Optimizing speaker distortion

[dzeekee007]

Thanks so much Uli,
That guidelines is very helpful

[Labdoc]

Question: Where should I locate the microphone when measuring speaker distortion and doing speaker linearization?

Mitchco suggests "Linearize each driver by measuring 30cm on axis to each driver, check distortion characteristics, and measure the linearized drivers for verification." (https://audiophilestyle.com/ca/ca-acade ... lkthrough/)

On the other hand, in his book, Mitchco states "10 cm on axis" for measuring each driver.

The Acourate Wiki (German version) is silent on where to place the measurement microphone (https://www.audiovero.de/acourate-wiki/ ... tsprechern)

[UliBru]

in ideal case a speaker driver would be measured without room reflections. A close-up measurement will thus emphasize the driver sound and suppress the room sound.
But the driver sound is also influenced e.g. by driver and loudspeaker geometry, e.g. edge diffrations. Or think about a horn speaker or a dipole speaker. So at the end there is no general rule for the measurement distance.
I have made good experiences using a distance of about 40 cm with standard speakers. With dipole speakers I even use the listening position.
This is for driver linearization.

For measurements of the harmonic distortion the distance is not so critical. At least as long as the room does not add some distortions by vibrating or rattling items. Usually it is more difficult to remove distortions. So the measurement is mainly used for information purposes. The correction filters do not correct for them anyway.

[Labdoc]

Uli,
Thanks for this quick and practical guide to a complex topic (measuring loudspeakers).
After a long gestation, I now have the time and speakers + amps to do some measurement and linearization experiments.
Murray

[Labdoc]

Like @Niels, I have an old pair of speakers in the basement: Celestion DL10's. I removed the original passive crossover for which the specification is XO1 530 Hz and XO2 3800 Hz. Then I set up to measure distortion of the MF driver following the procedure in the Acourate Wiki and Mitch Barnett's web article. (https://www.audiovero.de/acourate-wiki/ ... tsprechern and https://audiophilestyle.com/ca/ca-acade ... lkthrough/ ).

I placed the measurement microphone on axis with and 40 cm from the MF driver. In Acourate, I performed a logsweep from 400 Hz to 4400 Hz, bracketing the crossover points of the original passive crossover (530 and 3800 Hz). I cleared the resulting curve, copied "Documents\Acourate\LogSweep\LogSweep48_rec.wav" and "Documents\Acourate\LogSweep\Inverse48.dbl" into the working directory, loaded these into Acourate Convolve LogSweep(measurement) with Inverse and clicked LogSweep Convolution. This worked as described in the procedure method.

However, on the next step, Extract Pulse and Harmonic Distortion, I get an error message and the distortion extraction fails. The steps and error are in the attached graphics.

Uli, can you help troubleshoot please.

Murray

Convolution Setup

LogSweep convolution setup

1_LogSweepConvoSetup.jpg (32.93 KiB) Viewed 3882 times

Select Centre

2_Select_Centre.jpg (151.13 KiB) Viewed 3882 times

Extract Setup

3_Extract_Setup.jpg (33.84 KiB) Viewed 3882 times

Error on Extraction

4_Error_on_Extraction.jpg (5.29 KiB) Viewed 3882 times

[Labdoc]

My problem has disappeared! Using the same LogSweep data as before, but marking a wider area in the time window before Extraction, seems to have fixed the error.

Now I would like to ask for help in interpreting the distortion curves. My novice impression is that the distortion of the MF driver is relatively high. If true, would measurement of a narrower frequency range offer improvement? For example, the speaker was used for 530-3800 Hz reproduction with the passive crossover. I measured 400-4800 Hz. Is it useful to repeat the LogSweep with 500-3500 Hz, for e.g. ?

5_Distortion_400-4800_LogSweep.jpg (168 KiB) Viewed 3879 times

[UliBru]

A distortion measurement is always connected to the playback level. So your result would be very great at a level of > 120 dBSPL
It makes sense to change the min-max-limits of the chart to zoom better into details.
So far I can see that the brown k2 curve has a distance of about 50 dB to k1, equivalent to 0.3%. This is not bad at at.
k3 is weaker, its size is about -40 dB = 1%. But still not bad.

Repeating the sweep with 500-3500 Hz will not change the result. But you can easily try.

[Labdoc]

Hi Uli,
An expanded view shows that the highest distortion is 40 dB between k1 and k3, or 1%, as you noted.
The volume under test was 75 dB as I recall, measured by SPL meter.
What is your impression about the linearity? Is it worthwhile to linearize this MF speaker, assuming the HF and LF drivers behave similarly ?

6_Expanded_Distortion_400-4800_LogSweep.jpg (128.12 KiB) Viewed 3870 times

[UliBru]

The linearity is about +-3 dB. As it is no problem to apply a linearization it is IMO worth a try. You then can verify the behaviour with a new measurement with applied linearization.

[In-Ear]

One question,

why is it such an effort to get distortion results with AC? Couldn´t all necessary graphs be stored directly in according working directory and you just have to push a bottom to get distortion measurement result such as in REW or other programs? I really find me avoiding to use this function in AC as it is a mess to make quick distortion measurements here..maybe worth to think about it in an future update?

Jörn