Asymmetric construction induced distortion in electrostatic speakers - Page 4 - diyAudio
Go Back   Home > Forums > Loudspeakers > Planars & Exotics

Planars & Exotics ESL's, planars, and alternative technologies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 13th June 2012, 10:05 AM   #31
Legis is offline Legis  Finland
diyAudio Member
 
Legis's Avatar
 
Join Date: Dec 2009
I also got a first hand experince with low res v. high res coating when I changed the 12m aluminum coated membrane of my stats to 6m Hostapohan coated with antistatic vinyl polish that should give quite high surface resistance (somewhere near 10^9ohms).

The measuring equipment is in a downgraded state now because the main measurement soundcard is broken, hence the high noise level in the plots.

Low drive level measurement was done without crossover or equalization, the high drive level was done with 260Hz 6dB/oct high pass to avoid trafo core saturation. Measured in my room with concrete walls, pretty close to the walls, so never mind the rollercoaster ups and downs. The mic was in the listening position.



THD high res. v. low res; lower drive level (noisy measuring equipment cannot measure the high res coating's THD):

Click the image to open in full size.


THD high res v. low res, higher drive level (around 25Vrms):

Click the image to open in full size.


Pretty nice ~25dB difference at the high drive level.

The bias voltage was somewhat below 2kV during the measurements, so the distortion showing with the high res coating at high drive level is likely mostly from the transformer. Boosting the bias to 6kV increases the sensitivity by 10dBs, and at the same drive level the distortion is likely 7-8dB lower even though the speaker plays 10dB louder.

I had previously thought that the excess H2 levels with aluminum coating at high drive levels were caused by the step up transfrmer or some low res coating - trafo interaction mechanism. If I understand correctly, the charge migration on the membrane should increase H3, not H2.

Last edited by Legis; 13th June 2012 at 10:17 AM.
  Reply With Quote
Old 13th June 2012, 10:59 AM   #32
diyAudio Member
 
Join Date: Feb 2008
Location: Jackson,michigan
Very Good Legis!!!

Although I didn't get as far in my measurements, I have found that running a higher bias seemed to sound better as well.
Much cleaner sounding with much more authority on the lower end midbass especially.
Hopefully, I will get to explore this in greater detail with my next set of panels.

If I don't burn them up first !!! He,he,he

I have run them at 13.8kv before but that was pushing the envelope for instant failure.
So I kept them at a maximum of 10kv of bias and even this was marginal,but at a low level they sounded very nice and much clearer to me than running at around 3.5kv.

jer

Last edited by geraldfryjr; 13th June 2012 at 11:14 AM.
  Reply With Quote
Old 13th June 2012, 02:38 PM   #33
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by Legis View Post
I had previously thought that the excess H2 levels with aluminum coating at high drive levels were caused by the step up transfrmer or some low res coating - trafo interaction mechanism. If I understand correctly, the charge migration on the membrane should increase H3, not H2.
This would be the case if the push-pull ESL was mechanically perfectly symmetric. But even if an ESL is built perfectly symmetric, as soon as the bias charge is applied to the diagphragm deflects toward one of the stators. This upsets the symmetry. Curved stators add to this problem because even when uncharged the diaphragm tends to be closer to the rear stator than the front.

You will mainly see H3 due to charge migration show up at lower frequencies where the diaphragm excursion increases to a significant portion of the gap. It can be confusing to differentiate transformer distortion from ESL drive mechanism distortion. If you have a voltage probe, you can measure the transformer distortion directly by monitoring the applied primary voltage right at the transformer connections, downstream of any crossover parts or series damping resistors. In general, you will find it's distortion contribution to be rather small except at lower frequencies and higher drive levels.

BTW, congrats on the successful rebuild

Last edited by bolserst; 13th June 2012 at 02:41 PM.
  Reply With Quote
Old 13th June 2012, 03:41 PM   #34
Legis is offline Legis  Finland
diyAudio Member
 
Legis's Avatar
 
Join Date: Dec 2009
Thanks bolserst.

