Driving ESLs, thoughts wanted

[Altaic]

I'm a college student, and I'm all for making sacrifices for cost effectiveness. That said, I'm still (passively, now) looking for a cheaper method to drive ESLs than spending $300 per channel on audio transformers. I don't yet have an amp for the ESLs, so I am in favor of direct-driving them if possible.

Here are some links I've gathered (I have a couple more, but they're dead now):

http://gilmore.chem.northwestern.edu/accoustat1.gif
http://gilmore.chem.northwestern.edu/accoustat2.gif
http://gilmore.chem.northwestern.edu/koss.gif (lousy sound)
http://www.audiocircuit.com/9041-esl-circuit/Diy/Experts/NeilMckean-NM/9041DENM-HVA.htm

I have a bunch of nice MJL1302A/MJL3281A bipolars, so I'd be in favor of doing some sort of SS/tube hybrid, or even bipolar/MOSFET hybrid. What about making a tube voltage amp that takes inputs from the outputs of an ESP 3A? Any advantages, or would the 3A be wasted? The tube amp shouldn't have to deal with gain, because current waveforms are converted to voltage waveforms, right? I'm also in favor of buying the 3A boards from Elliott, to simplify the process...

I know little about SS amp design, and even less about tubes, so the less I have to figure out for myself the better. Any help is appreciated

 

[Denis]

Driving ESL by hybrid SS/tube amp

You can get excellent results with simple complementary Darlington emitter follower made from these MJL3281/1302 transistors driven by a triode voltage amp. The internal resistance of two halves of 6922 connected in parallel is low enough. Use two pairs of the mentioned BJTs with a pair of MJE15032/15033 as their drivers at40Vcc (single rail supply) and you'll avoid the use of any protection circuit except Vcc rail 5A fast-blow fuse. The power dissipation ability of these transistors will allow you to put a lot of idle current (up to 2 a) into them making pure class A amplifier. No loop NFB will be needed. The re is no reson to use more high Vcc voltage to drive ESL63s.

Best regards

Denis N. Afanassyev

 

[Altaic]

schematics?

Do you know of any schematics of amps such as the setup you described? Also, wouldn't the bipolars drive the triode, or rather the triode be the actual output device? My terminology's probably screwed up.

Even better, would it be possible to make a simple modifiaction/addition to something like the 3A? Like, slap the triode on after the usual output stage? I haven't had the opportunity to sleep since yesterday morning, so if I'm not doing so well picking up on ideas, please forgive me.

 

[phishead8]

You don't need to spend $300 on step-up transformers.

You can use a tube step-down transformer in reverse. You want one with at least 10K impedance on the primaries (you will use as secondaries), and it must be center tapped for the high voltage bias.

Try www.tubesandmore.com under the push pull audio transformers. P-T1609 worked well for me. Only $37.

Also www.handwoundtransformers.com/pp has a similar model for only $29! I haven't tried it, but I can't see a reason why it won't work.

These are some cheap and easy solutions to the transformer problem. My suggestion to you is to try them and listen. If you want better quality later you can always upgrade (maybe in time to return the old ones).

This way you can listen to your beautiful ESLs while you build your direct-drive amp. Good luck.

-Dan

 

[phishead8]

And another thing...

If you don't have an amp to power the ESLs with, yet, you can buy a mid-fi amp to power them now. After you finish the direct-drive you can bi-amp the system and use the mid-fi amp to power the woofers. (Unless your ESLs are huge you will find that the bass will be a little less than desired)

Just my thoughts, from one poor college student to another.

Keep us posted.

-Dan

 

[Altaic]

Thanks for the info Dan. I had been wondering about using step-down transformers in reverse, but it seems like the power specs are rather odd. 10W output @ 10000Ohms, with an 8Ohm "primary." How does that work out in terms of voltage and current? 60W into 8Ohms with a rail-to-rail voltage of 70V -> 10W @ 7kV? I could even use the 4Ohm one as a primary. I'm not fresh on the formula, though.

