Asymmetric construction induced distortion in electrostatic speakers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In the past there has been some debate whether asymmetry in construction, e.g. the diaphragm not being equidistant to both stators, would result in increased distortion or not.
For example, Moray James ones proposed that second harmonic distortion could be raised by purposely adding asymmetry. thread on DIY

I thought it might be interesting to put this to the test in a real world ESL element: the Wachara headphone.

coating surface resistance: 1E7 ohms/square

First have a look at the second harmonic distortion:
blue trace: symmetric stators, d/s = 1mm.
yellow trace: asymmetric stators, d/s = 1 & 2mm

I compensated the output level for the decreased sensitivity of the asymmetric element (about 6dB).
 

Attachments

  • 2nd_order_asym_small.jpg
    2nd_order_asym_small.jpg
    68.5 KB · Views: 985
For the same conditions, here is the result for third order distortion

Only a slight increase here, nothing spectacular except for the weird dip around 500Hz and peak around 1.5kHz. I would not pay too much attention to the raise above 2kHz, this is present in all my 3rd harmonic measurements and I suspect it's an artefact of the measurement setup (microphone?).
 

Attachments

  • 3rd_order_asym_small.jpg
    3rd_order_asym_small.jpg
    67 KB · Views: 895
SY said:

Thanks!

What does the acoustic loading look like on each side of the diaphragm?

The element is suspended in air and during all measurements there was a thin felt mat on one side of the element (between the element and the microphone) to damp the main resonance.

I would think that this contribution to asymmetry would be just as important as geometry.

Interesting observation, but I'm not sure I agree with this line of thought. For example in this case there is a constructional asymmetry introduced by the felt, but would the element also experience this as asymmetric?

Acoustic loading on one side could make a difference if:

1. the diaphragm suffers from compression/expansion (in the 'thickness' dimension)
2. the loading itself is asymmetric, e.g. different if we 'push' or 'suck' air through it.

I assume that you would agree that point 1 is very doubtful. Point 2 sounds more plausible and would likely be a case if we had a woofer in an enclosure. But we are moving so little air here, that in this case I'm not so sure.

If you are interested I could do some measurements without damping, damping one one side and damping on both sides.
 
I had to do these measurements at a little higher output level (+9dB) because of background noise coming from outside of the house. So distortion levels may be a bit higher compared to previous measurements.


1. second harmonic, one side felt damping (blue) vs. no damping (yellow)
 

Attachments

  • 2nd_order_no_damping_vs_single_side_felt_small.jpg
    2nd_order_no_damping_vs_single_side_felt_small.jpg
    77.4 KB · Views: 862
Don't mention it.

I agree there's no sign of asymmetric acoustic loading induced distortion to be found in those graphs. A good thing since some well known ESLs use this technique.

We do see some reduction in distortion, mainly 2nd harmonic, with a little damping. IMD shows the same (not posted).

I'll be posting some interesting stuff next.
 
In post #1 we found a significant increase in second order harmonic distortion as a result of asymmetric diaphragm to stator distances.

Theoretically, we would not expect this increase in distortion in the ideal electrostatic panel. In the ideal case we have:

1. perfectly uniform electric field between the stators
2. constant charge operation

Those conditions would (theoretically) make it a perfectly linear device.

To make it more interesting I repeated the experiment but this time with a high surface resistance coating of about 4E11 instead of the 1E7 Ohms/square in post #1, thereby improving condition 2 above.

yellow trace: symmetric
blue trace: asymmetric
 

Attachments

  • 2nd_order_asym_high_sr_small.jpg
    2nd_order_asym_high_sr_small.jpg
    75.7 KB · Views: 661
As we can see, no increase in second harmonic this time! We have actually improved the linearity of the device by raising the surface resistance of the coating.

That this is so, is confirmed by a significant drop in second harmonic compared to the measurement from post #1, especially in the mid range (0.03% vs 0.1%)

Wachara has commented that he could hear the difference between low and high resistance coating in his headphones. These measurements support that claim.

Unfortunately many commercially available coatings for ESLs are in the 1E7 to 1E8 range. I'd say this makes a strong case for higher S/R coatings!
 
Yes, I can confirm that the lower surface resistance on the diaphragm does make distortion higher - no instrument measurements, just my ears.

This discussion is very much too deep into technical matters, which I don't understand much. Arend-jan, if you could summarize your findings, I would really appreciate it.

Wachara C.
 
chinsettawong said:
This discussion is very much too deep into technical matters, which I don't understand much. Arend-jan, if you could summarize your findings, I would really appreciate it.

Wachara C.

Sure. The simple summary is:

1. constructional asymmetry can increase distortion in a real world electrostatic transducer. Speakers with higher SR coating are less effected by this than speakers with low SR coating.

Side note: the effects are probably marginal in our speakers, the experiment was done to confirm/bust the general opinion that distortion does not depend on asymmetry.

2. very high resistance diaphragm coating lowers distortion, also in the mid range. A value around 1E11 Ohms/square is adequate.

3. a small amount of damping lowers distortion in the low/mid range.

4. asymmetric acoustical loading (e.g. damping material on one side of the panel only) does not raise distortion like asymmetry in construction.


ddd said:
Hi Jan, you can probably simulate differency by adding serial resistors to the diaphragm (say Nx2,2E7Ohm). The THD could differ, especially at lower frequences.

Hi ddd,

Ah yes, the series resistor. This I need to investigate to complete the experiment, I agree. Note that the series resistor is only equivalent to high surface resistance coating if the diaphragm moves completely uniform (e.g. like a piston) in a completely uniform electric field. This is not the case in a real speaker. The influence of a series resistor should be least effective on low frequencies as the charge on the diaphragm has more time to move. This is interesting for real world speakers so I will do the experiment.

N.B. Please don't call me 'Jan', my name is 'Arend-Jan', it's a single name. Thanks.
 
I redid the experiment with the low S/R (1E7) coating and a series resistor of 1E9. With and without the series resistor was only a very small difference in distortion level, absolutely not comparable to what we have seen with the 4E11 coating.

Then I constructed a 5E10 Ohm series resistor but could not get the same level of charge on the diaphragm, and as a consequence the output was about -10dB compared to before. Distortion levels were still up high as with the 1E9 series resistor.

I repeated the experiment with the very high resistance coating (in the 1E11 - 1E12 range) and again distortion levels dropped by a factor 10. Same as before.

Concluding, I think this case makes it plausible enough that an external resistor is no substitute for very high resistance coating. And not only for full range (e.g. low frequencies) ESLs as is often believed. A coating with a S/R around 1E11 is adequate.

I hope someone interested will pick this up and try to confirm my findings.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.