General question on dynamic range and music

[restorer-john]

My question is why you chose 1 uW@8 ohms?

1uW (2.83mV) into a typical 8R loudspeaker of 90dBSPL/1W@1M would produce a signal 1000 times (60dB) less, or 30dBSPL. That would be very close to the limit of even a quiet typical listening room. It's dead quiet in my house at the moment and I measure 33.9dB A in the listening area.

[email protected] into an 89dBSPL@1W/M loudspeaker (tested now) on my bench is just faintly audible at 1M. I'm talking disappearing into the residual noise. At 3M forget it.

Pick 4R and lower sensitivity and gets even worse...

A typical amplifier capable of the power output I described above, would have residual noise well into the double figures of millivolts and as such, could not audibly reproduce the full dynamic range of the CD format. i.e. at the low end, the signal will disappear into noise and at the top end, you run out of available voltage swing (power).

 

[Blumlein 88]

1uW (2.83mV) into a typical 8R loudspeaker of 90dBSPL/1W@1M would produce a signal 1000 times (60dB) less, or 30dBSPL. That would be very close to the limit of even a quiet typical listening room. It's dead quiet in my house at the moment and I measure 33.9dB A in the listening area.

[email protected] into an 89dBSPL@1W/M loudspeaker (tested now) on my bench is just faintly audible at 1M. I'm talking disappearing into the residual noise. At 3M forget it.

Pick 4R and lower sensitivity and gets even worse...

A typical amplifier capable of the power output I described above, would have residual noise well into the double figures of millivolts and as such, could not audibly reproduce the full dynamic range of the CD format. i.e. at the low end, the signal will disappear into noise and at the top end, you run out of available voltage swing (power).

Then we are back to the issue of using a single noise number. I'll grant you that getting more than 80 db with various things like noise, and gain of the amp pushing the noise up etc etc. is near the limit of what most people will achieve. Also happens to be near the limit of anything recorded with a microphone can achieve.

 

[restorer-john]

Then we are back to the issue of using a single noise number. I'll grant you that getting more than 80 db with various things like noise, and gain of the amp pushing the noise up etc etc. is near the limit of what most people will achieve. Also happens to be near the limit of anything recorded with a microphone can achieve.

I deliberately used 1uW as that is about the limit for a typical dynamic loudspeaker driver of above average sensitivity to produce a perceivable sound. I also picked 3.15kHz as that is around where our ears are most sensitive. Above and below that, you won't hear 1uW.

A SOTA CD player will do 1.5-2.0uV of residual noise, meaning a potential 120dB dynamic range from 2.0V. But to be able to actually hear that residual on a loudspeaker, we'd need to get to that 2.5mV (1uW). That means an amplifier gain of 61dB (there's that number again). With it comes 1250 times the residual noise or 2.5mV. So again, the signal disappears into the residual.

The conundrum doesn't change no matter how one twists it, and merely highlights the absolute futility in obsessing over DACs with S/N numbers that few amplifiers on earth can do justice to. You've got kids here buying 'power' amplifiers that hardly classify as HiFi, with so little real power as to be a joke and thinking they can strap a SOTA DAC on the front and get 'better than CD' quality. It's funny.

 

[Blumlein 88]

I just tried this with Revel F208's. I could hear about 76 db below a one watt level using 3500 hz. This was at 1 meter. If I turned my head sideways, basically lining up the tweeter directly into my ear I could get another 3 db lower. Not a normal way of listening obviously. The F208's are also around 89 db/meter. I did the low level sound first checking levels with an ADC, and got the voltages at higher levels afterwards. This would be lower than 1 uwatt. If I've not messed up the calculation like not far from 25 nanowatts I estimated earlier. Something like 13 db SPL. I was using a file that was 24 bit, I didn't think about until after I was done. I could have used a 16 bit file. Even this from overly close is still shy of 96 db overall range.

 

[solderdude]

Using high efficiency speakers in a small room one may need a factor 100 less power to reach impressive peak levels compared to some 85dB bookshelves in a larger room.
So the 1μW number can be anything between a bunch of nW and several μW if the goal is audibility threshold in an undefined room with a range of different sensitivity speakers.

In general I agree with John about the dynamic range of CD being enough. That said... for DACs more might be needed if digital volume control is used and power amps are 'full open' regardless if that is the wisest thing to do .

 

[Sokel]

I deliberately used 1uW as that is about the limit for a typical dynamic loudspeaker driver of above average sensitivity to produce a perceivable sound. I also picked 3.15kHz as that is around where our ears are most sensitive. Above and below that, you won't hear 1uW.

A SOTA CD player will do 1.5-2.0uV of residual noise, meaning a potential 120dB dynamic range from 2.0V. But to be able to actually hear that residual on a loudspeaker, we'd need to get to that 2.5mV (1uW). That means an amplifier gain of 61dB (there's that number again). With it comes 1250 times the residual noise or 2.5mV. So again, the signal disappears into the residual.