BTW, my ESLs are 18"x36" and are costing me ~$50 each without the transformer. I haven't planned for an aditional sub per channel or anything. Also, do you think it would be prudent to not make one dimension divisible by the other? It seems like I might get peaking, beaming, resonating, etc.

 

[phishead8]

You're asking the right questions Altaic.

I'll try and answer them to the best of my knowledge.

When you use the step-down transformers in reverse the 8 Ohm side will not be exactly 8 Ohm. Mine was nominally 4 Ohms and dipped down to 1 Ohm at 40khz! So, to make it a little safer for my amp (a 200 Watt Yamaha) I used the 16 Ohm configuration instead. I personally have watched multiple amps give up the ghost while trying to play the low impedances (before I knew what I was doing). Do not make the same mistake! In regards to this the bigger and more robust the amp is, the better. You'll probably need an amp larger than 60W. Electrostatic Loudspeaker Design Cookbook by Roger Sanders spouts a minimum of 100W. I would do at least 150W to be sure. I don't know what else to tell you about that.

Don't worry about the 10W spec of the trannie unless you want your music loud. I found that I could listen to quite loud volumes with no problem.

As far as I know there is no problem to building your ESL with a ratio in the dimensions. You may want to build some supports in the middle of the 18" to stabilize the diaphram. It depends on your spacing, but 18" gives the membrane alot of room to travel and it may touch/arc. Think about gluing a strip of spacer in the middle, so that essentially you will have 2 9" x 36" or 3 6" x 36" ESLs working in tandem.
Since the bulk of the music is well above the resonant frequency of the membrane (mine was ~115hz) you won't have to worry about peaking. Beaminess is a characteristic of ESLs. While making the sweet spot very small, it does make the "soundstage" at that point very wide and open. There is no easy solution to that problem.

I am not an expert in all of these problems, so I encourage anyone else to point out the errors of my advice.

Good luck,
Dan

 

[analog_sa]

Denis

It seems like Altaic is looking for means to directly drive the electrostats, which means a minimum of 2kV peak in push-pull (meaning with respect to the diaphragm). The 40v emitter followers may be fine to energise a step-up but that's all they can do.

Altaic

I am not a transformer guru but some of the specs of the step-up are clear. You need a step-up ratio in the region of 1:100-150 and insulation between primary and secondary capable of coping with a few kV. Dunno if all valve outputs can cope with this. Frequency response requirements otoh seem to be more easily achieved than with normal loudspeaker loads as the load is almost entirely capacitive and there is very little resistive component so very little real power transfer. I've found that even low quality non-interleaved transformers can sound quite good if they're not expected to drive directly huge areas of the panels. What this means is that you can have separate transformers for the bass and treble panels, thus allowing you to use a larger core for the bass. To get you started you can use even simple mains transformers, two per panel, primaries in parallel and secondaries in series. This requires each transformer to provide just half the step-up ratio and needs only to cope with half the voltage requirements.
As in most electrostats only thin strips are expected to work at high frequencies, this will reduce the capacitance the high frequency step-up transformer is seeing. It's easy to come to a compromise between the filter, hf radiating area and transformer quality.
I once ended up with a 3-transformer configuration, each driving a single frequency band and in turn being driven by a separate power amp. Very dynamic sound with excellent bass.

Unfortunately your real question about the direct drive i can't really answer. Obviously you'll need some pretty high voltage valves like 845, 211 etc. With all the complications and obvious dangers. Even if you decide to follow that route, i feel you should first get a transfomer based setup up and running and then improve on it. It will be so much cheaper and easier.



peter

 

[Denis]

Re: schematics?

Do you know of any schematics of amps such as the setup you described? Also, wouldn't the bipolars drive the triode, or rather the triode be the actual output device? My terminology's probably screwed up.

In this circut the vacuum triode with resistive late load drives a bipolar (Darlington emitter follower) output stage via a capacitor (10 - 15 uF polypropylene). The loudspeaker is capacitor(>4700 uF electrolytic) connected to the output too.

