Doppler distortion doesn't make sense (Edit: It actually does!)

[Fluffy]

In his Youtube channel, Paul Mcgowan from PS Audio has talked many times about Doppler distortion. He describes it as an artifact of single driver speakers when a higher frequency is modulated by a lower frequency with the movement of the diaphragm that produces both of them simultaneously. The distortion occurs in the same way that the pitch of an engine sound coming from a fast moving vehicle rises and falls as the vehicle comes towards and then away from the listener. In the speaker, the slow movement of the diaphragm producing the low frequency changes the pitch of the high frequency as it moves forward and backward.

This doesn't make any sense to me. First of all, from the perspective of the waveform, the only way for two waves to be produced simultaneously is so that one modulates the other.

That's just how it works, and this shouldn't result in any form of nonlinear distortion, because the frequency components are the same.

From a physical perspective, the soundwave moving through the air always contains these kind of inter frequency modulations, regardless if the source is a single driver or multiple drivers with a crossover. When two drivers produce different frequencies, the end result is a complex pressure wave traveling through the air that contain all the frequency components. A single driver produces the resulting pressure wave on its own right from the get go, and multiple drivers will each contribute its own frequency band that will also result in the same pressure wave. It's not like two drivers each create a separate pressure wave that arrives to the ear independently through separate mediums.

And finally, from a physiological perspective, we only have one ear drum in each ear. It doesn't matter how many sound sources there are, the pressure wave that arrives to the ear and gets picked up by the eardrum contains all the frequency components, along with their inter modulations.

Am I completely wrong about this?

 

[DonH56]

Doppler distortion, or what I learned decades ago as frequency modulation distortion (FMD), in speakers has been debated endlessly but I am not a speaker designer. Without stepping into its audibility, you must remember that the basis of the argument is that the two signals are not independent and the modulation is motional and not electrical. If you sum two pure tones electrically and without distortion, an FFT will show only the two tones. In a speaker, there is that same sort of summing and modulation, but in addition you have the speaker that is generating the sound moving back and forth and that adds additional frequency components due to the Doppler effect. The frequency changes as the source moves back and forth, away from and toward the listener, just as the sound (pitch) changes from a car or train as it approaches then passes and fades away. An FFT will show frequency spreading (modulation) due to the movement of the source. In your example, the 1.5 kHz source is being moved back and forth by the 40 Hz modulation of the driver (assuming one driver is responsible for generating both signals), thus should exhibit spreading.

Again, this is not from just summing two signals, but from physically moving the signal source, thus creating Doppler shift.

The more separated are the two frequencies, the more likely you are to hear any such distortion. If they are near in frequency the difference is less and it is difficult to impossible to resolve the spread in pitch. That is the argument used for multiple drivers handling separate frequency bands. More complex signals (like music) also make it harder, natch. However, the modulation is also a function of displacement and thus driver area, thus full-range systems like planar dynamic speakers and ESLs exhibit Doppler distortion to a lesser degree than smaller cone drivers and get away with covering more frequencies with a single or fewer number of drivers.

At least, that is my hand-waving off-the-cuff explanation. It is real, but how much it matters for practical designs is debatable. Other factors like directivity (beaming) and distortion also make it hard for a single small'ish driver to span the full audio frequency range so FMD is (as usual) one of many design trades.

HTH - Don

 

[Fluffy]

What about headphones? If I understand you correctly, the very tiny movement of a headphone driver makes Doppler distortion insignificant, despite having a full range driver.

And another question, would the effect be exaggerated, maybe to the point of audibility, if one would take a woofer with large excursion (maybe even a subwoofer, but without the low pass filter), and play through it a very low frequency at very high volume (=large displacement), coupled with a high frequency? Say something like a woofer playing a combination of 15hrz at 120 dBSPL together with 1khz at 85 dBSPL. Would it be possible then to perceive the rapid change in the pitch of the 1khz due to Doppler effect?

 

[gene_stl]

An even better question is why would anyone imagine that a single driver can produce a ten octave range of sound/musical frequencies.

Organs have tiny little pipes that are whistles that are a few inches long and small in diameter. They also have pipes that are 64 feet long and almost a foot in diameter. One for the high frequencies and one for the low frequencies. For the same reason that there are ContraBasses Basses Cellos Violas and Violins.

The notion of the magical full range speaker is one of those audiophile legends that has a great following. They do work great if you 1) don't mind missing the top two octaves because of beaming and the bottom three octaves because they can't move enough air (and especially without creating Doppler or FMD) 2) you don't mind being limited to a small room and or low sound pressure levels and 3)you only ever listen to female vocals playing a single guitar.

