Crest Factor ... how does one use it?

[Don Snyder]

When using HornResp, we typically use sine waves, which are much harder on a speaker than music. Several people have made comments about "Crest
factor", but no one has explained if it can be used to predict the power requirements of music.

Assume that we have a speaker with an x-max of 10 mm. Playing music that contains a 40 Hz fundamental, deflecting the speaker 9 mm and an 80 Hz 2nd
harmonic deflecting the speaker 3 mm, then the speaker would be distorting. Would the distortion be noticeable?

[Don Hills]

Such a signal as you specified, with a 2nd harmonic louder than the fundamental (and presumably higher harmonics also present) is going to sound distorted before any contribution from the speaker.

[bob91343]

If the cone reaches its limit of travel, you will hear distortion.

[soho54]

That is not how it works. Crest factor has nothing to do with Hornresp sims really.

Crest factor is just the difference between the average of a wave, and its peak value. It is something to be worried about in live sound, to avoid clipping, but it's not a big deal really.

If we are talking about a steady sine wave, the average SPL level is -3dB below the actual waves peak level/cone movement. The crest factor would be 3dB. In real content the average for a select portion may be -20dB, but there could be peaks up to -10dB in that average. This gives you a crest factor of 10dB. The crest factor is higher, but the sine wave has the potential to get quiet a bit louder, as the real content would begin to clip earlier, do to it's larger crest factor.

Hornresp shows you the max displacement levels already, so crest factor is irrelevant. The amount of power needed to hit xmax/xmech is the same, no matter how you get there.

[Don Snyder]

Don Hills: It's common for musical instruments (bass guitar, tympany) to have the tonic depressed, and they don't sound distorted. Ever hear of the
missing fundamental?

Bob: Agreed, but that's not what I was after.

Soho: In estimating maximum power for music, one could assume that no single sinewave was more than 1/3 or 1/2 the total energy at a given time.
This would not be a good assumption for synth, but would work for most music. Limiting the power to maximum displacement predictions (x-max)
would be safe ... but booooring. Exceed x-mech at your peril!

Anyone for a little DSP?
~Don

[jbell]

Originally Posted by Don Snyder

Don Hills: It's common for musical instruments (bass guitar, tympany) to have the tonic depressed, and they don't sound distorted. Ever hear of the
missing fundamental?

Bob: Agreed, but that's not what I was after.

Soho: In estimating maximum power for music, one could assume that no single sinewave was more than 1/3 or 1/2 the total energy at a given time.
This would not be a good assumption for synth, but would work for most music. Limiting the power to maximum displacement predictions (x-max)
would be safe ... but booooring. Exceed x-mech at your peril!

Anyone for a little DSP?
~Don

Don - your line of thinking was I believe in davygrvy's thoughts when he designed his radar feedback loop. Instead of trying to figure out and limit based on a set of 'rules or predictions' to figure out when x-max was exceeded, he chose the measurement route.

you have some interesting dsp thoughts to share? maybe a real time distortion analyzer that tracks input and output and limits based on %distortion?

[sreten]

Originally Posted by Don Snyder

When using HornResp, we typically use sine waves, which are much harder
on a speaker than music. Several people have made comments about "Crest
factor", but no one has explained if it can be used to predict the power
requirements of music.

Assume that we have a speaker with an x-max of 10 mm. Playing music that
contains a 40 Hz fundamental, deflecting the speaker 9 mm and an 80 Hz 2nd
harmonic deflecting the speaker 3 mm, then the speaker would be distorting.
Would the distortion be noticeable?

Hi,

I guess the answer to your question is that excursion requirements
must be averaged over the bass range and they become far more
severe the louder and lower you want to go.
Cerst factor on most recordings in the bass is very low, its compressed
for fairly obvious reasons, for highly dynamic low bass is very hard to
reproduce peaks and still keep the average audible, see loudness curves.

rgds, sreten.

[soho54]

Originally Posted by Don Snyder

Soho: In estimating maximum power for music, one could assume that no single sinewave was more than 1/3 or 1/2 the total energy at a given time.
This would not be a good assumption for synth, but would work for most music. Limiting the power to maximum displacement predictions (x-max)
would be safe ... but booooring. Exceed x-mech at your peril!

Peak power is peak power there is no time constraint involved. If it take 3000w to hit maximum safe excursion,or the xmech for an instant it will take the same amount of peak power to do it all night long even at a near DC pulse rates.

There are no hidden power needs, or situations. This has nothing to do with crest factor though. The area under the curve, so to speak does not have increased power needs. The only question is whether the amp can output the power needed all night, or in more than instantaneous bursts.

If you can't keep your speakers under the danger zone in the nude, you need more speakers, or you need to get some good compression and limiting.

[Don Snyder]

Jim: I make cartoons of airplanes and speaker cabinets. This radar feedback loop stuff I'll gladly leave to Davygravy.

sreten: Interesting thoughts. Tapped Horns have got me thinking again, and now we've got that low "Bo" string on electric bass ... and there's Synth on
half the recordings.

I hate compression ... it sucks the life out of music. Sure, a little touch of it here and there solves some problems, but there's no dynamics in today's
music.

Soho: Good stuff, you're right of course.

Thanks guys,
~Don