Why isn't THD a regular part of speaker measurements?

[Fluffy]

I understand that room acoustics can make this measurement very unreliable. But I would imagine that when measuring in an anechoic chamber or just placing the mic 1 inch from the driver, you can get pretty accurate results. This is a regular measurement in headphones and basically all types of electronics, so what makes it less applicable to speakers?

 

[sergeauckland]

I suspect that the reason manufacturers don't often publish distortion figures or graphs is that they're so much worse than electronics, that the numbers would scare off a typical buyer who may not understand fully what these mean.

Subjective reviews don't publish any numbers at all, but reviews in Stereophile and HiFi News usually do include distortion at 90dB SPL.

S

 

[MZKM]

On the manufacturer end: Speakers distort a lot more than headphones typically. Some companies like KEF post THD, only for higher end models though.

On the reviewer side: the SoundStage/NRC measure THD.

I would assume Stereophile doesn’t because he measures things in his backyard in NY, so it would have to be loud to get enough range (his waterfall plots are only like 20dB of range, sometimes even less), which also means annoying the hell out of his neighbors.

 

[Fluffy]

I've seen very little reference to distortion in Stereophile measurements. In soundstage there are indeed pretty detailed graphs, though they have about 50 DB of range, so most speakers there have distortion way below that and it's difficult to tell them apart. In HiFi News the measurements seem incomplete and very broad, so again it's hard to differentiate. Overall I wouldn't say that's a parameter that one can realistically rate speakers by, drawing from the available information.

 

[Juhazi]

I don't remember which, but some German magazine publishes distortion or power compression measurements. Mostly we see those for pro monitors.

It is good to know where and when bass gets in trouble, and sometimes it is the tweeter or mid. DIY people need distortion info the most to select drivers used to get the spl required.

 

[briskly]

just placing the mic 1 inch from the driver, you can get pretty accurate results

This is a far less than ideal method of taking measurements of non-linearity. This will place a great burden on the mic+pre to stay acceptably linear at very high SPL. I would venture a guess that this affects some of the LS driver measurements in Voice Coil's test bench.

Also, most headphone measurement systems are somewhat noisy. This affects the results taken at innerfidelity, for instance.

 

[anmpr1]

Measuring speaker distortion is not a trivial matter, since there are questions of what frequency and power levels to use. Richard Heyser was one of the first (maybe the first) to include comprehensive distortion analysis in his reviews. I've copied a section of a review he did so you can get a feel of how it could be done (or at least how he did it). There is more, with graphs, and this is only a small excerpt. But you can get the idea.
__________
Measured harmonic distortion for the frequencies of 41.2Hz, 110Hz, and 262Hz... correspond to the musical tones of E1, A2, and middle C. I chose middle C rather than A4 (440Hz) because the actual acoustic crossover from woofer to mid-range is slightly above 250Hz, and I wanted to measure the distortion for the same driver at all frequencies.

Low bass harmonic distortion progresses smoothly from a few tenths of a percent at 100mW to near 10% at 60 average watts, with second harmonic slightly above third harmonic throughout the whole range.

Mid-bass harmonic distortion at 110Hz does not look at all like the low-bass distortion. Mid-bass harmonic level stays essentially uniform and of low level throughout the entire drive range, as the woofer really likes the acoustic load in this important frequency range, and pumps out acoustic power with little distortion.

Up near the crossover the harmonic distortion rises uniformly with drive level, although its total level is very low even at a thundering 100+ dB SPL.

Intermodulation of middle C and E1, when both are mixed in equal proportions, is impressively low. Music played at a 10 watt average level is reproduced at SPL which many other fine loudspeaker systems simply cannot reproduce, yet IM remains below 4%. The nature of this IM is principally amplitude modulation of middle C by the lower tones up to about 50 average watts; at 100 watts (ear protection level) the IM measures 12.88% and has picked up phase modulation of 6 degrees peak to peak on middle C, in addition to about 8% peak to peak amplitude modulation.

 

[MZKM]

I've seen very little reference to distortion in Stereophile measurements. In soundstage there are indeed pretty detailed graphs, though they have about 50 DB of range, so most speakers there have distortion way below that and it's difficult to tell them apart. In HiFi News the measurements seem incomplete and very broad, so again it's hard to differentiate. Overall I wouldn't say that's a parameter that one can realistically rate speakers by, drawing from the available information.

Human trials have shown that THD [EDIT: with music] that’s roughly 40dB down (1%) is what’s audible for the upper treble, around 20dB down at 500Hz, and around 10dB down at 100Hz.

THD stacks, and if assuming worst case scenario where the distortion is identical, if your DAC, pre-amp, and amp are operating at -60dB THD (0.1%) and the speakers are -50dB, that would add up to ~ -44.2dB, so just around the threshold of audibility. If we change the DAC to a -110dB, the preamp to -100dB, and the amp to -90dB, which is more realistic operating conditions, that changes it to ~ -49.7dB, which would be hard for most people to hear. Also keep in mind this is worst case scenario of stacking.

 

[direstraitsfan98]

Buy low distortion speakers!

 

[anmpr1]

Buy low distortion speakers!

In the context of amps, there are no low distortion speakers. There are only lower distortion speakers relatively speaking. And the lowest distortion speakers are not everyone's cup of Joe in the morning. Plus, they tend to be big and expensive. One day (not in my life) someone may really create a viable distortionless speaker.