Crossovers are at a very high state of development. Not having crossover networks is one of the reasons that full range driver types cite. However the inventions of woofers midranges and tweeters happened for very good reasons and the development of crossovers followed apace. They also like to imagine that single drivers are some kind of point source. They are not.

Doppler or FM Distortion is only one of many reasons to not use a "full range"single driver system. there are several others.

Some of these distortions enter into the way we hear since we only have a single tympanic membrane in each ear. So they may sound natural.
I don't know whether there are two way and three way earphones since I don't really follow that technology. But for sure the "room" that an earphone transducer is driving is so small that its mechanical displacements are much less.

 

[KSTR]

Doppler distortion is considered an issue by two german manufacturers of active speakers with motional feedback where that feedback can improve driver distortion down to neglegible levels so that Doppler distortion may be the dominating the error in a stressed 2-way with a woofer moving significantly.
Some Silbersand speakers have an all-analog solution to properly pre-distort the signal both for 2-way and 3-way designs, whereas Schanks Audio has taken the approach to do the correction in the digital domain with the on-board DSP.

 

[RayDunzl]

In the speaker, the slow movement of the diaphragm producing the low frequency changes the pitch of the high frequency as it moves forward and backward.

This doesn't make any sense to me.

Does it make sense when you hear the horn (tweeter) of the car (woofer) moving toward or away from you (gross example)?

 

[NTK]

The simplest 1-D wave is describe by the following equation:

The way to understand at this equation is, when you are at position x=0, which is at the source, u(t, x=0) = sin(ω t).
If you are at, say 1 unit distance away, you'll get u(t, x=1) = sin(ω (t - 1/c)). This means at x=1 you'll see the same wave but it is delayed by time 1/c (i.e. it takes 1/c amount of time for the wave to travel from x=0 to x=1).

Now, if the source of the wave is moving, for example, away from you at a speed of v, this equates to x = vt. The equation becomes:

You can see that, the frequency has effectively reduced by the ratio (1 - v/c). Therefore, when the sound source is moving away, the "observed" frequency is reduced, and vice versa — the well known Doppler effect.

For a speaker with concentric drivers, the outside driver diaphragm acts as a reflector for the inside driver. This causes the "apparent" source location of the inside driver to move back and forth. Couple this to the above equation, the listener will see frequency fluctuations and thus the Doppler distortion.

 

[gene_stl]

The simplest 1-D wave is describe by the following equation:
View attachment 33460

The way to understand at this equation is, when you are at position x=0, which is at the source, u(t, x=0) = sin(ω t).
If you are at, say 1 unit distance away, you'll get u(t, x=1) = sin(ω (t - 1/c)). This means at x=1 you'll see the same wave but it is delayed by time 1/c (i.e. it takes 1/c amount of time for the wave to travel from x=0 to x=1).

Now, if the source of the wave is moving, for example, away from you at a speed of v, this equates to x = vt. The equation becomes:

View attachment 33471
You can see that, the frequency has effectively reduced by the ratio (1 - v/c). Therefore, when the sound source is moving away, the "observed" frequency is reduced, and vice versa — the well known Doppler effect.

For a speaker with concentric drivers, the outside driver diaphragm acts as a reflector for the inside driver. This causes the "apparent" source location of the inside driver to move back and forth. Couple this to the above equation, the listener will see frequency fluctuations and thus the Doppler distortion.

What he means is that not only are single full range drivers problematic but also , so are coaxial concentric multi way drivers which are often a corollary to the single driver dream.

 

[Fluffy]

Well than I guess it's real. Good to know. Now how audible is this effect actually? More audible than THD? Than IMD? Is there a reason to apply digital or analog correction to resolve the issue? And what about my question regarding headphones and their full range driver?

 

[NTK]

What he means is that not only are single full range drivers problematic but also , so are coaxial concentric multi way drivers which are often a corollary to the single driver dream.

I can't see how a single driver is susceptible to Doppler distortion (unless of course not when the whole speaker is shaking violently).

Well than I guess it's real. Good to know. Now how audible is this effect actually? More audible than THD? Than IMD? Is there a reason to apply digital or analog correction to resolve the issue? And what about my question regarding headphones and their full range driver?

This is an inherent negative to coaxial drivers. But it didn't stop KEF or Elac or others to make really good sounding speakers. May be the benefit of a coincident sound source is enough to compensate for this weakness. Or may be the audible effect of Doppler distortion is overblown.

