ESLs and Amplifier Wattage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
In our research into electrostatic speakers, we've been told time and again (most notably by Roger Sanders in the Electrostatic Speaker Cookbook) that power is the most important quantity in choosing an amplifier to drive ESLs.

I come across people on the internet often who're building 200 watt amplifiers to drive their electrostatics. I am now considering an amplifier project to drive the set of ESLs I intend to build. My friends and I've cleared the major hurdles of ESL production, and am in the research mode of an amplifier now.

Now I'm also coming across people who've build various models of Nelson Pass paternity, rated around 10 watts or so, that "sound great on my ESLs". What's going on here? How can a 10 watt amp power an electrostatic speaker? What am I missing? If I'm right here, P=V^2/R . . . V = sqrt(P*R) . . . leads to a 10 watt amp on an 8ohm resistance producing about a 9V max output voltage. To power an electrostatic, you'd need a pretty large transformer to achieve the stator voltages necessary for proper membrane force. What's am I missing here? Should I be planning a 200W power amp, or should I be studying the elegantly simple Zen 10 watt-ers? Any help that can be provided would be immensely useful. Thanks.

- Jonathan
 
A lttle bit of calculation:

The voltage on the film in a well designed ESL should be in the range 2000-5000 Volts depending on size and the frequency range it should reproduce.
Normally the voltage difference should be no more than half of the film's voltage.
So if we assume we limit the stators peek-peek voltage to 1500 Volts and the ratio of the transformer is 70 (standard value), we need around 22 Volts to drive the ESL to it's maximum.
22 volts at 8 ohm is 60 Watts and my experience is that a high current capable 50 Watt amp can drive any ESL.

At normal listening levels you seldom pass 400 Volts on the stators which in our case means 5,7 Volts or approx. 6 volts which is 4,5 watts at 8 ohms.

So a 10 W high current capable amp can drive a ESL at normal listening levels.

BUT, and this is a big BUT, the problem is that a ESL can have a impedance as low as 0,5 ohms at 20 KHz which demands wastly more power.

With the "normal listening" 6 Volts at 0,5 ohm we need 12 ampere to drive it and that will almost kill any amp.
And besides that, some amplifiers will start oscillate as soon as they are connected to 0,5 ohm.
And at the ESL maximum at 22 Volts we need 44 Ampere.
That is almost 1000 Watt.
By that most amplifiers on the market will say "thank you very much - I'm leaving".

Beside this the output impedance of the amplifier need to be at least 5 times lower than the loudspeakers lowerst impedance which in out case result in a 0,1 ohm.
And it is good practice to keep the open-loops output impedance as low as 0,2 ohms.

So the recommendation is a 30-50 Watt amplifier with a low output impedance and high current capability.

/Janne
 
Hedlund,
Ok, you took it further than I did, but basically you've fleshed out about what i was thinking. So theoretically, ESLs don't require much in the way of power, just enough to get the proper voltage across it's impedance-at-frequency. If I understand you correctly, the only reason people recommend powerful amplifiers for ESLs is so that they can handle the enormous current draw when the ESL's impedance drops to extrely low values. The practical problem, then, is the horrible impedance properties of an ESL. If these were tamed, then a modest amplifier (say, one of the Pass Zens) could drive them. Consider this:

In this thread <a href="http://www.diyaudio.com/forums/showthread.php?threadid=856">here</a>, I've been trying to encourage development of a method of evening out an ESL's impedance curve. I took some data on our lab equipment, and posted the resulting impedance magnitude and phase plots in this thread.

For regular magnetic drivers, there's a capacitive device called a Zobel Network (Randy Slone also calls it a Boucherot Cell on page 209 of the 'High Power Audio Amplifier Construction Manual') who's purpose is to cancel out the mostly inductive impedance of the magnetic driver to provide a flat impedance curve to a crossover network (when built into the speaker), or to the output stage of the amp (when built into the amp).

If such a device could be tailored to an ESL, is it possible that the current-drawing 0.4 ohm impedance magnitudes could be ironed out? Is it possible that the destabilizing phase shifts could be corrected? In other words, is it possible to prepare an ESL''s impedance so that a simpler low-power amplifier could drive it without danger?

I don't think the necessary intermediary device would be as simple as the Zobel; ESL's (as near as I can figure) must offer healty doses of both inductive and capacitive reactance. The Zobel, which is simply a capacitor in parallel with the driver through a resistor, is designed to negate purely inductive reactance. Obviously, something more is going to be needed.

Another point of concern is what this will do to the equalization of the audio signal. I would think that a flat impedance curve would actually help the linearization of the sound frequency response, but maybe its more subtle than I'm thinking. Anyway, you seem to know your stuff, I'll throw this open to you and anyone else who can offer any help. Thanks again for your time.
- Jonathan
 
Are you kidding with me.
A Zobel network sits parallell to the driver further lowering the impedance. I think 0,4 ohm is bad enough.

The only way that you can make is safe to connect a Yamaha amp to a ESL is to use a series resistans of approx. 1,5 ohms which will give a not desirable attenuation of the highest treble. So please don't use Yamaha amps.

The impedance dip in the treble is a consequence of the panels capacitance and the transformers ratio.
Either you have to lower the capacitance or the ratio.
The capacitance is depending in the panels size, distance between the film and stators and the voltage applied to the film.
There are four things you can do to increase the impedance.

