• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Speakers distortion

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,071
Location
Zg, Cro
I believe it is safe to say that THD and IMD are no longer an issue with well designed DACs and amps. Thank God these days you no longer need to sell your kidney to buy such products as was the situation before. But how it is with speakers? How does THD and IMD of modern speakers, say Revel Performa3 F206 for example, compare to the similarly priced speakers from 15-20 years ago?

I would expect directivity to be noticeably improved over these years, but right now I would kindly ask your opinion on THD - how much has it been improved? :)
 
Last edited:
Not so significantly IMHO. There have been some incremental improvements in driver design, mostly resulting from improved computer modelling, but the materials haven’t changed considerably, nor the basic principles. Horn-loading remains the best way to reduce nonlinear distortion, but is practical in a domestic context only in the high frequency range, and centre-to-centre spacing issues typically arise from horns capable of loading in the low-mid frequency range.

So some improvement, sure, but not a lot.
 
I think it depends a lot on which specific loudspeakers one compares.

The Quad ESL63, for example, has distortion below 0.3% above 200Hz at 96dB SPL, but about 3% at 50Hz or thereabouts
The KEF 104.2 has 2nd Harmonic distortion below 0.1% and 0.3% third harmonic at 2kHz and 86dB SPL.
The Yamaha NS1000 is around 0.3-0.5% 2nd harmonic at 96dBSPL above 200Hz, with third harmonic below 0.1%. AT 86dBSPL, it's around the 0.2% above 200Hz.

The numbers come from Martin Colloms's HiFi Choice 1985 book.

One difficulty in comparing is that methods of measuring differ, especially as much of today's loudspeakers are measured at 90dBSPL.

I measured my current B&W 801 actives and compared them to the previous Meridian DSP5000s which were still there at the time.
Distortion.JPG
,

The measurements were done at home using tones and don't allow for room gain at different frequencies, so are only indicative, not definitive.

S.
 
Not so significantly IMHO. There have been some incremental improvements in driver design, mostly resulting from improved computer modelling, but the materials haven’t changed considerably, nor the basic principles. Horn-loading remains the best way to reduce nonlinear distortion, but is practical in a domestic context only in the high frequency range, and centre-to-centre spacing issues typically arise from horns capable of loading in the low-mid frequency range.

So some improvement, sure, but not a lot.

Interesting thoughts.. Why do you think modern midrange drivers are not cone horn loading drivers instead of smaller bass versions that are widely used?
 
I think it depends a lot on which specific loudspeakers one compares.

The Quad ESL63, for example, has distortion below 0.3% above 200Hz at 96dB SPL, but about 3% at 50Hz or thereabouts
The KEF 104.2 has 2nd Harmonic distortion below 0.1% and 0.3% third harmonic at 2kHz and 86dB SPL.
The Yamaha NS1000 is around 0.3-0.5% 2nd harmonic at 96dBSPL above 200Hz, with third harmonic below 0.1%. AT 86dBSPL, it's around the 0.2% above 200Hz.

The numbers come from Martin Colloms's HiFi Choice 1985 book.

One difficulty in comparing is that methods of measuring differ, especially as much of today's loudspeakers are measured at 90dBSPL.

I measured my current B&W 801 actives and compared them to the previous Meridian DSP5000s which were still there at the time.
View attachment 19513,

The measurements were done at home using tones and don't allow for room gain at different frequencies, so are only indicative, not definitive.

S.

What those titles in the header mean - what are "Passive Spec", "S/N6143" etc..?

From what I can see modern sepakers have really prestty much the same figures as the ones you posted. That doesn't leave much hope they would sound significantly better, does it?
 
Interesting thoughts.. Why do you think modern midrange drivers are not cone horn loading drivers instead of smaller bass versions that are widely used?

Don’t quite understand, sorry. What do you mean by “smaller bass versions”? :)
 
Going three-way (or more) instead of two will drastically reduce various types of distortion (including Doppler).