The THD reduction effect with increased bias now puzzles me a little. I measured with aluminum membrane that if I increased bias so it increased SPL/sensitivity by 5dBs, it also dropped the THD for 3,5-4,5dBs (the panels played 5dB louder with less distortion) - the THD level kept at the same level while the signal escaped fom it. The reduction I measured was a constant reduction in the whole band.

What is the cause of this? I wonder if the same effect applies to high res coating, geraldfryjr's experience suggest that it does (I presume you use high res coating on the membrane).
  Reply With Quote
Old 13th June 2012, 04:51 PM   #35
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by Legis View Post
The THD reduction effect with increased bias now puzzles me a little. I measured with aluminum membrane that if I increased bias so it increased SPL/sensitivity by 5dBs, it also dropped the THD for 3,5-4,5dBs (the panels played 5dB louder with less distortion) - the THD level kept at the same level while the signal escaped fom it. The reduction I measured was a constant reduction in the whole band.
Hmmmm...this doesn't make sense. Perhaps it is due to measurement technique. It appears you are using the Farina method of distortion measurement with logarithmic swept sine. The ARTA manual states that for reliable results using this technique, a low level of room reverberation is needed. Based on your description of the measurement setup, this may be the problem.

You might try the STEPS software like arend-jan used to see if you get similar results.

Oh, one other question: In post#31 you mention the high drive level was about 25Vrms, what was the low drive level?
I ask because there doesn't seem to be much difference in SPL level between the two cases...perhaps 5dB.
  Reply With Quote
Old 13th June 2012, 04:54 PM   #36
Bazukaz is offline Bazukaz  Lithuania
diyAudio Member
 
Bazukaz's Avatar
 
Join Date: Nov 2005
Location: Vilnius
Send a message via Skype™ to Bazukaz
Quote:
Originally Posted by Legis View Post
Thanks bolserst.

The THD reduction effect with increased bias now puzzles me a little. I measured with aluminum membrane that if I increased bias so it increased SPL/sensitivity by 5dBs, it also dropped the THD for 3,5-4,5dBs (the panels played 5dB louder with less distortion) - the THD level kept at the same level while the signal escaped fom it. The reduction I measured was a constant reduction in the whole band.

What is the cause of this? I wonder if the same effect applies to high res coating, geraldfryjr's experience suggest that it does (I presume you use high res coating on the membrane).
Hi,
I have had once output transformers themselves producing buzzing(mostly in 2nd harmonic frequency) with steady state sine waves.That contributed significantly to measured harmonic distortion. That was more in the mid-high range(>400 hz)
As panel output will be larger compared to trafo buz with higher bias, its a likely possibility in your case.
You can test this easily by measuring(and listening) the buzz from transformers with diaphragm fully discharged.

Last edited by Bazukaz; 13th June 2012 at 04:59 PM.
  Reply With Quote
Old 13th June 2012, 09:48 PM   #37
Legis is offline Legis  Finland
diyAudio Member
 
Legis's Avatar
 
Join Date: Dec 2009
Thanks for the answers. I have been busy membraning the other panel. It 'snow done and all I can say is WOW.

I measured the effect with the old aluminum membrane, when the distortion figures were in the 0,3-1% region. I cannot see how it would be because of trafo buzz, because high res coating dropped THD by 20-25dB to very low levels. If the trafo would buzz I quess the THD would not have dropped. Bias voltage has not changed grom those times. I can also quarantee it was not a measurement flaw, I made the measurement to both panesl and it was consistent. My room is very quiet and I always use adequate levels when measuring THD.

I try to measure the effect tomorrow with the new membranes, I upload some plots here.

I did not measure the voltage with the lower drive level. I would predict pretty close to half of the Vrms, ie - 6dB. The THD rised pretty much with the higher sweep (with aluminum membrane), 6dB increase in SPL rised the THD by ~20dB. The trafo is Plitron's 1:75.