I've not seen yet a schematic exactly what I've described.

I was experimenting with a Darlington emitter follower consisting of MJ15024/15025 output pair driven by MJE15032/15033. The output transistors were biased to 1.5A. I drove this output stage with also a BJT resistor-loaded common-emitter stage (MJE340) operating with much higher B+ voltage (160V) rather than a tube. The THD appeared quite low too. This prototype makes quite remarkable sound and I'm continuing refining this circuit.

Of course, this amplifier needs a step-up transformer to drive the ESLs.
Moreover, the directly driven ESLs is a bad idea, because the properly designed transformer greatly improves the performance. In such masterpieces as QUAD ESL63 the step-up transformers are actually the parts of equalizing circuit, which is essential to make a dipole radiator producing a flat response.

It seems indeed feasible to design the step-up output transformer for a tube amp, which will directly matched to ESL panels.

Some example of vacuum triode / bipolar hybrid without global NFB can be found at www.audiodesign.tin.it, but this is not class A design with some complications.

Regards,

Denis N. Afanassyev.

 

[vbd]

Hi all, I will not enter the debate about transformer vs direct drive, I just give a link to a DIY amplifier for ESL:
http://www.audiocircuit.com/9041-esl-circuit/9041IMAI-DI.htm
This design has MOSFETS outputs, feedback, and opamps... I found it quite interesting...
If you are ready to cope with high voltages...

 

[ultranalog]

direct drive ESL amps

845 PP direct drive amp

 

[Denis]

Hi all, I will not enter the debate about transformer vs direct drive, I just give a link to a DIY amplifier for ESL:
http://www.audiocircuit.com/9041-esl-circuit/9041IMAI-DI.htm
This design has MOSFETS outputs, feedback, and opamps... I found it quite interesting...
If you are ready to cope with high voltages...

MOSFETS and OPAMPS are certainly not my favourite parts.

 

[vbd]

MOSFETS and OPAMPS are certainly not my favourite parts.

Hi Denis, this is Denis !
The 845PP tube design from ultranalog is for you, then... I must admit I' m from the solid state age and I' m reluctant to design with tubes... But I also think that a good design will sound good, whatever is the technology...

 

[Denis]

Hi Denis, this is Denis !
The 845PP tube design from ultranalog is for you, then... I must admit I' m from the solid state age and I' m reluctant to design with tubes... But I also think that a good design will sound good, whatever is the technology...

I equally design with tubes and solid state and I trust in transformers.

Denis.

 

[Circlotron]

Off-beat transformer

I am not a transformer guru but some of the specs of the step-up are clear. You need a step-up ratio in the region of 1:100-150 and insulation between primary and secondary capable of coping with a few kV. Dunno if all valve outputs can cope with this. Frequency response requirements otoh seem to be more easily achieved than with normal loudspeaker loads as the load is almost entirely capacitive and there is very little resistive component so very little real power transfer. I've found that even low quality non-interleaved transformers can sound quite good if they're not expected to drive directly huge areas of the panels.

You want low cost, good insulation and 1:100 stepup ratio? (=1:10,000 impedance ratio). Try a car ignition coil, it's just a transformer in disguise. It will have relatively loose primary to secondary coupling so it may be a complete flop, but you never know... Also they come in two flavours - the high voltage end with a single outlet, and the other type where both ends of the HV side are avaiable. This type has the secondary and primary separated, the former is only an autotransformer. You already have one so give it a go.

GP.

 

[Altaic]

Woah, so if I need say 6kV from a 60V output, I'd need an 80000Ohm <-> 8Ohm transformer? Yikes! I'm only getting a 1:35 turn ratio, then, using a 10000Ohm <-> 8Ohm one? Does that mean I need HV tube/mosfet outputs anyway, if not the full 6kV? 180V, anyway.

