Well, I have been pondering over how to correctly measure the speaker driver distortion with Omni Mic kit from Dayton Audio.
I've done some basics measurements at 85-90dB SPL levels but when I read the Dayton Help section (sort of manual), the suggested SPL is >125dB in order to run the distortion test correctly.
I am not in the slightest mood to run the 89dB driver to 125dB which will ultimately destroy the driver. But I understand the reason why the manual suggests to test the speaker at such high SPL for some non-linear distortions only appear when everything is pushed to the limitation of the driver.
But do we really need to consider that kind of non-linear distortion in the how good (or horrible) a speaker would sound under most sane levels listening?
Aside from the Omni Mic I have, I might be able to try other measuring method. Is there any other kind of distortion measurements you folks use?
And it is my understanding that the driver distortion is best tested in Free Air Loading (OB) and not inside a box! Am I getting this all upside down?
TIA!
I've done some basics measurements at 85-90dB SPL levels but when I read the Dayton Help section (sort of manual), the suggested SPL is >125dB in order to run the distortion test correctly.
I am not in the slightest mood to run the 89dB driver to 125dB which will ultimately destroy the driver. But I understand the reason why the manual suggests to test the speaker at such high SPL for some non-linear distortions only appear when everything is pushed to the limitation of the driver.
But do we really need to consider that kind of non-linear distortion in the how good (or horrible) a speaker would sound under most sane levels listening?
Aside from the Omni Mic I have, I might be able to try other measuring method. Is there any other kind of distortion measurements you folks use?
And it is my understanding that the driver distortion is best tested in Free Air Loading (OB) and not inside a box! Am I getting this all upside down?
TIA!
Got that backwards -- level should be LESS THAN 125dB (at higher, the mic itself distorts too badly). 90dB or 100dB is a good level, less is ok. You just have to be above the where the distortion products will be above the back ground noise.
Most types of distortion in loudspeakers are SPL-dependent (distortion level increases as the SPL increases). Thus, to be able to do comparisons, you need to calibrate and measure at a specific SPL and distance (like 95 dB at 1 m).
If you need an absolute evaluation, measure at the highest level you are typically listening.
If you need an absolute evaluation, measure at the highest level you are typically listening.
bzfcocon is right, though -- a distortion measurement without a displayed or specified level doesn't mean much. I was just referring to the ranges where you can use OmniMic. Above 125dBSPL won't work (and even at 120dB is getting sketchy).
Thanks everyone for the input! I will follow the suggestion and measure at around a more comfortable SPL for the drivers but here is what I read from the Omni Mic manual.
They wanted you to run the speaker at very high levels but I am assuming the speaker as a complete speaker and not a single driver.
They wanted you to run the speaker at very high levels but I am assuming the speaker as a complete speaker and not a single driver.
That last sentence was intended as a thing to not do.
You want the mic close to avoid echoes/reflections but watch out that the levels, when that close, don't get too high or you'll be measuring the mic rather than the speaker.
Instructions didn't want you to run at very high levels, they wanted things arranged so that direct signal is considerably higher than reflected -- turning up the volume has no effect on that ratio, only mic and speaker placement do.
You can go as low in level as you want, provided you are above ambient noise for all the distortion products you are trying to see. But most meaningful level would be around where the speakers will normally be used.
You want the mic close to avoid echoes/reflections but watch out that the levels, when that close, don't get too high or you'll be measuring the mic rather than the speaker.
Instructions didn't want you to run at very high levels, they wanted things arranged so that direct signal is considerably higher than reflected -- turning up the volume has no effect on that ratio, only mic and speaker placement do.
You can go as low in level as you want, provided you are above ambient noise for all the distortion products you are trying to see. But most meaningful level would be around where the speakers will normally be used.
Gotcha Mr. Waslo! I hope it's the case also.
I guess I can ask you more about the Omni Mic in the future? I find it very easy to use and does a lot of things for me.
If "correct" includes some indication of perception then you have an even bigger problem. The latest AES Journal has a paper on the "best sounding nonlinearities" - they add distortion intentionally to make the bass "sound better".
My suggestion: learn how to take good frequency response measurements at off-axis angles. That's more important than THD.
89db is the sensitivity of the driver, not the spl "capacity". If the driver has enough travel and enough power it will do 125 or whatever without being destroyed.
In case you haven't guessed, BWaslo is the Omnimic.
In case you haven't guessed, BWaslo is the Omnimic.
I always found it intriguing as to what distortion is good and acceptable and desirable.
I think most low sensitivity drivers used and designed for the Home Audio environment with low distortion have a hard time coping to re-crate the bass guitar distortion without compression.
The higher sensitivity bass driver in the Pro Audio environment is designed to handle that but most in the Audiophile industry / community seems to steer away from the use of the PA drivers.