DSP and active crossovers make such speakers practical and easy to create, therefore even if driver technology doesn't change, DSP reduces speaker distortion - indirectly.
 
Don’t quite understand, sorry. What do you mean by “smaller bass versions”? :)

Why aren't midrange drivers cone instead of membrane (as bass drivers are)? :)
They could be horn loaded if they were cone, right?
 
Going three-way (or more) instead of two will drastically reduce various types of distortion (including Doppler).

DSP and active crossovers make such speakers practical and easy to create, therefore even if driver technology doesn't change, DSP reduces speaker distortion - indirectly.

How could DSP reduce distortion of the single driver without feedback? Or you mean indirectly by using 3 drivers instead of 2, which is easy with DSP and active crossovers?
 
https://www.soundstage.com/index.php?option=com_content&view=article&id=16

There are some measures of THD+N here. Some Revels are below .5% (the chart cuts off at close to what would be .5%). Don't know how much lower. This at 95 dpSPL or in some cases 90 dbSPL. The level of performance is available from other good speaker designs. N dominates the measurement at lower frequencies.

Here is the F206 specifically.
https://www.soundstage.com/index.ph...&catid=77:loudspeaker-measurements&Itemid=153

1546012194380.png
 

Attachments

  • 1546012174004.png
    1546012174004.png
    49.6 KB · Views: 145
Last edited:
How could DSP reduce distortion of the single driver without feedback? Or you mean indirectly by using 3 drivers instead of 2, which is easy with DSP and active crossovers?

Yeh the latter is what I think @Cosmik was alluding to. By having more control over the crossover slopes it’s possible to minimise driver excursion. It’s also possible to do things like high-pass filter the bass driver(s) below resonance, for example. And with full control over phase correction and the ability to use sharper xover slopes, it’s easier to incorporate more drivers into a single speaker with each driver reproducing a narrower range of frequencies, which can also tend to minimise distortion.
 
Why aren't midrange drivers cone instead of membrane (as bass drivers are)? :)
They could be horn loaded if they were cone, right?

Still really confused here! All cone drivers have membranes, in fact the cone in a cone driver is its membrane ;)
 
Certainly if we go back 40 years instead of 20 there have been huge improvements. Most speakers in the 70's were designed by engineer/hobbyists using their knowledge and ears. I remember visiting A/D/S in the early 80s and talking to their speaker engineers who were tremendously excited about using laser interferometry to actually view and measure cone flex & resonance.

It is surprising to me that since then basic driver design....that is use of dynamic cones, horns, ribbon and AMT tweeters has not changed much. But the details have changed a lot due to improved measurement and materials. I see that the Revel F206s offer "motor-system distortion-reduction mechanisms that stabilize the flux field during operation". Whether that means ironless motors or another mechanism...that can potentially reduce driver distortion by nearly half.

Similarly, the F206 "integrated acoustic lens waveguide based on a breakthrough mathematical approach" probably does not impact on-axis anechoic performance at all...but can greatly improve in-home performance.

BUT- irrespective of the important but incremental over time changes to passive speakers, I think Cosmik's reference to DSP and active equalization is spot on. Whether built into active speakers, or as part of the electronics, DSP can mean speakers perform far better in actual homes than was true in the past. Anechoic measurement of distortion and frequency response is just half the story.

The market for passive speakers is rapidly diminishing as demand for smaller, smarter, active speakers increases. Some of them to my audiophile ears sound terrible. But, not all of them do. And nearly all the good ones sound far better than similarly sized speakers of 15 years ago.
 
Last edited:
Could you elaborate on this pls @Blumlein 88?
With thd plus noise if noise is high enough it obscures distortion. At 200 hz and less it is difficult to isolate sounds and those low frequencies travel great distances. So usually the distortion levels in that range are more noise than anything else. You even see this using REW for measurement. You'll see it in graphs that Ray posts.
 
https://www.soundstage.com/index.php?option=com_content&view=article&id=16

There are some measures of THD+N here. Some Revels are below .5% (the chart cuts off at close to what would be .5%). Don't know how much lower. This at 95 dpSPL or in some cases 90 dbSPL. The level of performance is available from other good speaker designs. N dominates the measurement at lower frequencies.

