Who makes the lowest distortion speaker drivers

since you mentioned "reduced" my suggestion is definitely valid as well. Horns with compression drivers do definitley have less FM distortion but they have increased 2nd order harmonic distortion due to the nonlinearity of the air in the compression chamber.

Regards

Charles
Exactly. As Beranek derived, there is a mathematical way to express this distortion based on the underlaying physics. CD's distort by necessity. I am not aware of any such physics for piston drivers. Their distortion can be minimized to very low numbers. Besting electrostats, certainly at slightly elevated SPL.
 
Besting electrostats, certainly at slightly elevated SPL.
Since you ask, here's my ESLs and a Foster ribbon (XO around 3500 Hz). Domestic volume. Lotsa DSP. I would get lower THD+N readings, but I'd have to figure out how get all the cars off my street and the lawn mowers too.

Your turn.*

B.
*actually, no way to reliably compare one amateur REW plot with ancient equipment (like mine) to another plot. But I think mine looks pretty good.
 

Attachments

  • ESL and ribbon THD+N.jpg
    ESL and ribbon THD+N.jpg
    42.4 KB · Views: 66
But to return to the OP's question, I'd say the Rice-Kellogg cone driver is a poor concept and today's drivers are only incrementally better than 50 years ago. Sadly, I know of no better way of making bass.

However, if somebody would simply produce cone drivers with free-air resonance around 10 Hz, THAT would be a major improvement. Finally. That, at last, would put the speaker resonance outside the woofer passband, like for all tweeters and mid-ranges.

There has been major progress in addressing the anti-phase rear wave, esp the computation simplification published by T/S (based on models perfected long before). But the best audiophile designs - true horns, sealed or leaky boxes, and "infinite" baffling (including labyrinths like mine) - are many decades old.

The future of quality speakers lies in feedback. Just like in all other aspects of audio except speakers. You might say that can be in the design of new drivers because it can be achieved with built-in accelerometers or with special feedback coils on the voice coil former.

B//
 
Since you ask, here's my ESLs and a Foster ribbon (XO around 3500 Hz). Domestic volume. Lotsa DSP. I would get lower THD+N readings, but I'd have to figure out how get all the cars off my street and the lawn mowers too.

Your turn.*

B.
*actually, no way to reliably compare one amateur REW plot with ancient equipment (like mine) to another plot. But I think mine looks pretty good.

I'd say an array of the Rice and Kellogg drivers doesn't have to be ashamed of their results...

distortion in percent.jpg

Also a DIY attempt with REW with a calibrated microphone. And a busy road messing with the results! 25x 3.5" drivers running full range.
Measured at the listening spot at 2.7 m in a functional living room with DSP.
 
  • Like
Reactions: 1 user
True,
Browsing around in my many measurements from the origins of my project do show very similar results taken at a later date....
SPL calculated back to the standardized 1m distance is 86dB/1m.
distortion in percent-3.jpg

Again as measured at the listening spot in my living room at 2.7m distance. 2 years later and those same silly small 3.5" drivers.
I'm pretty sure others could present even better results using Rice and Kellogg drivers.
 
Distortion measurements don't say much without information about the SPL.
Distortion measurements don't say much on an absolute scale without a whole lot context to the testing besides just SPL. And when you think about it, not a lot that audiophiles do about THD. But it is a tool for an individual to use in their home comparatively.

My THD+N measurements get better the louder the test - because of the residual level in the testing context... which I can't usefully assess.

Frankly, I don't think THD is much of an issue today and with music systems. On the other hand, I have never found any measurement that helps assess transient performance ("fast bass") as with motional feedback R&D.

B.
 
True,
Browsing around in my many measurements from the origins of my project do show very similar results taken at a later date....
SPL calculated back to the standardized 1m distance is 86dB/1m.
View attachment 1078369
Again as measured at the listening spot in my living room at 2.7m distance. 2 years later and those same silly small 3.5" drivers.
I'm pretty sure others could present even better results using Rice and Kellogg drivers.
Those speaker curves are fabulously better than anything I have ever seen before on any occasion ever.

B.
 
True,
Browsing around in my many measurements from the origins of my project do show very similar results taken at a later date....
SPL calculated back to the standardized 1m distance is 86dB/1m.
View attachment 1078369
Again as measured at the listening spot in my living room at 2.7m distance. 2 years later and those same silly small 3.5" drivers.
I'm pretty sure others could present even better results using Rice and Kellogg drivers.
Very nice results. However, for comparison reasons I would like to see figures for 94dB @ 1 meter. It appears to be some sort of convention. For example, AudioXpress always measures domestic drivers at that level. I will take some of my own system over the weekend.
 
It appears that "distortion" here means nonlinear distortion. The fact is that nonlinear distortion in a loudspeaker is far far less important than linear distortion. And, in fact, there is no correlation between THD or IMD measurements and subjective perception. This kind of makes this whole discussion relatively pointless.

