Stator gaps in electorstatic tweeter - midrange?

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Dear all,

First I was building headphones, with a 0.3 mm spacer stator to film to stator, and 700V these where fine, driving them with Stax transformers, but since it was my first prototype I burnt them up. All in the fun I guess, so my next prototype was started.

Now I have been thinking about making hybrid small desktop computer speakers. The tweeter device is meant to be 100 mm by 170 mm in size. Size is essentially fixed due to desk space and construction process. We loose about 10 mm on each side, and 20 mm on height due to construction methods. Making an unsupported diaphragm of 80 mm my 150 mm.

I was hoping to get the tweeter panel down to around 500 Hz, so the frequency response out performs any dome tweeter, maybe this is an unrealistically low frequency for such a dinky driver, (feedback please) but my initial trials at low volume went very well.

With the second version I started with 0.5 mm spacers for stator to film to stator gaps, using old Quad ESL57 transformers, (just the tweeter drive). I used the same 700V bias supply, I will be building new higher voltage bias supplies for the next design as the sensitivity was far too low, and my experiments with headphone drivers showed the higher the bias the better the sound.

This worked brilliantly at low volumes, and a very nice feature is that at low volumes with low frequencies although the membrane hit the stators no damage was done.

Playing with a signal generator, I then cranked the volume up to what I consider excessive outside a party, and we got a nice flash and the nice little tweeter prototype is now black and as expected dead. Now I am thinking 1mm spacers for the next prototype.

Do you think 1mm spacers is to large for a 80 mm by 160mm unsupported diaphragm?

Any suggestions for bias on such a wide gap? What would you think is a reasonable step up for such a wide gap?

My current guess is a bias around 2000V - 3000V, and will stick with the ESL57 transformers for now (I think around a 80 fold step up).

What stator gaps are you guys using for your electrostatic tweeters?

Best regards

My desktop ESL's worked Great !!!

They were 3.25"x9.75" any wider than that you will run into beaming issues.

The dispersion was about 15 to 20 degrees off center for the highend and work just fine for being nearfield at less than 1meter to .5 meter of distance from the drivers.

I have choosen to use a 4 section electrically segmented design for my latest one.
This will increase my high frequency dispersion by 2x to 4x!!

I used a .072" to .080" D/S. and it used every bit of that space below 350hz at very high volumes, I had found Diaphragm to Stator clipping occured at about 180-200Hz with test tones, Again it was very loud.

Being that they have such a small surface area you have to use a much higher Bias voltage to have enough efficiency to keep up with a woofer.

They will work and play just fine whith a 2-3KV bias but I have found that it took 7-8Kv of bias or more, I tested up to 12-14Kv but ran into coating breakdown issues with that model (Not a problem with the new one) to match that of the woofer effeiciency in to order of 89db to 91db per watt, or for a 4-5V peak signal into a 1:160 (or so) step up transformer.

Typically I was still very happy with 7Kv of bias and didn't have to worry about coating issues at all the way they were.

I was able to reach +105db at that point with just my 80watt amplifier (38Vpeak) and my little 5.25" wooofer couldn't keep up past 95db.

My 8' sub however did rather well when I had it setup and running with it.
It was loud eneough that I basically had to yell to be able to carry on a conversation over the music.
At that time I was just designing my HV regulated variable bias supply and I think I was only using 3kv to 5-6.5Kv at the time.

This was when I discoverd that the ESL's gain on my mixer had to be set +3db to +6db higher than that of the woofer and was when I chose to try and Double the bias and it worked!!

Since I didn't actually have a dedicated crossover at the time I just used the shelving filters on my mixer, I had found that they naturally crossed over at about 650-700Hz.

With the actuall dimensions, this was the actual predicted range that they should have been crossed over at.
I was surprised to find out how well it worked !!

Although this seemed rather high for my liking the system sounded very natural and there was no localization of the two drivers!!!
They worked flawlessly in unison as a whole. ;)

I hope that helps you to create a similar system of your own. :D

jer :)
Last edited:
I usually reference everything on the high side to ground (ie earth).

This way if I have to drive the transformer with a Bridged amplifier then there is no static charge on the transformer to be capacitively to the amplifiers outputs through the transformer.

Yes, my power supply is earthed the the schematic is found here,

jer :)
Last edited:
Go stacked if you want it small , as i did early, each doubling of stacks adds another 6dB, especially in the low end. it will cost another bias supply, or at least the cascade to create a positive bias and one negative

the thread is called sandwich esl yummie or something
Tyu, I have found no difference in SPL whether I use the center tap on the secondery side or not.

The transformers I use have a low enough DC resistance on the secondary that it makes no difference at that high of a voltage, or, cuase any imbalances of the voltage refference point to the Stators if it is a floating system.

It it is not a floating system then it could cause problems I think.

Theoretically if the system wasn't floating and you didn't use the center tap, then the one stator would be at grond potential and have little effect, maybe you get half of the output that you would normally get.

I maybe wrong about that as I haven't really studied single fed designs much.

They seem to complicated things more for what they are worth, and, require more components (capapcitors and/or resistors too, Asymmetrical system) to make them work vs just having a transformer fed with a bias supply (Symmetrical system).

There are quite a few discussions about such configurations in other threads.

jer :)
Last edited:
Considering it takes 4 times the excursion to maintain the same SPL for every octave lower, I used a D/S of 1.85mm and didn't run into any issues until I got below about 220hz to 180hz at a very high SPL.

You should be just fine with a 1mm D/S at 400hz to 500hz and above.

jer :)
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.