ESL, Ribbon THD

[Mor den Siur]

Greetings

There is some DIY constructions on this forum, but there is not many measurements taken on them. If someone have some I would be grateful for posting it here.

After Zaph published popular ribbon measurements it changed a lot my point of view on them.

[Calvin]

Hi,

this is my small panel, measured in a anechoic chamber using Monkey Forest and a B&K-Mic-capsule.
Input power was ~50W, which translated to a SPL of ~122dB@4m!
It´s the point just before the membrane touches the stators.
The panel is used at freqs >250Hz.

IM-distortion tests ended with trying to find a ´special´ 2-tone combination where IM-artefacts became just visible with ~-70dB.
At normal listening levels the panel is virtually distortion-free.
A very deciding point is the quality of the amp and the audio transformer. Most of the distortion stems from the tranny and a unstable amp can reach easily much higher distortion levels than a well executed panel.
This is very different to the rather high distortions of ribbons and magnetostats. The ESL is pure and clean up to earsplitting levels as no other loudspeaker principle.

jauu
Calvin

[Mor den Siur]

Thanks for your reply.
Could you post a picture of your ESL or/and give some specs, including membrane thickness?

[a.wayne]

Originally posted by Calvin
Hi,


At normal listening levels the panel is virtually distortion-free.
A very deciding point is the quality of the amp and the audio transformer. Most of the distortion stems from the tranny and a unstable amp can reach easily much higher distortion levels than a well executed panel.
This is very different to the rather high distortions of ribbons and magnetostats. The ESL is pure and clean up to earsplitting levels as no other loudspeaker principle.

jauu
Calvin

Hello Calvin ,

So based on the above information, it seems you are saying Ribbons sound better than electrostatics due to it's higher distortion level

[Bazukaz]

I have recently fired up my new wire panels. THD was ~0,5% from 100 Hz to around 500Hz , ~0.1% from around 500 Hz and up. At some frequencies it was as low as 0.02%. The signal level was around 70% of full output power . However , I meeasured in noisy room(with computer on) , so results may improve. The quality of transformers is not the best, having THD up to ~0.1-0.2% at low frequencies.
Panels are disconnected now, since I have to complete electronics and order better transformers , so I cannot repeat measurements. Here is the frequency response (not the best , again measured in my living room).

I will post more detalailed measurements when panels are finished completely.

[bear]

Ummm... Calvin?

Ur saying that your reference sensitivity is over 100dB/1w/1m??

Please tell me/us how you manged that in an ESL that is not very narrow band?

You say this is your small one? What is small?

And the three curves, 2nd, 3rd and 5th? Or??

_-_-bear

[Calvin]

Hi,

@Mor
there are pics and descriptions of the panel in earlier threads.
@Wayne
No I didn´t say that and didn´t even thought that there was any room left for such an interpretation. There are of course many who prefer a rather distorted signal (especially with K2) over a clean signal. But thats a matter of taste. Objectively the signal quality of a well executed ESL (in the midhigh frequency range to be precise) is the best I´ve seen in any loudspeaker.

@bear
uups I cited myself wrong
The measured value is 110dB@4m, which translates to 122dB@1m for a global distribution character. For the panel the SPL value at a distance of 4m is higher than at 1m because of the cylindrical distribution character. If You recalculate the SPL@4m to the value @1m for a global distribution character, You have to add 12dB, hence 122dB.
In praxis the global player won´t loose 12dB with a quadrupling of distance because of a not ideal global distribution and room acoustics, but on the other hand it suffers from thermal compression, spider progression related compression and BL-nonlinearities which add up to 2-3dB. A ESL-panel dose not suffer from compression until the diaphragm hits the stators. The SPL over input power curve is very linear. So, 110dB@4m is the real measured value, 122dB@1m is the calculated equivalent value for a global distribution character.
The curves are K2,3,4 and 5. K5 can just be seen as a very small peak around 4kHz and -75dB.
Measured with a MB550 mic-capsule You roughly can add 0.1% to 0.2% to these distortion values and even higher values when using cheaper mics. Even though the used B&K capsule came fresh from testing and calibrating by B&K, I don´t know the distortion figures for this capsule. But it should be kept in mind that the complete measurement setup adds distortions of its own which could actually be the major part of overall distortion in this measurement at these high SPL levels.
´Small´ means 25x125cm (~10"x 49").

jauu
Calvin

[Mor den Siur]

Calvin - are you using your panel with any kind of sub below?
As you said panels are great in mid-hi range. If you crossing them what is the frequency?