And to the comment of compression drivers being inherently distorted while direct radiators are not, this is incorrect. A direct radiator compresses the air in a nonlinear sense exactly like a compression driver and generates both IMD and FMD as a result. The absolute pressures are lower on the direct radiator meaning lower levels of nonlinearity, but then it also means lower levels of efficiency and hence the need for more excursion and hence more nonlinearity for the same SPL. My take on all this is that in the end a compression driver on a waveguide will have far less audible nonlinearity than a direct radiator. Both both are pretty negligible compared to linear distortion.
 
  • Like
Reactions: 1 users
But to return to the OP's question, I'd say the Rice-Kellogg cone driver is a poor concept and today's drivers are only incrementally better than 50 years ago. Sadly, I know of no better way of making bass.

However, if somebody would simply produce cone drivers with free-air resonance around 10 Hz, THAT would be a major improvement. Finally. That, at last, would put the speaker resonance outside the woofer passband, like for all tweeters and mid-ranges.

There has been major progress in addressing the anti-phase rear wave, esp the computation simplification published by T/S (based on models perfected long before). But the best audiophile designs - true horns, sealed or leaky boxes, and "infinite" baffling (including labyrinths like mine) - are many decades old.

The future of quality speakers lies in feedback. Just like in all other aspects of audio except speakers. You might say that can be in the design of new drivers because it can be achieved with built-in accelerometers or with special feedback coils on the voice coil former.

B//
Poor concept? If you say so. More conclusions.

How about the HST-15, it's got a Fs around 16 hz? What magic is there with having the free air resonance outside of the bandwidth ( or put another way, isn't 10hz in the bandwidth ? ) It should automatically be at the top of your list, no?

Also, once you put a driver in a box, the free air resonance is meaningless. ( unless you like a naked driver hanging in mid air )
 
Poor concept? If you say so. More conclusions.

How about the HST-15, it's got a Fs around 16 hz? What magic is there with having the free air resonance outside of the bandwidth ( or put another way, isn't 10hz in the bandwidth ? ) It should automatically be at the top of your list, no?

Also, once you put a driver in a box, the free air resonance is meaningless. ( unless you like a naked driver hanging in mid air )
+1 for gedlee, as usual.

Some Audio 101 needed here.

The Kellogg-Rice design shakes a heavy object to transmit vibrations into vastly lighter air. The air doesn't "speak back" to the driver and so there is no beneficial degenerative feedback correcting the massive cone assembly. Planar and ribbon drivers, by contrast, are pretty dead and are therefore much more compliant to the audio signal than a spring-loaded half-pound of moving mass... unless you have motional feedback.

When you have a big resonance within your passband (say 25 to 150 Hz for woofers), it makes the cone motion wacky, taking on a voice of its own (at the system resonance freq), and is strange to power.

With a driver with free-air resonance like 10 Hz, the wacky part is outside the passband even if the rear-wave "enclosure" raises it some. A 16 Hz driver is good but getting high because it will end up influencing sound in the passband. The so-called bookshelf AR-1 driver was 12-14 Hz and in the 1.3 (?) cu foot box maybe 35 Hz and an inspiration in 1960. Good for its day*.

B.
* mine is quite beat-up but I wish somebody would buy it
 
Last edited:
When you have a big resonance within your passband (say 25 to 150 Hz for woofers), it makes the cone motion wacky, taking on a voice of its own (at the system resonance freq), and is strange to power.
This depends on the Q of the resonance. If the Q is low then there is no "wacky" cone motion - the system just follows the signal exactly, lagging a bit which again depends on the Q.

When the Q is high then "wacky" things do happen. room modes are "wacky". (Like driving a room between two high Q modes, the sound will decay, not at the driven frequency, but at the frequency of the two modes. Weird huh!)

Driver modes are not "wacky" (in other words high Q, which should never be allowed in a driver.)

Another point should be realized here, and that is that in any result that involves a room, the room modes will dominate. (And don't get me started on LF sound perception in real rooms.)
 
Last edited by a moderator:
  • Like
Reactions: 1 users
Why is speaker distortion measurement very high? Maybe you use high distortion amplifier?
In this forum @xrk971 made some measurement:
https://www.diyaudio.com/community/threads/virtual-audition-of-very-simple-quasi-mosfet-amp.295286/

The measurements I've shown in this thread are taken at the listening spot in a room at 2.7 meter distance. They include
the room effects.
In contrast, XRK's measurement was semi anechoic and most probably taken about an inch from the cone or even less.
 
Hi Earl,

Apologies if this has been brought up it before.

We know that THD and IMD is a pretty blunt instrument.

What came of your GedLee metric? Have you collaborated with any software companies or authors to devise a way to measure a speaker eg. Fuzzmeasure, REW?
There was one company that implemented it, somewhere here on DIY. But there was apparently very little interest. People just seem blissfully happy to make THD and IMD measurements the old way. They think that it is "objective" data, which it is, except that it is poorly correlated to perception. That doesn't seem to matter.

To me it's like the drunk looking for his keys under a street light even though that's not where he lost them, "because the light is better."
 
  • Like
Reactions: 1 users