 

[MC_RME]

Well than I guess it's real. Good to know. Now how audible is this effect actually?

I have test signals of sines in the 3 kHz range that are mixed with low frequency 20 Hz and lower. These easily showcase all kinds of intermodulation distortion including Doppler on headphones, when the 3 kHz changes its frequency (!) when raising the bass part.

That said the Doppler effect is inaudible in normal signals (voice, music). Intermodulation distortion in the mids caused by strong low frequency levels usually show up as amplitude modulation, long before Doppler becomes audible. YMMV, though.

 

[Cosmik]

I'm not seeing a simple explanation of this phenomenon here yet. There's no need to resort to maths or anything like that. Simply consider a speaker cone producing a 20 Hz tone, say, and a 1 kHz tone at the same time. The vibrations of the cone needed to produce the 1 kHz tone are riding on top of the vibrations needed to produce the 20 Hz tone. This is no different from a speaker producing a 1 kHz tone mounted on top of a vehicle that is rapidly driving forwards and reversing i.e. the same as a siren on a vehicle. Why should it be any less real in an audio system?

Intuitively, it will be worse for smaller, high displacement bass drivers, and it will be worse for drivers that cover a wider frequency span, so a two-way speaker will suffer more from it than a three-way all else being equal.

Here's an article on it, and how real it is:
https://www.stereophile.com/reference/1104red/index.html

 

[Blumlein 88]

Just one more reason using multiple subs is a good thing. Less excursion.

 

[Krunok]

I'm not seeing a simple explanation of this phenomenon here yet. There's no need to resort to maths or anything like that. Simply consider a speaker cone producing a 20 Hz tone, say, and a 1 kHz tone at the same time. The vibrations of the cone needed to produce the 1 kHz tone are riding on top of the vibrations needed to produce the 20 Hz tone. This is no different from a speaker producing a 1 kHz tone mounted on top of a vehicle that is rapidly driving forwards and reversing i.e. the same as a siren on a vehicle. Why should it be any less real in an audio system?

That is not how drivers work - they don't play according to Fourier components, their movement follows summation waveform.

 

[Fluffy]

I'm not convinced this is actually an audible problem. I tried listening to a combination of 20hz and 3khz tones with my lcd2c headphones, with the volume cranked high so the bass displacement is high, and the higher tone is lower by 20 db than the lower one, so to not be completely ear piercing. I first listened to the pure 3k tone, and then added the 20 hz tone, and I didn't perceive a difference to the higher tone. Not that this proves anything, but it at least shows that this is not an easily perceptible phenomenon.

I'd like to see some study conducted showing how audible this effect is in test signals and also standard music, compared to other forms of distortion.

 

[Krunok]

I'm not convinced this is actually an audible problem.

Probably not. IMHO it manifests as another kind of IM distortion.

 

[March Audio]

I have tested this myself. Not commenting on audibility, but measuring with a microphone you do get the low frequency as sidebands around the higher frequency

 

[Fluffy]

Ok forget about what I said before. I just tried it again. This time with the Focal Clear, which has a smaller and stiffer 40 mm driver, compared to the 80 mm of the lcd2c (which means more excursion for the same spl, I suppose). this time I tried a combination of 80 hz (because the clear doesn't like frequencies well below 50 hz), and 6k. I played the 6k tone, and as soon as I added the 80 hz I could hear a lot of very clear sidebands to the higher frequency. I also tried combination of other very low and very high frequencies and they also were clearly audible. The same combination of tones on the lcd2c results in a much cleaner upper frequency.

I don't know if this is in fact Doppler distortion, or some other kind of IMD that's unrelated (I made sure I'm not crossing 0 Dbfs with the combining waves, obviously).

 

[Krunok]

Ok forget about what I said before. I just tried it again. This time with the Focal Clear, which has a smaller and stiffer 40 mm driver, compared to the 80 mm of the lcd2c (which means more excursion for the same spl, I suppose). this time I tried a combination of 80 hz (because the clear doesn't like frequencies well below 50 hz), and 6k. I played the 6k tone, and as soon as I added the 80 hz I could hear a lot of very clear sidebands to the higher frequency. I also tried combination of other very low and very high frequencies and they also were clearly audible. The same combination of tones on the lcd2c results in a much cleaner upper frequency.

I don't know if this is in fact Doppler distortion, or some other kind of IMD that's unrelated (I made sure I'm not crossing 0 Dbfs with the combining waves, obviously).

Are you saying that you have the same driver producing 80Hz and 6kHz?

 

[Fluffy]

The headphones only have one driver, so yes.