The conundrum doesn't change no matter how one twists it, and merely highlights the absolute futility in obsessing over DACs with S/N numbers that few amplifiers on earth can do justice to. You've got kids here buying 'power' amplifiers that hardly classify as HiFi, with so little real power as to be a joke and thinking they can strap a SOTA DAC on the front and get 'better than CD' quality. It's funny.

Tried your example and what I get is this (everything is calibrated) :



I adjusted it to the point to be clearly audible but the funny thing is that as the time passes it becomes more and more audible (probably bothers me).
Sad thing is that my true RMS DMM can't register any value out of the speaker terminals,it reports plain zero even with the range adjusted to it's lowest.

Edit: DMM is true RMS up to 1Khz,no more,so...

 

[restorer-john]

Sad thing is that my true RMS DMM can't register any value out of the speaker terminals,it reports plain zero even with the range adjusted to it's lowest.

Edit: DMM is true RMS up to 1Khz,no more,so...

A DMM will have no accuracy down at 1mV (100uV) or less unless you've paid big dollars for a 6.5 digit or greater and even then, a sensitive ACmVM will be more useful- they are designed for audio down to 1mV and lower. Some have 300uV FSD or even 100uV FSD on a few analogue ones with 200kHz bandwidths. The wandering DC offset of an amplifier will also throw off your DMM, even on AC coupling in many cases.

Otherwise, you can use an ADC with very low noise, such as the Cosmos unit and calibrate the chain as best you can.

A decent old school 'scope with a 5mV per division with 5x mag can get you to 1mV per screen graticule/division (cm). So down to perhaps 100-200uV but you may have to manually trigger/ext trigger.

Otherwise buy a distortion analyser or a big expensive AP.

 

[Sokel]

A DMM will have no accuracy down at 1mV (100uV) or less unless you've paid big dollars for a 6.5 digit or greater and even then, a sensitive ACmVM will be more useful- they are designed for audio down to 1mV and lower. Some have 300uV FSD or even 100uV FSD on a few analogue ones with 200kHz bandwidths. The wandering DC offset of an amplifier will also throw off your DMM, even on AC coupling in many cases.

Otherwise, you can use an ADC with very low noise, such as the Cosmos unit and calibrate the chain as best you can.

Or you can buy a distortion analyser or an AP.

That's it,not even close to your standards.
I choose it for it's durability (yep,the old me) and decent price.



I would measure it with my ADC if I was not afraid and if the amps manufacturer allowed me to,on the contrary it states that nothing should ground outputs (balanced outputs amp).

 

[goat76]

The original question has been resolved some time ago already, it was caused by different definitions of dynamic range. My misunderstanding has been removed, but it seems not of everyone discussing here.

I guess you have realized that measuring the actual full dynamic range of recorded music doesn't necessarily have much meaning for how dynamic the music you listen to is actually heard. What we hear when it comes to dynamics is not necessarily the lowest point vs the highest peak in a recording, but rather the differences we hear in short bursts going from a "mean value" to the peaks, and therefore, a windowed measurement makes more sense like LUFS or similar that are meant to give us a better representation of what we as listeners perceive.

But I do understand that it could be interesting to see how much of the available space the entire dynamic range of the most dynamic music recordings may take, on a CD for example.

 

[restorer-john]

That's it,not even close to your standards.
I choose it for it's durability (yep,the old me) and decent price.

For 99% of stuff you'll do, the DMM will be fine. And, if you have an accident and destroy it, they don't cost much to replace. It's a good choice.

 

[restorer-john]

But I do understand that it could be interesting to see how much of the available space the entire dynamic range of the most dynamic music recordings may take, on a CD for example.

I have several of the very first demo CDs made for the early days of Compact Disc. A few of the classical tracks are scary in their dynamic range. A bit ridiculous actually. Even with 500 watts per channel on tap, you have to stand next to the preamp and 'ride' the volume control, lest you be catching a voicecoil as it leaves the magnet gap...

 

[Blumlein 88]

I handled the levels by using a Topping D10B which is measured to have low noise and excellent low level linearity. So I made the assumption if I reduced level in a sound editor the D10B would accurately reproduce levels. I don't think that is a bad assumption. I also assumed the amp, which is a quiet class D amp, does not suffer variable gain at lower input levels.

So I started with a -40 dbFS tone, and worked down until I could not longer hear it. I noted the level of the signal in the sound editor, and then used an ADC to measure the level at -40 db at the speakers. Low enough not to harm the ADC, high enough noise is no issue, and I know what input results in what levels. I then also measured -30 db at the speaker with the ADC as another point of reference. From that I could work out how my minimum level related to 1 watt. So it might not be as accurate as an AP or Cosmos or some dedicated measuring gear. It should be far more accurate than a DVM and not terribly wrong.

