Driving ESL transformer more efficient on SS-amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2017
Hi guys

Problem: My ESL is unefficient and eats amps for breakfast.
Each amp makes huuuuge difference in soundquality and quickly clips.
They also roll off in the highs, and when I EQ it, the amps hate that even more :) It quickly sounds stressed above 90dB, and the 10dB boost in the highs >10k definately make it a lot worse.

The least stressed so far were Fostex Laboratory amps. But dont sound not perfect in general (clearity, stagesize, dynamics) - 300 Watts.
Best sound was a Parasound 806 but suffers lot from stress when louder (80 Watts).
Just for fun an old cheap Technics SEA1000 clips already on lowest volume :eek:

So finally I drive it on 2 Parasound HCA 2200 MKII.
It is a good balance between previous favorites, though not perfect.


Now, the ESL comes with a serial resistor of 6.7R (10%) MOX 10 Watt.
The efficiency is pretty low, as usual.
Thats what i just measured:
capaciti_phase_impedaauutc.png


the DC meter sais 0 ohm without the serial R, connected to the transformer.
Lowering the resistor from 6.7 to 4.7 doesnt really increase much the audible loudness but clearly the highs, sounds much brighter. Which sounds logical considering the impedance in the highs.

So, id love to skip that resistor but the Parasound wont like that :D
I guess 2-2.5 R they should survive. That should boost the highs many dB and I could skip the EQing.
But... I can climb from 380 to 1000 Watts if i bridge the amps (got 2).
In that case Id need to increase the serial resistor to double, which leaves me with no big benefit other than heating the resistor :mad:

Do you think I could instead use bridging, plug a 4.7 R and parallel a 10uF cap + 1-2 Ohm R. ??
That would limit the DC-R to amp friendly level, but lower the highs AC-R to a mean minimum, enabling the juice to flow....

What do you think?

cheers
Josh
 
Hi....for years I have read this..ESLs need BIG power to work an give full output!..only ESL that I have seen like this were Sound labs...I had one pr of older SL A3 ...sold them after hearing my First Acoustat Xs....
To day in my room 18'x25'x13' I have a pr of Acoustat 2+2 Martin logan CLS an Acoustat Spectra 11....all can be ran off 60 tube watts, with out any stress...great top end well in to the upper 90dbs...
On my CLS I have only a 1ohm 20 watts on the pos input with a 1ohm-10mf cap in a zobel on the 1ohm...like the zobel you have on your 4,7...sounds great an vary ezey to drive...

What ESL do you have?
What bias volt do you have on your pr ESL?....sounds like it maybe vary low...ck it!
 
Last edited:
Disabled Account
Joined 2017
Hi tyu
Well its not exactly a zobel or. But similar effect though.
I heard some use a normal zobel to eliminate the UHF resonance peak of the transformer beyond 30k up. But i cant measure mine higher than 20k. Would be considerable as my amp got xtreme negative feedback (damping >1000).

I got a Capaciti EL160, former german small brand. 20x120cm longthrow foil.
I guess it’s a lil easier to drive for low feedback tube amp with more headroom / soft clipping.
But the total volume wont exceed the 90db spl much.
I prefer SS to get around 100db and also personal preference.

So you also bypass the resistor with a cap to get more HF efficiency??
 
Hi,

when they are in good working condition Capacitis are quite efficient and not that difficult to drive.
After Your description I assume a faulty condition of the panels.
Could be a major loss of polarization voltage (polarzing supply, membrane contact).
Since You seem to have tried more and more powerful amplifiers it might even be that the audio transformers suffered from internal flashovers.
I'd suggest You measure the polarizing voltage of the supply and membrane and the output voltage of the audio transformers to see if the electronics is still ok or if its faulty.

jauu
Calvin
 
+1 for Calvin's post.

The first fix may be just raising the bias. I have big spacing and run over 7kV. Do you have a DIY high voltage probe?

ESL panels, as seen through a transformer, are a horrible load. In my recent efforts, I tried great many networks but ended up with just enough resistance to keep my large amps* happy but not enough to screw up the FR signal too badly into the transformer.

So I'd start by more closely examining the signal reaching the panels downstream of the transformer if you can or at least reaching the transformer "primary".

Can't say as I believe any decent amp sounds different than any other, providing they are working with loads they can manage.