Last edited by Legis; 13th June 2012 at 09:52 PM.
  Reply With Quote
Old 13th June 2012, 10:19 PM   #38
diyAudio Member
 
bolserst's Avatar
 
Join Date: Jun 2005
Location: Fort Worth, Texas
Quote:
Originally Posted by Legis View Post
I have been busy membraning the other panel. It's now done and all I can say is WOW
Welcome to constant charge ESL

Quote:
I can also quarantee it was not a measurement flaw, I made the measurement to both panesl and it was consistent. My room is very quiet and I always use adequate levels when measuring THD.
It isn't just that the room needs to be quiet, the Farina method distortion results are not isolated from reflections since they are calculated based on the log sweep separating the distortion components in time from the undistored response. This was my concern since you mentioned the ESLs were positioned near concrete walls.

Quote:
...The THD rised pretty much with the higher sweep (with aluminum membrane), 6dB increase in SPL rised the THD by ~20dB.
Yes, H2 went up ~20dB, but H3,H4 went down ~5dB. This is not expected distortion trending behavior, making me think reflections might be corrupting your measurements.

My concern might be completely unfounded...
My experience with the Farina method is minimal, but I did have issues getting distortion results that matched more traditional methods when measuring loudspeakers. Practice makes perfect?
  Reply With Quote
Old 13th June 2012, 11:00 PM   #39
Legis is offline Legis  Finland
diyAudio Member
 
Legis's Avatar
 
Join Date: Dec 2009
Quote:
Originally Posted by bolserst View Post
Welcome to constant charge ESL


It isn't just that the room needs to be quiet, the Farina method distortion results are not isolated from reflections since they are calculated based on the log sweep separating the distortion components in time from the undistored response. This was my concern since you mentioned the ESLs were positioned near concrete walls.


Yes, H2 went up ~20dB, but H3,H4 went down ~5dB. This is not expected distortion trending behavior, making me think reflections might be corrupting your measurements.

My concern might be completely unfounded...
My experience with the Farina method is minimal, but I did have issues getting distortion results that matched more traditional methods when measuring loudspeakers. Practice makes perfect?
The H3...Hn harmonics are at the noise level of my measuring equipment, the base noise is only -50-60dB down (it fluctuates). They do not repsesent the correct levels, only H2 level that has risen is correct. The equipment is very noisy indeed and also the mic's cord is in very bad condition, connection problems if the cord is twisted wrong way. I have used it because it's long, all other XLR cables are only 1m long (hifi interconnects). I try to measure tomorrow with 1m cord. The noise issue I cannot change now because the EMU 0404 usb is broken and I measure with my secondary equipment (Behringer Xenyx mixer and cheap 48kHz usb soundcard with only unbalanced input).

I have usually got pretty reliable and repeatable results with both methods, sweep thing is way faster though. I have never seen that it would rise the distortion evenly all the way, in undamped room it can make some points in which the distortion makes a jump but not a constant increase.

I see what I can do tomorrow with both steps and the sweep.

Last edited by Legis; 13th June 2012 at 11:09 PM.
  Reply With Quote
Old 14th June 2012, 02:12 PM   #40
Legis is offline Legis  Finland
diyAudio Member
 
Legis's Avatar
 
Join Date: Dec 2009
It seems that I cannot make reliable result with noisy equipment like I previously could with alumembrnae when the THD was much higher. Although the tendency seems to be there that the distortion level stays at the same (absolute) level while signal's level increases with higher bias, thus THD's / H2's level decreases.

Does anybody else have measuring equipment and variable bias supply?

By the way, I have read that higher bias introduces "negative tension" (or sth) to the membrane and it should lower the Fs of the membrane. However I could not produce any measurable difference to the panels fundamental resonances (it is segmented in different size sections) with two bias settings that had 4dB SPL difference. Too little difference in bias levels?
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Asymmetric drive to asymmetric loudspeaker tolits Multi-Way 5 11th January 2008 05:14 PM
Class D for electrostatic speakers koldby Class D 23 4th December 2007 04:10 PM
Electrostatic Speakers & Amp ralph-bway Multi-Way 28 6th July 2003 06:43 PM
electrostatic speakers junglejuice Planars & Exotics 3 22nd October 2001 09:32 PM


New To Site? Need Help?

All times are GMT. The time now is 09:33 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2