Basic Distortion Measurements REW

a4eaudio

I have come across several threads where someone pretty knowledgeable about speaker building makes the comment that it is hard to do good distortion measurements. I assume the key here is "good", given DIY tools vs more advanced undertakings.

My particular DIY tools consist of a Dayton Audio EMM-6 mic, preamp, and REW. Is it really just an adequate SPL level and distance, or is there more to it than that? I can take a measurement in REW and simply show distortion, so is this not good enough?

If I simply want reasonable but informative distortion measurements using typical DIY tools, what would be a standard process?

dcibel

There isn't so much of a standard process, but more of a "what do you want to know?". It's often not easy to compare distortion results directly, as differences in mic used and test conditions can affect the results greatly.

First problematic area is unfortunately the EMM-6 mic. Just like cheap speakers, cheap mics will have high distortion of their own, primary reason I sold the EMM-6 was it presented quite high 2nd order distortion results, making speaker evaluation and any comparison to other measurements a bit problematic. Noise floor is another factor, where a higher quality mic will provide a lower self-noise for being able to see the low distortion results above the noise floor.

That said, a few tips for measuring distortion. One key factor is to maintain some known reference for being able to interpret the results. With EMM-6, absolute SPL can be a challenge, but taking a voltage measurement of the voltage applied to the speaker is easy, to provide provide some point of reference for comparison. Hificompass provides distortion measurements at several voltages, which you then have to interpret to actual SPL based on driver sensitivity, so depending on the driver you may compare 2V to 5V measurement for an equal SPL comparison.

Measurement distance can be "semi-far field". 1m doesn't make much sense, baffle diffraction is not really part of this puzzle. The balance here is that the closer the mic is to the speaker, the less influence the room interaction has on the measured results, however once you get too close to the speaker it becomes "near field" affecting frequency response on both the low and high end. Another consideration is max SPL capabilities of condenser mics. Most start to hit limitations at 120dB, with max SPL in the 125-135dB range, so placing a mic too close can run into these limitations more easily. About 315mm is often used as a happy place for many speakers, and makes for easy translation of SPL at mic to 1m by simply adding 10dB, so 110dB at 1m would be 120dB at the mic. Hificompass mentions 315mm distance provides accuracy down to 150Hz for his setup.

A few common standard SPL used for testing is 86 and 96dB@1m, to show average distortion at fairly normal SPL and for dynamic peaks.

Other than that, very low background noise is important for distortion measurements, much more so than for frequency response. Since harmonic content can be 0.1% (-60dB) of the fundamental, any background noise >0.1% of the signal can be presented as distortion. Taking a measurement with no signal to show just background noise is useful to see where the noise floor for your system and room are.

a4eaudio

Thanks for the explanations.

Knowing what third-party tests show, I should have low distortion drivers. However, the RS-225 aluminum cone has significant breakup by 5Khz. I tried to design my xo such that at least 3rd order distortion is down 40db. So all I'm really trying to do is test the final speaker and see what I achieved.

Below is distortion measurement with the EMM-6 at 94db/1m. (My poor little refurbished Dayton APA150 couldn't get it any louder LOL.) So, do I interpret your comment about the EMM-6 and high 2nd order distortion to mean that the red line below probably overstates the 2nd order distortion a little bit?

(I set the vertical axis to "dBr", is there a preferred way to present the graphs?)

dcibel

@a4eaudio said:
So, do I interpret your comment about the EMM-6 and high 2nd order distortion to mean that the red line below probably overstates the 2nd order distortion a little bit?

Yes, that's what I found, it was quite significant, though I don't have any specifics beyond that, I didn't save any direct comparisons once I got my current mic (Line Audio OM1) and saw that distortion made a lot more sense and noise floor was lower too.

(I set the vertical axis to "dBr", is there a preferred way to present the graphs?)

Similar to the polar map "normalized or not" question, dBr and % selection in REW provide a normalized representation, makes it easy to see the relative distortion to the fundamental, but you sort of need to know the amplitude of the fundamental (frequency response) for complete interpretation. So if you ask me, normalized distortion is my preference for evaluating raw driver performance, along with FR for complete interpretation. For a complete speaker with filtered response, the un-normalized distortion plot is a better representation of actual performance, but takes a bit more effort to determine actual distortion for a given frequency.