B.
* even my big amps sometimes go into over-load protection north of 8kHz if doing loud REW sweeps; 1980's Kenwood Basic class G capable of hundreds of watts into low loads
 
Disabled Account
Joined 2017
K something real to grab. Measured the Capaciti with 2.83volt amp output in 1 meter distance.
Lot of room incl therefor 1/3 oct smoothed.

capaciti1m1wfpudi.png


It looks more or less like I expected. But if you say they should be easy to drive, there are def. 5-10dB missing. Or ??
 
Hi,

don't know the actual values but the polarizing voltage should be >3kV, probabely around 5kVdc.
The ratio of the audio trannies will be somewhere around 1:100 to 1:150
The amplitude response really looks too low in volume by ~5dB.
Could be a serious loss of polarization, as the HF response falls down too much after my taste.
Btw. the 1m measurement isn't used so often with strip like speakers .... gives almost always bad results.
You better measure the panel twice, once in very close proximity ~5-10cm (near-field condition) and at greater distance, say 4m (far-field condition).
Don't be astounded to see higher SPLs with the 4m measurement.

jauu
Calvin

ps. Michael Ehrlinspiel once told me that he gave all mashinery and plans to a former employee to ensure servicing of the speakers.
Unfortunately don't I know who that person is and how to contact him.
 
Last edited:
Disabled Account
Joined 2017
Hi Calvin

While googlin around i noticed youre someone with a name reg. ESL ;) Respect.
Ya its huge pitty Michael left the planet, I dont understand :....(
Was too good to remain true.

Serious polarization loss would mean ive overdriven (foil-stator contact) or the foil-coating aged away? I did lot of sine-log-sweep high level 60 seconds (hundret times), hope i dont regret that.

Do you know bout DIS elektrostaten, they do service, can u recommend or do you or whats good adress? :)


PS: bout response variation in distance i know them all, its smth totally new for a "normal speaker"-guy. It looks very reasonable. But SPL is 2,83/1M not 4 or 1 inch :)
 
Hi,

You cannot ´overdrive´ the polarization as its not part of the audio-signal-path, which runs via the audio-transformer wires to the stators.
Still though constant highvoltage stress can eat the membrane coating around the contact points.
Also some coatings don´t remain stable under constant HV and movement ... though I would rather exclude that for the coatings Michael used.
What You can override is the flashover treshold of the audio transformers .. especially when the panels begin to play softer and You push up the volume of the amp.
Unfortunately is the volume loss mostly going on slowly and unnoticed which increases the danger of pushing too much power from the amps into the transformers.

Regarding check and possible repair ... I don´t know if the company You named has gained sufficient knowledge over the years ..... my last contact with them is from years ago .... and at that time they didn´t seem to have deeper knowledge of what a esl is and how to construct safely and stable/reliable. I typically don´t repair non-mine esls, but as You are in the south west also, I can offer You to check them out.

jauu
Calvin
 
… the ESL comes with a serial resistor of 6.7R (10%) MOX 10 Watt. The efficiency is pretty low, as usual. Thats what i just measured:
From looking at the available assembly manuals of the Capaciti ESLs, it looks like there were two different types of interfaces. I don’t know which one came first, but I will call them Type A and B. Type A you can see has a resistor internal to the interface enclosure( I believe it is 3.3 ohm). Then you can add in parallel to it a 6.8 ohm or 3.3 ohm depending on your preferred high frequency balance. Type B does not have an internal resistor, so if you don’t have an external resistor in place you won’t get any sound from the speakers. Type B uses resistors of 3.3, 2.2, or 1.5ohm depending on the desired HF balance. For nominally flat on-axis response with either interface type, the total resistance in series with the transformer primary should be about 2.2 ohm.

Which type of interface do your E160s have?
Based on the fact that they came with a 6.7 ohm resistor I am guessing Type A. But, looking at the impedance measurement and 1m frequency response you posted, it looks like there is no internal resistor in parallel with the external 6.7ohm. So either there is no internal resistor, or it has failed.

What signal level did you use when measuring the impedance?

I guess 2-2.5 R they should survive. That should boost the highs many dB and I could skip the EQing. But... I can climb from 380 to 1000 Watts if i bridge the amps (got 2).
In that case Id need to increase the serial resistor to double, which leaves me with no big benefit other than heating the resistor
It would be best to not use more than 250W – 300W on these to avoid damaging the coating over time.
 