 

[MRehorst]

Re: Off-beat transformer

You want low cost, good insulation and 1:100 stepup ratio? (=1:10,000 impedance ratio). Try a car ignition coil, it's just a transformer in disguise. It will have relatively loose primary to secondary coupling so it may be a complete flop, but you never know... Also they come in two flavours - the high voltage end with a single outlet, and the other type where both ends of the HV side are avaiable. This type has the secondary and primary separated, the former is only an autotransformer. You already have one so give it a go.

GP.

I can not think of a way to use either one or two autotransformers to drive an ESL that would not result in applying the high voltage ESL bias directly to the amplifier output. It may not hurt a tube amp, but it is very likely to destroy a solid state amplifier. I recommend that you do NOT try this.

MR

 

[MRehorst]

BTW, my ESLs are 18"x36" and are costing me ~$50 each without the transformer. I haven't planned for an aditional sub per channel or anything. Also, do you think it would be prudent to not make one dimension divisible by the other? It seems like I might get peaking, beaming, resonating, etc.

In theory the rectangular shape will have two major resonances- one related to the long dimension, one to the short. In practice, you're unlikely to ever hear them. If you are really concerned, build the thing so that no two sides are of equal length. That will spread the resonances out. Putting a support strip into the middle (or slightly off-middle) of the diaphragm will help, too, and allow you to bias the speaker at higher voltage (increasing the sensitivity) without the diaphragm hitting the stators.

The best thing anyone can do for their ESLs is make the diaphragm TIGHT and bias it with as high a voltage as you can muster. I like to crank it up until they hiss and whimper a bit , then turn it back down until they're quiet , but you may find your own technique that is less sadistic than mine...

 

[Altaic]

[frusterated rant]
So it still stands that the cost of making the panels is nothing compared to the damn transformer or special amp.

If I want to get more than 2.1kV, I can't use one of the el-cheap-o transformers, rather I either have to use one usually reserved for mains, which will have crappy freq response and kill my bass even more than it already is, or I have to spend $300 on a bloody transformer. So 70V from the 3A is okay either way in this case.

If I want to make a direct-drive amp, I have to get very high voltage parts, and a rare design. BTW, ultranalog, your pdfs are encrypted. I'm not sure how much it would cost for these parts, but I'm guessing upwards of a couple hundred dollars.
[/frusterated rant]

I don't think I need to coat my stators, because I'm making sealed ESLs, so it shouldn't arc until I get upwards of 10kV. Because the stators are uninsulated, shouldn't the voltage not need to be so high? Or is voltage a function of distance between stators? I initially thought to space the stators farther apart because I can have much higher voltage with no problem, and then the diaphram would have more travel, but now I'm thinking it would be better to shorten that distance. Does travel affect frequency response, or does it affect max volume? I'm less concerned with volume and more with frequency response.

 

[MRehorst]

[frusterated rant]

I don't think I need to coat my stators, because I'm making sealed ESLs, so it shouldn't arc until I get upwards of 10kV. Because the stators are uninsulated, shouldn't the voltage not need to be so high? Or is voltage a function of distance between stators? I initially thought to space the stators farther apart because I can have much higher voltage with no problem, and then the diaphram would have more travel, but now I'm thinking it would be better to shorten that distance. Does travel affect frequency response, or does it affect max volume? I'm less concerned with volume and more with frequency response.

Don't worry about the transformers so much. One or two 8-10k primary side transformers per speaker will be more than adequate. Many people have built speakers using these and they work fine, even with relatively low powered amps to drive them.

Coating the stators is not done to prevent arcing. It is done to prevent corona discharge (a faint blue spark that continuously streams ions). Corona discharge makes a whining or hissing sound that is NOT high-fidelity at its finest. You do not have to coat the stators if you use a relatively low bias voltage (<=2kV or so).

Stator to diaphragm spacing has little effect on frequency response at low volume levels. At high volume, low bass will really slam the diaphragm back and forth, possibly hitting the stators. This is why full range ESLs have a reputation for not being able to play loudly.

Most people solve the problem by using conventional bass drivers in a box. Then you have to deal with crossovers...

Isn't engineering wonderful? It isn't about making things cheaper. Its about making trade-offs.

MR