1. Lower the voltage on the film.
2. Make a smaller panel.
3. Increase the distance between film and stators.
4. Lower the ratio of the transformer, but since the one you have is 50:1 this is no option to you.

All will decrease the capacitance but also the sensitivity of the ESL which I suppose is low as it is.

Take my word for it. Designing ESL's is not easy.
It's the art of compromizes.
The key to a successful design is to make the distance between the film and stators as small as possible.
The distance depends on a lot of things like the frequency range the panel is supposed to cover, the tension of the film, the sound pressure you want it to manage, panel width, distance between the ribs in the stators, the amplitude etc, etc.
No wonder there are so few successful designs on the market.

/Janne
 
The Zobel (or snubber) network is used in two separate places and for two distinct reasons. One use is to fit a Zobel directly across the drive unit terminals so that the combination presents a nominally resistive load to the passive crossover. The second use is to fit a Zobel at the output of the power amplifier in order to reduce the load impedance at high frequencies and ensure amplifier closed-loop stability.

The amplifier Zobel is fitted to maintain stability into inductive loads. An output series inductor is usually provided for stability into capacitive loads.

With a number of amplifiers, the problem is not that the amps become unstable with a capacitive load (some do, but this can usually be cured with additional compensation components) but with the protection circuitry tripping due to the high, out of phase current requirements. Hence the use of high power amps to try to avoid this.

High power is not a necessary requirement, but the ability to deliver a high current, out of phase with the voltage, is. As an example, the Quad ESL57 has a maximum input voltage of 33Vp-p before damage occurs (limiters can be fitted to prevent this voltage from being exceeded) so the amp only needs to be capable of providing an output of 12Vrms (it has long been accepted that a 15W amp is adequate for an ESL57, despite its relative insensitivity of 84 to 87dB/W). However, the minimum impedance of the ELS57 is 2ohm at 15kHz so, in theory, for maximum power the amp must be capable of supplying about 8A (90deg out of phase with the voltage due to the capacitive nature of the ESL).

Fortunately, this level of current is not required in practice since the amount of power in music at this higher frequency is substantially less than that at lower frequencies and I would suggest that a current of about one half of the maximum will be sufficient.

You may be interested in seeing a JLH Class-A amplifier specifically designed to drive the Quad ESL57, which I have just added to my website.

http://www.gmweb.btinternet.co.uk/index.htm

Geoff

[Edited by Geoff on 11-04-2001 at 01:49 PM]
 
My experience with ESL is limited to my Sound Lab A1’s that I use and information found on the internet and some referance books on how to buid them. This is a link to the Sound Lab A1 specifications is: http://www.soundlab-speakers.com/a1.htm

As can be seen Sound Lab recommends a 100 to 400 watt amplifier to drive these. I have found that they can be driven to good listening levels by as little as ten watts.

These ESL's have a internal crossover from low frequency transformers to a high frequency transformer. This gives then a somewhat different impedance curve than a ESL using only a single transformer. As I recall the lowest impedance within the audio range that I run into is around 1.8 ohms at the high end with another dip near the very low midrange due to a internal dividing network.

My studies have led me to the following conclusions:

Maximum amplifier power required by any ESL would be the amount of power required to generate enough voltage to drive the panels hard enough to drive them to their maximum SPL level without doing damage or acing. This at the lowest frequency.

Also since the low impedance point is at a very high frequency there is usually not enough signal there to cause any problems with not being able to drive a ESL with many low power amplifiers. As long as the amplifier being used is stable at all signal levels when driving the load the ESL provides.

I is also reasonable to assume that the size of the ESL panel and the quality of the transformers used along with their turns ratio will also have a great effect on the amount of power required to drive them. Thus there are a lot of variables that makes things difficult.

John Fassotte
Alaskan Audio

For those not familiar with the Sound Lab A1 design I will briefly outline their internal workings.

The Sound Lab A1’s panels are very large and have fairly wide spaced stators. The applied polarizing voltage is about 8 kilovolts and is adjustable.. The panels are curved and mechanically divided in to segments. Each segment has a slightly different low self-resonant frequency. When these resonance points are combined they give a smoother extended low-end response and thus are classified as full range. As I recall the panel capacitance to each stator is around 2700 pf.

The entire panel is driven by a high frequency transformer and a set of two very large heavy low frequency transformers. Each transformer is optimized to handle its specific frequency range. The two low frequency transformers are in parallel.

The input signal is divided into low and high frequency segments and routed to their respective transformers though networks that allow some adjustment in drive levels to each. The upper frequency range of the low frequency transformer drops off very rapidly while the high frequency transformer has excellent characteristics up to about 50 khz when loaded by the panel.

The output of the high and low frequency transformer is added together by some simple capacitors in the original, which has been modified by me to a RC network to smooth out the low to high crossover point between the transformers at the lower mid range by me. The amount of the polarizing voltage is also adjustable and is also used to tailor the overall response.
 
For what it is worth....
I have a pair of Martin Logan SL3's which for about a year used a 100watt Yamaha AV amp to drive, the sound was better then my previous speakers so all was good.
I soon built the Opti-mos amp using 56 volt rails (150watt @ 8-ohm) Wow, after installing this I didn't think they could sound any better!
Well, long story short along came the second Opti-mos using 86 volt rails (330 watt @ 8-ohm).
I did not think it was possible, but the Martin Logan's sound even better.
In my own opinion the bigger the amp the happier the electrostatics seem to be.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.