Attachments

  • Capaciti_E160_Interface_A.png
    Capaciti_E160_Interface_A.png
    813.8 KB · Views: 240
  • Capaciti_E160_Interface_B.png
    Capaciti_E160_Interface_B.png
    923 KB · Views: 243
  • Capaciti_E160_assembly.pdf
    848.5 KB · Views: 538
  • Capaciti-Element-Series-Assembly.pdf
    287.8 KB · Views: 120
Could be a major loss of polarization voltage (polarzing supply, membrane contact). Since You seem to have tried more and more powerful amplifiers it might even be that the audio transformers suffered from internal flashovers.
Looking at the impedance curve in post#1 taken with 6.7ohm series resistor, the primary inductance couldn’t be more than 6mH - 8mH. For a full range ESL I would expect 10 to 20 times that. Perhaps the transformers do have internal damage as you describe. Comparing measurements for both speakers may give an indication. If impedance and response match each other reasonably closely, it is unlikely damage has taken place.

..don't know the actual values but the polarizing voltage should be >3kV, probably around 5kVdc.
I recall Capaciti mentioning his supplies were variable between 1kV – 4kV, so probably operation range in the 3kV-3.5kV range.
 
Disabled Account
Joined 2017
Hi all. Nice support so far, thank you guys.
Id be very happy to come back to your offer Calvin :)


Thx Bolsert for the PDF!!!! Great.
But funny, I have neither one of them. I drove to Zurich 2015 to 2nd hand dealer.
I got only 2 normal connectors, and a power in. No external resistor at all.


Inside the positive pin goes into the 6.8 resistor which leads straight to the transformer. That's it. Takin the resistor out leaves me with near 0 Ohm DCR.
With resistor 6.8.


I conclude lowering the resistor value will do the same as those versions, lift the highs few dB.


But also the poor 10 watt resistor has to stand a lot. Gets hot, changes value, bä.
I bought a bunch of 44 MOX with 47R each. Would give me 2.1 Ohm and 220 watts, wrapped in a cooler, with 0.45% eff tolerance. Not yet installed though. Nice idea?
Wouldn't work with bridged amp tough.




This damaging topic makes me nervous. New foils are 1000 bucks in some places.
New transformer.... ? guess not available anymore. And theyd go to the bin?? :..(


cheers
Josh
 
It might be helpful to re-state something Calvin said.

If you pump enormous voltages into the transformer in an effort to raise the loudness, you may cause shorting and burns in the high-voltage secondary.

So take care how much power you are running into the speakers. And check the bias voltage reaching the panels.

If you know about safety, it is easy to make a high-voltage probe for this application. Put together a chain of resistors to form a voltage divider adding up to say, 200-megOhms and a last one that is .2-megOhms. If you don't understand this clearly, don't try it.

B.
 
Disabled Account
Joined 2017
well considering that the official manual from the company sais it can run more than 100dB.... I think ive been usually faaaar away from that. Still the fostex amp with 150 watts easily clipped.
That sounds to me like some big problems exist.
Ill have to take care about it, thx.


The resistor question was just more a parallel generic thing to talk about out of interest. Making smth more efficient, makes easier life for amp, sounds better. Having more watts than needed too. So rather a finetuning than any intention to run a big party volume or ignoring the damage.
 
Or not realizing the time it takes to charge some ESLs.

Here's a time course for my panels*. I use a very large series resistor on the bias supply.

The divisions are 10 dB. The curves are:

before bias
10 minutes
8 hrs
17 hrs

Not shown is after turn-off of 10 minutes. Looks similar to the turn-on 10 minutes.

B.
*not sure where i put the mic so curves may not be representative of the listener's FR
 

Attachments

  • Bias time course.jpg
    Bias time course.jpg
    121.8 KB · Views: 199
  • Like
Reactions: 1 user
Disabled Account
Joined 2017
I think that issue xplains some ESL distortions. I read somewhere that:
driving high level or xcursion can significantly discharge the ESL.
On some u can even notice volume drop after loud impulse.
The Capaciti are normally quick settlers. Also they turn down the voltage when after few minutes no signal and jump back on input within a second.
Therefore i usually triggered with first sweep, then measured the second one.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.