so you think loudspeakers are linear eh!

Jump to Latest
#63
On the first page of this article linked a few posts back,
http://www.four-audio.com/data/MF/aes-swp-english.pdf
is this
"Capturing binaural room impulse responses for high-quality auralization purposes requires a signal-to-noise ratio of >90dB which is unattainable with MLS-measurements due to loudspeaker non-linearity but fairly easy to reach with sweeps due to the possibility of completely rejecting harmonic distortion."
 
#64
Yep. Signal-to-noise ratio is the important bit, though.

Let's say there's 40dB of background noise - nearby traffic, air conditioning, whatever.

That means you need to get 130dB into the microphone to get >90dB SNR.

Not many speakers are linear at that sort of SPL.

Chris
 
#71
Pano said:
Yikes! If you are hitting 11 volts into a pair of 93dB/W speakers in a domestic setting, that's going to be loud. To be fair, the loudest I've ever had to go was 42V peak with 83 dB@2.83V speakers. And that was in my lava cave that soaked up sound like a sponge. Yeah, I still wanted more. 🙂

11V RMS where the tests are run would be 15.55 V peak. I think out of a 93dB driver that would be close to the levels I had with 42V on a 83 dB driver.*
I can understand wanting more, but for me an 11V sweep is enough to know how a driver will perform in the region that I use it.

*check my math
Click to expand...

I hear where you're coming from...and yes, I get the same math.

I guess i just have a hard time accepting rms response curves as indicative of peak capability.

My amps have rms and peak meters, so it's fun to watch them bounce around...especially since it's for 4-way active.
The vast difference in voltage swings between the sub, mid, HF, and VHF drivers is kinda amazing. Especially how low rms levels can be higher in frequency, and how high peak levels can be lower in frequency. I of course realize there are different driver sensitivities in play, but it sure is clear how much more voltage is needed down low.

I've come to believe there is no real way to put a meaningful rms level on music.
So the only voltage I pay attention to is peak (other than for thermal limiting).
And i try to make measurements, sine sweeps, stepped sine, or CEA2010 at peak levels to find out when to say no to SPL increases.

I'd be really interested in any distortion plots or compression sweeps like chris661 did, for your 42V 83dB @2.83V driver, if you have any.
I have to think it was probably showing/sounding some duress....but it would be nice to be mistaken.🙂

From a pure listening point of view, I get intrigued how sometimes what sound loud, reads a lower SPL than I'd expect. It's always from a smaller 2 or 3-way rig, home or PA.
And conversely, how dang scary high the SPL meter can read, without sounding loud, on a bigger 3 or 4 way. So I blame it on distortion...

So anyway, I dunno...i just know loud is no good if not clean, and as far as i can hear, gear doesn't fully hold up even if increasing response curves aren't showing any compression.

Oh, don't mean to sidetrack...but it is directly related, since much of this relates to peak vs rms.
I've noticed how when 2 sine waves are summed, rms voltage increases by 3dB as expected, no matter what the 2 frequencies are.
However, peak voltage increases by 6dB, unless the sine waves are the exact same frequency in which case it's 3dB, just like rms.
As more frequencies are summed, the greater the spread between peak and rms. Not sure where it ends but guessing somewhere less than 12dB peak to rms. (I've measured 4 sines summed at about 9dB peak to rms.)

The point of all this to me, is even with just steady tones, I probably need 12dB of totally clean uncompressed output headroom....without accounting for spiky transients.
Mark believes in headroom !!! 😀
 
#73
mark100 said:
...My amps have rms and peak meters, so it's fun to watch them bounce around...especially since it's for 4-way active.
The vast difference in voltage swings between the sub, mid, HF, and VHF drivers is kinda amazing. Especially how low rms levels can be higher in frequency, and how high peak levels can be lower in frequency. I of course realize there are different driver sensitivities in play, but it sure is clear how much more voltage is needed down low.

I've come to believe there is no real way to put a meaningful rms level on music.
So the only voltage I pay attention to is peak (other than for thermal limiting).
And i try to make measurements, sine sweeps, stepped sine, or CEA2010 at peak levels to find out when to say no to SPL increases.

I'd be really interested in any distortion plots or compression sweeps like chris661 did, for your 42V 83dB @2.83V driver, if you have any.
I have to think it was probably showing/sounding some duress....but it would be nice to be mistaken.🙂


So anyway, I dunno...i just know loud is no good if not clean, and as far as i can hear, gear doesn't fully hold up even if increasing response curves aren't showing any compression.

