yes, maybe not realistic levels with full scale orchestra. but nevertheless, with my 83 dB w/m speakers 20-30W is pretty much party level and with heavy metal for instance I can't understand a person sitting next to me unless he/she shouts. also mind that in Europe the average room might be smaller than in US and that could be a factor too.
This is not the case here. Everyone is aware of the difference between dynamic range and loudness. The 'loudness' in this test is subjective and is what is used to determine the point at which one sets their volume control. The maximum of the dynamic range is defined as being 0dB digitally and is used to determine whether or not your amp will ever clip given the largest dynamic possible of the digital format.
Here is the part that you're missing.
No one is using a sine wave as a signal by which to set the volume control, the sine wave is an objective 'test', 'measurement' or 'control' signal only.
I think Pano's choice of sine wave signal level could perhaps be what's creating the confusion. For want of a better explanation, because I cannot read his mind, Pano chose to use a signal at -12dB or whatever because this wouldn't put stress on your system and would be unlikely to cause your loudspeakers to be destroyed. The less complicated way to have done this would be for him to have used a sine wave played at 0dB but that could have been destructive.
One listens to a piece of music. For the sake of this test it would make sense for someone to use their most dynamic piece of music with the largest average to peak ratio. They then set their volume control to the loudest that they would ever listen to this piece of music at.
After having done that they play Pano's sine wave at -12dB and then take a measurement on their volt meter of the amplifiers output. Lets say it reads 5Vrms.
Now we used a test signal that was 12dB below the maximum possible for digital, so we now compensate for that by multiplying the voltage 4 times as 20log(4) = 12dB. Having multiplied this by 4 we end up with 20Vrms.
What this tells you directly is that if you were to feed a 0dB sine wave through the system, the output on your amplifier would = 20Vrms.
20*20 = 400 / 8 = 50. Or 50 watts into an 8 ohm load. The users amplifier is capable of 150 watts into an 8 ohm load, so providing the loudspeakers are an easy enough to drive 8 ohm load, we know that the amplifier isn't going to clip when the system is confronted with a 0dB signal at the given volume.
What we are interested in when doing this test is seeing if the amplifier clips when presented with a 0dB signal at your 'loudest ever' position on the volume control.
We could have scaled up the digitally created sine wave by 12dB initially and used this as the test stimulus but chose not to so as to protect the system and then account for this later on.
The trouble here, as you should now see, is that if we scale up the pink noise appropriately, like we could have done with the sine wave digitally instead, that the pink noise will now clip digitally as it will try to go above 0dB. The amplifier though will reproduce the digitally clipped signal faithfully however as it wont clip when 0dB is reached and this is what we were trying to find out originally.
If ones amplifier is free from clipping when a 0dB signal is sent through the hifi system at the loudest volume you will ever use, then it does not matter what signal you try and feed it, fireworks or shuttle launches. The amplifier will not clip.
Now lets say I did the test again, but this time I went about it using the shuttle launch. I set the volume at the start so that the level of the talking people is realistic. Then the shuttle launches, then my amplifier clips and the my loudspeakers explode. Yes, no one is trying to say this won't happen, of course it will. But no one here is trying to reproduce a shuttle launch accurately. They are trying to reproduce their most demanding piece of music at the maximum volume they will ever listen to it at. They then do a test to see if a 0dB digital signal will clip their system at this volume and find that it won't. Now they know that their system won't clip, ever, given their taste in music and listening habits.
On smart phone so responding is well ....
83db works out to 74 db from typical listening position or 80 db stereo best case , 86 db Would then require 8 watt stereo operation , which means 12 db peak would require 128,watts ....
B
50 watt amp just died ............
Last edited:
obviously, a.wayne disagrees.
if I get it correctly, he's saying that although for a specific volume setting and with 0dBfs sine wave an amp won't clip at any frequency, with that same volume setting the same amp will clip with certain signals.
I'm still not sure what's the source of the confusion that I somehow have the impression that part of it arises from the fact that many seem to share this view.
Really? Speakers are "most sensitive" at 120Hz or 240Hz? Are you sure about that? And even if they were, what difference would it make? Are you sure that you understand how the test works?
not sure I get it. my listening position is at about 2 meters from each speaker. as I said, small room.On smart phone so responding is well ....
83db works out to 74 db from typical listening position or 80 db stereo best case , 86 db Would then require 8 watt stereo operation , which means 12 db peak would require 128,watts ....
B
50 watt amp just died ............
later edit: it may also mean that I'm getting old. somehow the one zillion dB at the club I can take but at home I think 90 dB is party level.
Last edited:
No he's not kidding. This is one other way you could do the test. Find out at what volume your amplifier clips when presented with a 0dB digital signal. Lets say you've got a digital volume control and you get clipping occurring when it's set to volume 10, but no clipping occurring at volume 9. You have now found out that if you listen to the hifi system at volume 9 and lower that you will NEVER get your amplifier to clip, unless you attach a pair of loudspeakers that the amplifier cannot drive.
On smart phone so responding is well ....
83db works out to 74 db from typical listening position or 80 db stereo best case , 86 db Would then require 8 watt stereo operation , which means 12 db peak would require 128,watts ....
B
50 watt amp just died ............
