Some speaker driver measurements...

Thanks for the measurements :cheers:

Have you considered equalising the response of the drivers to flat (or a reference slope) before running HD sweeps? It would help to make comparisons between drivers with different natural rolloffs. E.g. consider that the BlieSMa T34 is about 6dB down at 1kHz compared to >2kHz, which makes the distortion considerably lower at 1kHz compared to a tweeter which is flat down to 1kHz. Of course it is possible to look at the measurement which is twice the voltage to account for -6dB at 1kHz but I find it more intuitive if the levels are the same at every frequency to begin with, that way you can just open one plot for each tweeter and compare them side by side.

Up until now I've only been measuring midrange drivers which are more or less flat over their usable bandwidth so differing frequency responses hasn't been too much of an issue for me, but I'm planning to do woofers and tweeters soon and this seems like a big hurdle in comparing the distortion of different drivers as they near their mechanical resonances. I'm considering perhaps providing additional measurements with tweeters EQ'ed to 1kHz/2kHz LR2 and woofers to 40Hz BW2.
The solution seems obvious to me, yet for some reason no one seems to have ever done it.
 
Last edited:
:(
Thanks for the measurements :cheers:

Have you considered equalising the response of the drivers to flat (or a reference slope) before running HD sweeps? It would help to make comparisons between drivers with different natural rolloffs. E.g. consider that the BlieSMa T34 is about 6dB down at 1kHz compared to >2kHz, which makes the distortion considerably lower at 1kHz compared to a tweeter which is flat down to 1kHz. Of course it is possible to look at the measurement which is twice the voltage to account for -6dB at 1kHz but I find it more intuitive if the levels are the same at every frequency to begin with, that way you can just open one plot for each tweeter and compare them side by side.

Up until now I've only been measuring midrange drivers which are more or less flat over their usable bandwidth so differing frequency responses hasn't been too much of an issue for me, but I'm planning to do woofers and tweeters soon and this seems like a big hurdle in comparing the distortion of different drivers as they near their mechanical resonances. I'm considering perhaps providing additional measurements with tweeters EQ'ed to 1kHz/2kHz LR2 and woofers to 40Hz BW2.
The solution seems obvious to me, yet for some reason no one seems to have ever done it.

In fact, I had been thinking about this feature when making the technical task for my web designer... I declined it for the next reasons:
- it is more time consuming process
- after this feature have been implemented the problem with measurements at high voltage levels will appear due overloading at low end of frequency range. So it will require the next additional changes during measurements.
- for that purpose I give the measurements at different voltage levels
- I have broken apart a couple of very pricy tweeters due errors (my and DSP), so no, no again:)

p.s. SS D2904/7140 voice coil burns for less than 2 sec at voltage level 14 Volt:( I forgot to turn on HPF in DSP after measuring voltage had been set up at 200 Hz frequency.
 
Last edited:
Fair points. It is quite a bit more work, especially playing around with the DSP to reach a target response, and ensuring that the DSP doesn't negatively effect the performance of the measurement system.

As far as I know the measurements display the harmonics as relative to the fundamental, so that's already taken into account.
The problem is that the level of the harmonics varies non-linearly with the level of the fundamental. Generally, if the level is raised 6dB, the harmonics will rise by more than 6dB, therefore rising slightly relative to the fundamental. If the level is lowered 6dB, the harmonics will fall by more than 6dB, therefore falling slightly relative to the fundamental. This is intuitive - when a driver plays louder, the sound produced is more distorted because the diaphragm has to move further and therefore all forms of distortion related to excursion increase.

Without EQ, a tweeter with a response like the BlieSMa T34B is effectively being tested at a level that is 6dB lower at 1kHz compared to 2kHz+. This gives it an advantage in a distortion sweep because the harmonic distortion will be slightly lower for frequencies below 2kHz compared to if it had a flat response. This advantage goes away as soon as you apply some sort of crossover/filter to shape the frequency response into a target slope because now any tweeter shaped into the same response is on an even playing field.

Therefore in order to compare the T34B to say the SB29RDC (which has similar sensitivity but flat response to 1kHz), for frequencies above 2kHz you could compare 2.83V data for both drivers, but to compare distortion at 1kHz you need to compare the 2.83V data from the SB29RDC to the 5.6V (+6dB w.r.t. to 2.8V) data for the T34B. There's nothing inherently 'wrong' with testing unfiltered responses, but the resulting data requires a bit of interpretation for the above reason. Without test data at multiple voltages/levels like HifiCompass publishes, you can't always conclude which driver has better HD when the frequency responses diverge. Measurements at the right levels also need to be available to accomplish this. For example, we don't know what the distortion is on the T34B @ 1kHz when it is at the same level as the SB29RDC at 5.6V, because an 11.2V sweep isn't available for the T34B.
 
Last edited: