Driving the Beveridge ESL

Jan,
When reading your answer correctly, the ESL’s primairy and secundary side are untouched, only a relay is inserted in the secundary side to the stators. This relay could posibly present a capacitive load to the step-up transformer creating these dips.
Easy to find out when you connect without relay.

But when everything is untouched, this means that the ESL’s are producing sound while testing ??

Hans
 
AX tech editor
Joined 2002
Paid Member
Yes Hans, they are 'playing' when testing, only not music ;-)
But they play music quite well too!
The relays have 2.5pF to the coil and 0.1pF across the contacts when open.
I don't think that has an impact.
I have to think this over, I mean, I can understand that when you have a transformer with all kinds of impedance correction parts, it's unlikely that the transfer function from primary to secondary is exactly 1:120.00 or whatever the 'official' ratio is over the freq range.
So if you feed it a flat response you should expect a non-flat response at the secondaty side, no?

Jan
 
Yes, good point. But the fact that the dd amp is not affected, apparently, the load itself doesn't seem the issue. The transformer and surrounding parts may be.
I need to do some more tests today. I am suspicious about that LF bump as it shows in all cases.

Jan
Jan,

I’m not with you with your conclusion that the load isn’t the issue.
As Demian mentions the source impedances of both versions are incomparable, where the transformer is way more sensitive to load variations.
So either the relay, the attenuator or interference with the DD amp may cause those anomalies that are absolutely not there in the very close to reality simulation model.
But, since you have the option available, I would use Dirac Life to compensate the step up version to match your DD version, that way the dips are nicely counter corrected, whatever their cause may be.

Hans
 
AX tech editor
Joined 2002
Paid Member
Jan,

I’m not with you with your conclusion that the load isn’t the issue.
As Demian mentions the source impedances of both versions are incomparable, where the transformer is way more sensitive to load variations.
So either the relay, the attenuator or interference with the DD amp may cause those anomalies that are absolutely not there in the very close to reality simulation model.
But, since you have the option available, I would use Dirac Life to compensate the step up version to match your DD version, that way the dips are nicely counter corrected, whatever their cause may be.

Hans
Hans, I am going through all the settings on the RME, I suspect that there's some loudness setting or something I missed.
It's a menu system that's totally alien to me; it is logical, very much so, but it just seems incompatible with my brain ;-)
Will take me some time to work through. I will report back later today, hopefully.

Jan
 
AX tech editor
Joined 2002
Paid Member
OK, my bad, forget about my post # 97.
I found that some of the EQ and loudness settings from an earlier experiment were still enabled in the RME. Red face :-(
So I redid the measurements, first doing a freq response on the input signal to the direct drive amp as well as the Purify/paX as a baseline check; that should be a flat line, and it is, reasonbly.
Next I did a freq response at the connection to the two stators. The Purify that drives the speakers through the stepup transformer/network shows, among others, a roll off above 10kHz. The direct drive much less as it has not been eq'd yet. (Level has not been eq'd either so disregard absolute differences).
Hans, does the stepup network freq response look credible? The roll-off looks very similar to what I've been measuring on the ESL63 months ago.

Jan
 

Attachments

  • panel fresp dd vs. Purify.PNG
    panel fresp dd vs. Purify.PNG
    8.5 KB · Views: 52
  • amp input fresp.PNG
    amp input fresp.PNG
    8.6 KB · Views: 58
AX tech editor
Joined 2002
Paid Member
OK, next step. This is fun!
I setup an eq in the FLEX for the DD channels.
Re-measured the freq resp at the stators, and after some iterations I got this.
Maybe time again to listen to some music ;-)
Hans and Demian, thanks for keeping me on the straight & narrow!

Jan
 

Attachments

  • on panel fresp dd eq1.PNG
    on panel fresp dd eq1.PNG
    8.7 KB · Views: 61
  • peq3.PNG
    peq3.PNG
    14.2 KB · Views: 58
Normally, in an ESL the diaphragm is biased to a high DC voltage, and the two stators are driven in opposite phase.
That's correct.
Looking at the Beveridge circuit, it seems different here.
It sure does.
It appears that the diaphragm is DC grounded and AC driven from one phase, while both stators are at opposite DC voltages (+/-3200V) but AC driven with the same signal, in opposite phase to the diaphragm.
Correct.
What do you think?
If the panels were absolutely symmetrical, it would result in zero acoustic output.
The only way to make it produce sound is to make the panels quite asymmetrical. It would make it a nice 2nd-harmonic generator.
 
AX tech editor
Joined 2002
Paid Member
Nice work. How does it sound on DD?
Im surprised your sending <40Hz to the panels. Seems a bit ruthless?
To be honest, to me it sounds very much similar to the 'normal' drive, there are some subtle differences but I need much more listening time with my old ears to say something sensible. For now I feel there's more detail, especially in voices.
(I will demo the whole setup later this year to a large group of people and we'll see what they will say).

I should cut off the lows probably, but for now I was just trying to emulate the freq response of the normal drive.

Jan
 
  • Like
Reactions: 1 users
Final matching, between the direct drive and the paX solid state through the stepup xformers.
Measured directly at the stators.

I'm confussed. Why limit your DD response to match the transformers? Isnt that like someone buying an electric supercar and throwing away all the power by programming the power response to match a combustion engine. Sorry i dont understand :(
 
But that response curve set by the trannie looks less then ideal. If you blew 300grand on the latest e-sports whatever would you tell the engineers to reprogram the e-output to limit its abilities to the old ICE model? I get that would match the power so you have a fair handling vs handling comparison. But its still a comparison skewed and limited to the old model.

Wouldn't a better real world comparison be to get both options performing as well as they can at the listening position?

If I had a DD Id aim to maximise its performance by cutting out the harms freq extremes can do:
1. work out how hard and low the ESL can be driven by the DD at full listening volume
2. when the membrane slaps or HD escalates at normal full volume Id put in a high pass to protect the ESL and unburden the DD so the volts can go where they make music
3. then Id do one of those clever home online hearing tests and low pass where my high hearing freq cuts and unload the ESLs in the scary high freq zone
4. then Id be looking for dialling in the best freq response at the listening position

But thats just me
 
But that's for tomorrow ;-)

Tomorrows today! How does it sound? I'm so excited!!!

Matching the freq to the iron feed was, I must admit ,a fiendishly cunning plan. Any subtle differences? Midrange details emerging maybe? Timbral transparency?

Do you need to be careful with the DD bandwidth? Does the transformer limit bandwidth at the extremes and protect the current amp from the horrible low impedance and nasty phase at the freq extreme? Do you every use a NOS DAC with high US? Without the trannie could the DD be driven into oscillation even without NOS torture?