Oh, don't mean to sidetrack...but it is directly related, since much of this relates to peak vs rms.
Click to expand...
Meters on amps, unless if they are true "Vu"-spec meters have ballistic characteristics, that are unpredictable. And if Vu, then predictable but wrong on peaks. An oscilloscope or REW, I think, can tell peaks. If LED, there's a chance they are telling you truly about brief peaks and the limits to your amp's capability. But the RMS reading may be accurate but with music, means very little.

While we test continuous sine waves and continuous waves are what drains an amp's power supply when sustained, it is ability to handle peaks that is or more interest since there are brief peaks on non-over-cooked music always.

The distortion versus power curve should reveal how much excursion the driver makes as output grows and that should reveal how it handles peaks of that same power level. We don't want to bump into soft-clipping (unless in a club, deafened by the loudness, and drunk).

Which leads me to advocate using multi-driver tweeter arrays with lots of peak power handling because you'll hear the clipping on Brahm's Lullaby with a women's chorus.

So that's all in the sine-wave world. Bigger fish to fry in transient behaviour that motional feedback addresses. But I have never heard of a way to measure transient behaviour that seems useful to me.
 
Last edited:
#75
mark100 said:
I guess i just have a hard time accepting rms response curves as indicative of peak capability.
Click to expand...
They ought to be much more stressful than music signals, considering that a sine RMS is just 3dB below peak. Recorded music is 14-22dB average below peak. That's a lot.

I've come to believe there is no real way to put a meaningful rms level on music.
Click to expand...
There is! At least on digital music files. My favorite wave editor GoldWave will tell you the RMS value of a music file or any selection therein. Also peak values, but those are usually 0dBFS or very near. I'd be glad to help if you want to know. From that you should be able to figure RMS into your speakers for a given volume setting.
I'd be really interested in any distortion plots or compression sweeps like chris661 did, for your 42V 83dB @2.83V driver, if you have any.
I have to think it was probably showing/sounding some duress....but it would be nice to be mistaken.
Click to expand...
You are not mistaken. The poor little things sounded very stressed when I was right up at the baffle. Truly bad. Strangely 20 feet back into the room they sounded just fine. Not sure I know why. :scratch2:

That said, they weren't running 42V flat out, not even 30V. Maybe 4-5V RMS on music with peaks around 36V. I did once make the mistake of running a sine sweep at 30V RMS and the 3.5" fullrange (tweeter) went Pop! Burned right up. Was it linear up to that point? I doubt it. 😛
 
#78
bentoronto said:
Meters on amps, unless if they are true "Vu"-spec meters have ballistic characteristics, that are unpredictable. And if Vu, then predictable but wrong on peaks. An oscilloscope or REW, I think, can tell peaks. If LED, there's a chance they are telling you truly about brief peaks and the limits to your amp's capability. But the RMS reading may be accurate but with music, means very little.

While we test continuous sine waves and continuous waves are what drains an amp's power supply when sustained, it is ability to handle peaks that is or more interest since there are brief peaks on non-over-cooked music always.

The distortion versus power curve should reveal how much excursion the driver makes as output grows and that should reveal how it handles peaks of that same power level. We don't want to bump into soft-clipping (unless in a club, deafened by the loudness, and drunk).

Which leads me to advocate using multi-driver tweeter arrays with lots of peak power handling because you'll hear the clipping on Brahm's Lullaby with a women's chorus.

So that's all in the sine-wave world. Bigger fish to fry in transient behaviour that motional feedback addresses. But I have never heard of a way to measure transient behaviour that seems useful to me.
Click to expand...

Ben, the meters are digital readouts on a computer screen....Q-Sys Designer software running a qsys Core processor

A number of properties can be set with regard to what they measure and display. I keep one channel of the processor hooked to a scope that I can route any signal flow to, for verification.

Personally, I find the term structure of transients fascinating, particularly with regards to amplifier ability to meet their term structures.
I have no idea how well speakers then really keep up...would make a good study.
Guess I need to learn a wave editor like Pano suggested, and make some recordings to compare against electrical transients.
 
#80
Thanks. The Neumann are lovely, but I'll never be able to fit a pair in my ears to measure HRTF. I've damn near tried, tho. 🙂

FWIW, the lowest self noise mics that I know of are the Lewitt. Not sure about SNR.
 
You must log in or register to reply here.
Top Bottom