Here is the F206 specifically.
https://www.soundstage.com/index.ph...&catid=77:loudspeaker-measurements&Itemid=153

View attachment 19530

I've seen that one, but pity they didn't show distortion below 200Hz. I guess it would be more interesting to seee how much it is in the region where ear is most sensitive.

This is how "silence" currently looks in my room:



My guess is that 50Hz and its harmonics are there from fan in the notebook and refrigirator from the kitchen, or ..?

Lower freq noise is more elevated than I thought it would be. Is that why you said that below 200Hz distortion is noise dominated?
 
Last edited:
With thd plus noise if noise is high enough it obscures distortion. At 200 hz and less it is difficult to isolate sounds and those low frequencies travel great distances. So usually the distortion levels in that range are more noise than anything else. You even see this using REW for measurement. You'll see it in graphs that Ray posts.

Ok yes, I thought you were talking about those Revel measurements you posted, which are from the NRC’s anechoic chamber. Very unlikely to be much noise affecting the measurements there!
 
Still really confused here! All cone drivers have membranes, in fact the cone in a cone driver is its membrane ;)

Huh, I'm struggling to find words here.. :p
Why isn't the midrange driver similar in design to high freq driver, like the ones Dynaudio used long time ago? These days all of manufacturers for midrange driver seem to use similar design to bass driver, but smaller. Why is that?
 
What those titles in the header mean - what are "Passive Spec", "S/N6143" etc..?

From what I can see modern sepakers have really prestty much the same figures as the ones you posted. That doesn't leave much hope they would sound significantly better, does it?
Sorry if it was confusing. I was a small spreadsheet I created with the distortion measured on my two 801s, hence the two serial numbers, as originally passive, then after I removed the passive crossovers and drove them active. The spot-frequency distortion didn't change a lot, and I didn't measure intermod distortion, which I would expect would be improved by going active, but don't know for sure as I didn't measure it. I also included the comparison with my Meridian active 'speakers.

What I was trying to convey was that some vintage loudspeakers already had fairly good distortion figures, although there are some modern loudspeakers, like Revels, which have exceptionally low distortion. Overall, as I've seen from the reviews in HiFi News when I used to subscribe, loudspeaker distortion hasn't radically reduced.

Where I think modern loudspeakers are worse that vintage ones, is in the flatness of their frequency response. Far too many modern loudspeakers have a rising HF response, (listen to all that detail!) and a frequency response that looks like a cross-section of the Alps. Some of the worse are high-efficiency loudspeakers intended to be used with low power valved amplifiers.

S.
 
Sorry if it was confusing. I was a small spreadsheet I created with the distortion measured on my two 801s, hence the two serial numbers, as originally passive, then after I removed the passive crossovers and drove them active. The spot-frequency distortion didn't change a lot, and I didn't measure intermod distortion, which I would expect would be improved by going active, but don't know for sure as I didn't measure it. I also included the comparison with my Meridian active 'speakers.

What I was trying to convey was that some vintage loudspeakers already had fairly good distortion figures, although there are some modern loudspeakers, like Revels, which have exceptionally low distortion. Overall, as I've seen from the reviews in HiFi News when I used to subscribe, loudspeaker distortion hasn't radically reduced.

Where I think modern loudspeakers are worse that vintage ones, is in the flatness of their frequency response. Far too many modern loudspeakers have a rising HF response, (listen to all that detail!) and a frequency response that looks like a cross-section of the Alps. Some of the worse are high-efficiency loudspeakers intended to be used with low power valved amplifiers.

S.

Thank you, I find your data very interesting!

And how exactly do you measure distortion? How do you compensate for distortion of the mic you're measuring with?

Btw, F206 linearity measurement that @Blumlein 88 posted looks very good to me and it doesn't seem to suffer from the rine in HF effect you're mentioning.
 
Back
Top Bottom