That is half of the point of the thread though. You're quoting figures for what people are assumed to like listening at. This test actually gets people to listen, rather then using statistics and then go on and calculate the maximum amount of voltage swing necessary for their amplifiers given the maximum possible signal level (0dB digital) that the system could reproduce at their given volume setting. Most people aren't needing that much power, meaning most people probably don't listen as loud as you're thinking they will.
felt I should add that at the same time it doesn't tell me that any amp which is specified at V^2/4 (V being the clipping voltage) can do the job. some may go tits up if that is specified only on a resistive load. the results stand for my amp only.
It is if we're discussing the output voltage of the amplifier which is what this test is about. We are not discussing output power.
20log(V2/V1) = dB difference in voltages.
20log (4/1) = 20log(4) = 12dB.
Indeed and that is the limit of this test. It does not in any way go about handling the reactive nature of loudspeakers, on the whole this isn't a problem unless you've got a really feeble amp or killer loudspeakers.
Hi 5th
Actually I don’t see that I am missing anything in fact have spent a good bit of time examining the signals I wish to record and reproduce as well as what is required from the loudspeakers.
The problem is the reproduction system has a peak linear output, the maximum average level (which we hear as loudness all things equal) cannot be anymore than the peak level minus the peak to average ratio or one has instantaneous clipping. The method here is fine until the peak to average envelope exceeds 12 dB.
Add to that the fact that most speakers become dynamically nonlinear once above about 1/8 to 1/10 rated power and the problem seems pretty daunting but hey hifi speakers often don’t sound all that real do they?.
The fact is, to reproduce even everyday sounds, requires far more than home stereo speakers can produce. I have B&K sound level meter that can capture instantaneous peaks, just throwing a teaspoon on to a tile floor, produced a peak over 130 dB from about 8 feet. It didn’t sound loud at all because your ears don’t work like microphones and speakers and everything else in the chain and I guess that’s the problem in part, I am looking at the technical requirements not how short I can come and still sound ok.
To be clear too, the shuttle launch is as dynamic as many good music recordings but what makes it a speaker killer is the spectrum is loudest an octave or more below most hifi speakers low corner.
I am not sure why the great resistance to looking with a scope too, it's not like i am advocating something as heretical as "blind testing", just looking to see instead of assuming because you cant hear anything speacific all is well.
Best,
Tom
Actually I don’t see that I am missing anything in fact have spent a good bit of time examining the signals I wish to record and reproduce as well as what is required from the loudspeakers.
The problem is the reproduction system has a peak linear output, the maximum average level (which we hear as loudness all things equal) cannot be anymore than the peak level minus the peak to average ratio or one has instantaneous clipping. The method here is fine until the peak to average envelope exceeds 12 dB.
Add to that the fact that most speakers become dynamically nonlinear once above about 1/8 to 1/10 rated power and the problem seems pretty daunting but hey hifi speakers often don’t sound all that real do they?.
The fact is, to reproduce even everyday sounds, requires far more than home stereo speakers can produce. I have B&K sound level meter that can capture instantaneous peaks, just throwing a teaspoon on to a tile floor, produced a peak over 130 dB from about 8 feet. It didn’t sound loud at all because your ears don’t work like microphones and speakers and everything else in the chain and I guess that’s the problem in part, I am looking at the technical requirements not how short I can come and still sound ok.
To be clear too, the shuttle launch is as dynamic as many good music recordings but what makes it a speaker killer is the spectrum is loudest an octave or more below most hifi speakers low corner.
I am not sure why the great resistance to looking with a scope too, it's not like i am advocating something as heretical as "blind testing", just looking to see instead of assuming because you cant hear anything speacific all is well.
Best,
Tom
You've missed the point, again.
The -12dB that Pano has baked into the test has nothing whatsoever to do with peak to average levels. For all intents and purposes Pano could have set the test signal at -1dB and by that logic we've got no dynamic range whatsoever. As I explained above, the test signal is used as a measurement point and everything is scaled around it. The measurement point/signal could be at -100dB -80dB -50dB, it does not matter, the end result, ie the voltage requirement from the amplifier to reproduce a 0dB signal, would be identical each time.
The -12dB that Pano has baked into the test has nothing whatsoever to do with peak to average levels. For all intents and purposes Pano could have set the test signal at -1dB and by that logic we've got no dynamic range whatsoever. As I explained above, the test signal is used as a measurement point and everything is scaled around it. The measurement point/signal could be at -100dB -80dB -50dB, it does not matter, the end result, ie the voltage requirement from the amplifier to reproduce a 0dB signal, would be identical each time.
I'm not sure everyone here owns a scope. I don't.
instead a sound card can be used, they are good enough for this kind of measurement.
for instance I used the mic input on my laptop and a voltage divider. I couldn't believe how good it was, it's flat from 10Hz to 20k and all the distortion consists of a 2nd order harmonic at -80dB below fundamental (I checked it using my D/A converter which is known to be pretty good). using a RTA with FFT I can clearly see the point where all the harmonics abruptly rise which indicates clipping. just looking at the time display that is barely noticeable but a FFT tells a different story. I'll post some pics tomorrow.
the reason why Pano chose -12dbfs is that 0 dB would not be appreciated by the woofers given the chosen test frequencies and also your ears would hate it too (I should know as I once fed 20V sines into my speakers). btw I used my speaker's port resonance frequency for the high power test as the cone barely moves at that point. feeling the port output by hand and hearing the windows rattle is quite a show.
also, it should be obvious that all this is simplifying things to a point, for instance I don't think anyone pretends that their amp can maintain fixed voltage into any load at any frequency but that is really beyond the scope of the discussion and I think the calculated listening power is accurate enough for the purpose even ignoring the more subtle phenomena.