Yes, 89dB is the sensitivity of the speaker and I normally don't listen above 95dB peaks.
I have the amps to drive any speaker above 120dB level but I have a feeling I better ask before I bring the horror upon myself.
Yes, I figured BWaslo is the person whose name appears on the Omni Mic software so I appreciate him for chiming in.
When I saw High SPL >125dB, I thought "wait a min, this is confusing and can't be right". And then, my brain went into over drive mode and started thinking about Omni Mic being calibrated and may be able to measure 120dB with very low inaccuracy. I thought the higher SPL requirement is probably one if it's ability to measure the non-linear distortion at very high SPL and hence started this thread.
Well, yes, but measuring distortion does have its uses, even if not good for estimating sound quality at roughly-linear range levels. Such as finding out if the speaker or driver is able to usefully produce enough SPL. For instance, if the distortion is normally under a percent, but suddenly is jumping high when the driver can't move far enough. A fault that can happen when working with too-small woofers trying to get too low too loud.
Though IM distortion is probably a better way to test for that (there is way to measure IM with sweeps, but not implemented in OM at least yet).
OmniMic can do that, too, and has some interface features to make it pretty easy.
HI Bill
Yes gross changes in distortion are an indication that something has gone wrong, but thats about it. I now, like Hawksford, deliberately design nonlinearity into my drivers. It actually makes them sound better, but not for the reasons that you might think. It raises the "good" distortion and lowers the "bad" - a lot like collesteral.
On you polar maps, I read a great paper on the use of colors in engineering. It said that color should convey the same thing to the eye as is intended in the graph. Yellow is the "loudest" color, after white. So I go from white to yellow to green to blue to black which is a progression that is identical to the perceived levels of light intensity. I use red for "hot spots" - spots that are greater than 6 dB above the average level. I find this easy to read at a glance. But, of course when you normalize the data, as you do, then it looses its meaning.
Yes gross changes in distortion are an indication that something has gone wrong, but thats about it. I now, like Hawksford, deliberately design nonlinearity into my drivers. It actually makes them sound better, but not for the reasons that you might think. It raises the "good" distortion and lowers the "bad" - a lot like collesteral.
On you polar maps, I read a great paper on the use of colors in engineering. It said that color should convey the same thing to the eye as is intended in the graph. Yellow is the "loudest" color, after white. So I go from white to yellow to green to blue to black which is a progression that is identical to the perceived levels of light intensity. I use red for "hot spots" - spots that are greater than 6 dB above the average level. I find this easy to read at a glance. But, of course when you normalize the data, as you do, then it looses its meaning.
Normalization is just an option, done for looking at drivers that have not had any compensation added yet. But I understand about your color choices. After looking at various other packages' color schemes, I'm pretty used to how to read them, so went with their schemes. Maybe I should make that an option (though, as someone once told me, 'you give too many buttons and wheels and before long no one will be able to drive it anymore')
I think it is not about what sounds good but what sounds right. If not you are in polluted waters. Malcolm into good sounding nonlinearities ? I do not think that he is already into Alzheimer or i am not up to date with his latest visions.
More second, less third ? : just put some DC into the coils, in the positive or negative direction depeds mostly on the suspension design if it has any.
More second, less third ? : just put some DC into the coils, in the positive or negative direction depeds mostly on the suspension design if it has any.
Is it this one :http://www.essex.ac.uk/csee/researc...C141 Room Correction Using Synthetic Bass.pdf
I have to digest it first if i understand it.
I have to digest it first if i understand it.
Earl,
Very interesting comment. Could you elaborate a bit on "good" vs "bad" distortion. Quite serious. Are you talking linear, compressive, harmonic, time relative? Are the things you are tuning electrical or mechanical?
Could any of these parameters be related to how various amps behave with the speaker? I have observed a measurable difference between amps when injecting a perfect pulse starting at zero crossing and the acoustic output to be delayed further on some than others. I have not built a current probe good enough to see what is really going on.
Very interesting comment. Could you elaborate a bit on "good" vs "bad" distortion. Quite serious. Are you talking linear, compressive, harmonic, time relative? Are the things you are tuning electrical or mechanical?
Could any of these parameters be related to how various amps behave with the speaker? I have observed a measurable difference between amps when injecting a perfect pulse starting at zero crossing and the acoustic output to be delayed further on some than others. I have not built a current probe good enough to see what is really going on.
I have several papers on the perception of distortion and they are on my website. Hawksford references these in his paper. There are good and bad nonlinearities and the key is knowing what and when to use them. I guess that I would consider a lot of this a trade seceret other than its all pretty well spelled out in the papers.
- Status
- Not open for further replies.