[a.wayne]

Originally posted by Calvin
Hi,

@Mor
there are pics and descriptions of the panel in earlier threads.
@Wayne
No I didn´t say that and didn´t even thought that there was any room left for such an interpretation. There are of course many who prefer a rather distorted signal (especially with K2) over a clean signal. But thats a matter of taste. Objectively the signal quality of a well executed ESL (in the midhigh frequency range to be precise) is the best I´ve seen in any loudspeaker.

@bear
uups I cited myself wrong
The measured value is 110dB@4m, which translates to 122dB@1m for a global distribution character. For the panel the SPL value at a distance of 4m is higher than at 1m because of the cylindrical distribution character. If You recalculate the SPL@4m to the value @1m for a global distribution character, You have to add 12dB, hence 122dB.
In praxis the global player won´t loose 12dB with a quadrupling of distance because of a not ideal global distribution and room acoustics, but on the other hand it suffers from thermal compression, spider progression related compression and BL-nonlinearities which add up to 2-3dB. A ESL-panel dose not suffer from compression until the diaphragm hits the stators. The SPL over input power curve is very linear. So, 110dB@4m is the real measured value, 122dB@1m is the calculated equivalent value for a global distribution character.
The curves are K2,3,4 and 5. K5 can just be seen as a very small peak around 4kHz and -75dB.
Measured with a MB550 mic-capsule You roughly can add 0.1% to 0.2% to these distortion values and even higher values when using cheaper mics. Even though the used B&K capsule came fresh from testing and calibrating by B&K, I don´t know the distortion figures for this capsule. But it should be kept in mind that the complete measurement setup adds distortions of its own which could actually be the major part of overall distortion in this measurement at these high SPL levels.
´Small´ means 25x125cm (~10"x 49").

jauu
Calvin

Calvin ,

Where are the pics ?...........................................

Are you saying your speaker is 122db@ 1w/1M or @ 50 watts

[Calvin]

Hi,

@Mor
I´m doing only hybrid-ESLs, so Yes, I use a dynamic bass. It have been 8x 7" drivers per side in the first models and will be just 6 drivers in future, working in a dipole casing. Crossover freq is ~250Hz
With the large panel it has been 8x 8" drivers in dipole casings.
It will be 8x7" in a dipole casing with a crossover of ~150-180Hz and a CB-woofer below 50hz featuring 2x18" verylongthrow drivers.

@wayne
here´s a pic of the old ESEL (german for donkey, but I prefer ´Entity of Superior Electrostatic Loudspeaking´) which was DIY
diy esl suggestions thread ...and here are pics of the actual models which have become commercial.
Carver TFM amps for ESL's ??
The new models which will be presented at Munic HighEnd2009 will be smaller in height (160cm the big one and 135cm the small one instead of 195cm as it is now), the casings will be manufactured from magnesium and the big one will come with the big subwoofer in a concrete globe casing.

What I tried to say was that a typical loudspeaker (which is assumed to have a global distribution character) will produce 110dB@4m if it produces 122dB@1m (regardless of how much power is needed for this SPL) because the SPL drops by 6dB per doubling of measuring distance, starting with the highest value close to the surface of the membrane.
The panel produces less than 110dB@1m because the SPL drops when You approach the cylindrical line source. You have to understand that a cylindrical distribution pattern behaves very different to a global pattern with regard to SPL over distance. The SPL rises from the surface of the membrane to a distance of ~3-4m (with panels of more than 1m height). Above this distance the SPL starts to drop but with just 3dB per doubling of measuring distance.
You can easily hear this character in a large, long room, where the cylindrical pattern shows a much more uniform SPL-level than a global pattern.
The panel produces 110dB@4m with an input power of app. 50W. Its not easy to state a real efficiency because of the varying impedance of the speaker. So the 50W would mean equivalent 50W@8Ohms.

jauu
Calvin