I did this because as restorer-john described the test it sounded like an interesting way to do it. You hear people describe various speakers as being better at low levels than others. It might be interesting to try the same with some other speakers. Of course it would be more repeatable to use a good microphone just comparing speakers at low levels, but not as informative as what these low levels mean or don't mean for listeners.

 

[gnarly]

The conundrum doesn't change no matter how one twists it, and merely highlights the absolute futility in obsessing over DACs with S/N numbers that few amplifiers on earth can do justice to. You've got kids here buying 'power' amplifiers that hardly classify as HiFi, with so little real power as to be a joke and thinking they can strap a SOTA DAC on the front and get 'better than CD' quality. It's funny.

Yep, It really is a joke, obsessing over DAC S/N numbers.

Personally, to achieve best dynamic range (quietest to loudest, not crest factor etc),
I've learned to live with whatever noise floor the acoustic environment dishes up.
And then simply try to achieve as high as possible SPL capability, that stays linear throughout headroom for peak transients.

Like you said in another post, it takes a heck of alot of amp power/voltage, to stay linear, especially for typical home-audio speakers' sensitivity/efficiency
.
I run powerful 5-way actives. On the sub sections, which have 101 dB sensitivity, I use amps with 2.5kW continuous per ch.
Still tickle clipping lights every rowdy now and then.

 

[TimF]

It is worthwhile seeking out a place beyond the sounds of machines and engines. If it is in the desert, the silence at first can be disturbing. If it is in a marsh, field, or woodlands, there is a sound environment as detailed and alive as the visual field. The sound environment we've grown accustomed to in human settlements is akin to --what comparison is apt?--the constant heavy bombardment of commercial messages we receive these days; and similar to the rude, overt, loud visuals of pornography. It is noisy. It can be quiet outdoors beyond the human environment, and it can also be noisy relatively speaking. Animals send a lot of sound messages. Most of us are so far removed from being able to hear and read the sounds in the natural environment. More so, most of us rarely if ever venture into a natural environment as hit might have been before the industrial revolution. What have we lost? No wonder so of us crave to recreate a detailed and rich sound environment free from the over and under lay of machine and engine. It is, I think, worth the effort to every now and then seek out and revisit a natural space.

 

[Jochen]

It is worthwhile seeking out a place beyond the sounds of machines and engines. If it is in the desert, the silence at first can be disturbing. If it is in a marsh, field, or woodlands, there is a sound environment as detailed and alive as the visual field. The sound environment we've grown accustomed to in human settlements is akin to --what comparison is apt?--the constant heavy bombardment of commercial messages we receive these days; and similar to the rude, overt, loud visuals of pornography. It is noisy. It can be quiet outdoors beyond the human environment, and it can also be noisy relatively speaking. Animals send a lot of sound messages. Most of us are so far removed from being able to hear and read the sounds in the natural environment. More so, most of us rarely if ever venture into a natural environment as hit might have been before the industrial revolution. What have we lost? No wonder so of us crave to recreate a detailed and rich sound environment free from the over and under lay of machine and engine. It is, I think, worth the effort to every now and then seek out and revisit a natural space.

Enjoying full dynamic range, so to speak!

 

[dasdoing]

this should be an eye opener https://www.audiocheck.net/testtones_dynamic.php

 

[RayDunzl]

I would measure it with my ADC if I was not afraid and if the amps manufacturer allowed me to,on the contrary it states that nothing should ground outputs (balanced outputs amp).

Doesn't your ADC have some balanced inputs?

 

[Jochen]

this should be an eye opener https://www.audiocheck.net/testtones_dynamic.php

done that, it is amazing how less is really audible. But still as with going for crazy high SINAD and 24bit when available (although I hardly hear the difference to 256kbit AAC, if at all), better be on the sure side and go for the maximum, if it is easily available.

 

[Sokel]

Doesn't your ADC have some balanced inputs?

It does,but as also @DonH56 confirmed it needs some special conditions to measure balanced output amps,here for more:

Beta Test: Multitone Loopback Analyzer software

What is considered most representative for a FR? A Log Chirp or say a (periodic) white noise at say 8k and with a thousand averages? A log chirp with a large FFT (up to 8M if you want to wait) may give you higher frequency resolution, but for measuring crossovers either one should work fine.

www.audiosciencereview.com

 

[dasdoing]

done that, it is amazing how less is really audible. But still as with going for crazy high SINAD and 24bit when available (although I hardly hear the difference to 256kbit AAC, if at all), better be on the sure side and go for the maximum, if it is easily available.

you don't understand the consequences.
Do the test on a smartphone.
music has to sound good (and loud) on those, too