a limit, way more tangible than electronic dynamic range, over an AP test bench.
Not to say, that most CD's are recorded PCM 16 bit, equals to 96 dB dynamic range.
If you are a vinyle guy, it is way below 75dB. A 60 dB (rumble figure) is more practical.
3. The way to go, is not by getting a more expensive (and quit) system, but find a house with a dedicated room in the basement. That would be really nice and quite.
4. Well equipped sound recording studios, have a silent room, isolated (floating floor) that can reach very low levels of background noise.
An other limit of dynamic range is the compression that is used during the recording. Most studio practice some, and it is a standard equipment on every audio mixer. Some are using it aggressively. It makes the process way more easy that way.
Some manufacturers are proud of having ADC's or DAC's with the access of 24bit. Well, it is mostly to say that the top (high bit) 16 bit are linear and noise free. A true 24 bit equals to 144 dB! No home speaker can play that loud, or the low end would be soaked deep in the noise floor.
So after understanding the limits, that 133 dB measurement, is for display, not of any practical use.

 

[Music1969]

Wow already passed 6k YouTube subscribers @amirm , in such short time

You're comin for Hanz B... and then Darko next !

 

[Morla]

not sure if that was asked before. Couldn't find it by using the search function but.. what is the camera model you use @amirm?

 

[JustAnandaDourEyedDude]

A helpful video that explains how to make sense of the measurement of one of the types of inaccuracy in audio electronics that was mentioned in a previous video you did, i.e. noise. As you go along making other videos explaining the essential technical aspects of audio science and audio measurements, I think it helps maintain continuity of ideas if you remind the viewer that some complex concept you use now was explained by you in an earlier video (just as you did here with the concept of "decibel" when you said "remember how a decibel is a ratio?"). Last year, it took me multiple weekends of reading many of your gear reviews to infer that the noise level is usually independent of the signal while harmonic distortion amplitudes tend to scale with the signal amplitude. I should have read your articles on Understanding Audio Measurements . This video clears up puzzles like that for beginners in audio science.

 

[Lambda]

Why not apply a weighting like (A) to the noise measurement?
Because two peaces of equipment with the same 120dB SNR maybe have different SNR in dBA.

 

[JStewart]

Some remarks:
1. Electronic NR (Noise Range) is very high today, and can reach values way over 120dB.
But our environment is way worst. Urban background noise (you can check it easily with an application and your cell phone), is about 50 dB in a good day. Late night or a nice quiet neighborhood. Taking that as the noise floor, would ask to amplify the volume to some more significant levels. Lets say
90dB plus. 90-50 dB would leave you with a more practical 40 dB dynamic range, rather than 130 dB of an AP measurement screen.
2. There is a level of how loud you can go, with the Amp's max power and speakers efficiency. You may have a police visit after a neighbour's complain. So, after all, there's a limit, way more tangible than electronic dynamic range, over an AP test bench.
Not to say, that most CD's are recorded PCM 16 bit, equals to 96 dB dynamic range.
If you are a vinyle guy, it is way below 75dB. A 60 dB (rumble figure) is more practical.
3. The way to go, is not by getting a more expensive (and quit) system, but find a house with a dedicated room in the basement. That would be really nice and quite.
4. Well equipped sound recording studios, have a silent room, isolated (floating floor) that can reach very low levels of background noise.
An other limit of dynamic range is the compression that is used during the recording. Most studio practice some, and it is a standard equipment on every audio mixer. Some are using it aggressively. It makes the process way more easy that way.
Some manufacturers are proud of having ADC's or DAC's with the access of 24bit. Well, it is mostly to say that the top (high bit) 16 bit are linear and noise free. A true 24 bit equals to 144 dB! No home speaker can play that loud, or the low end would be soaked deep in the noise floor.
So after understanding the limits, that 133 dB measurement, is for display, not of any practical use.

This article from Amir is relevant to this subject.
https://www.audiosciencereview.com/forum/index.php?threads/dynamic-range-how-quiet-is-quiet.14/

 

[Lambda]

@B4ICU
Lot of miss conceptions.
1.IEMs capable of exceeding 135dB exist.
2.You can hear high pitch signal way under the mostly pink/brown background noise.
(Hering tests are not made in sound proof rooms at night)
3. A typical balloon pop can be 150-160dB
4. 135dB at 20Hz is not even that loud.

its not as easy as Adding or Subtracting one number from/to the other to get the usable dynamic.

 

[ted92]

Thanks Amir
Very interesting and important to understand
I would be very interested in the math behind the panther (which panther for which result ...)
Maybe it is already done and I missed something.

 

[jasonhanjk]

For low impedance IEM.
Noise is best to compare in absolute value, example in uV.

4Vrms with 120dB SNR is 4uV, which can be heard.

 

[peng]

Thank you for this excellent video! I hope part2 will come soon.
Jus two questions:

1) When measuring SNR, is there a published standard you typically follow, or such standards only applies to the filter such as the AES17?

2) In measuring SNR with the AP, how is the noise separated from distortions? If not separated, wouldn't SNR = SINAD? If separated, is the distortion part (from the DUT's output obviously) filtered out by the AP, or is it calculated by some sort of DSP/algorithm build in the AP?

I hope Amir or other experts can clarify this for me as I have already search the internet and could not get a clear anwer.

 

[Lambda]

4Vrms with 120dB SNR is 4uV, which can be heard.

This is why @amirm also tests at 50mV
But i agree a absolute value like dBµV would be better!

also why not go all the way and show dBµV/√Hz Noise Density

This has the advantag of beeing true and meaningful without quoting bandwidth.

 

[jasonhanjk]

I think you meant dBV.
1uV is -120dBV.

One or Amir video he mention turning the volume knob for his 50mV SNR. So far I won't disagree this method.

 

[Dj7675]

Great video. Looking forward to part II.

 

[GGroch]

Some remarks:...3. The way to go, is not by getting a more expensive (and quiet) system, but find a house with a dedicated room in the basement. That would be really nice and quiet.

Good points. For those who are not ready to move to a new home, you could buy a pair of ANC headphones with excellent audio specs. Some headphones like the wired AKG N90q feature ANC for this purpose. ANC cannot be turned off on the N90q because it is part of the critical listening experience, not an attempt to muffle airplane engine noise. I am surprised that more audiophile headphones do not offer this feature, but unlike the N90q, most of audiophile headphones are passive/analog.

 

[EXIF68]

Hello Amir,
very good explanation of the db-Scale.
In this case, i have an idea what would be very interesting for an additional measuring problem:
Most High-Ender saying that the correct polarity of the mains plug is absolutely necessary for minimum potential difference between audio components. And wrong polarity of the mains plug would have a major negative impact in the soud quality due to massive compensating currents. For me, i have tried it but i cannot hear any difference depending on the mains plug polarity. Is there something relevant to measure or my ears are too weak?
Greetings, Walter