ESL hybrid

[esltransformer]

excellent work

Yes, 2.35ohm if leakage inductance and winding capacitance is ignored. Actual impedance will be significantly less. Most solid state amplifiers rated for 4 ohm loads seem to have no problems if the impedance drops to 1.5 - 2.0 ohm in the top octave since there doesn't tend to be high amplitude audio content there.

Using the parastics posted for the Antek-0506 it is easy enough to calculate response and impedance with good accuracy using this spreadsheet:
http://www.diyaudio.com/forums/plan...101-electronics-challenged-2.html#post2555863


In the attached plots, C= ESL capacitance, R= resistance in series with primary
Pic #1: Shows impedance and response trends for increasing capacitive load with R = 0 ohm

Pic #2: Shows trends for varying R, with C = 1200pF
This is the typical capacitance for a "jazzman-style" flat, unsegmented ESL, for which the Anteks are used.
Note that R = 1 ohm provides a -3dB slope in the top octave which is exactly what is needed flatten the +3dB slope of the on-axis response of an ESL line source. Reduction in R might be desired if listening is done slightly off-axis.

Pic #3 and #4: Shows trends for varying C with R = 1 ohm and 2 ohm.

Pic #5: Shows impedance and response trends for increasing capacitive load with R adjusted for maximumally flat stator response.

 

[esltransformer]

I found another old measurement.
I did make in the past a transformer for a hybride esl, 1:115, never used it but last year i gave it to a friend.
Unfortunately no impedance measurements are left but i know for sure that it will be not very low (above 8-10Ω at least at 10kHz).

An externally hosted image should be here but it was not working when we last tested it.

 

[geraldfryjr]

Last night I found a calculator that finally helped me find an answer that I have been searching for, For the last three years!!

What is the Permeability of these cores?

On the page of this calculator it states that the permeability iron is about is about K=200.
Using the my measured data of inductance's of the winding's and knowing the turns, I was able to find out exactly the permeability of the core material that Antek is using and is about K=287.5 .

Here is that calculator,

Inductance of a Toroid

So then I plugged in my data and it all coincides the what I have measured using the AS-1206(modified).
Using the data that Bolserst had posted about the AN-0506 core I was able to interpolate what the actual inductance values should be.

I found much to my surprise that the 50 watt core does indeed have more inductance then that of the 100 watt core.
By as much as 70% to 75% more!!
The AS-1206 26 turn primary measures at about 1.01mh.
The interpolated inductance for the AN-0506 44 turn primary is 1.757mh.
This is great to know especially as it makes for a higher impedance for the ampilifer on the lower frequency side of the curve at 5.52 ohms at 500hz.

When you add a second core to double the transformation ratio the impedance will still be in about the 4 ohm range at 500hz.
As it comes at to 5.52 ohms with one core and 2.76 ohms for two cores and by the time you add another ohm 1 for HF dampening we are now at 3.76 ohms for 500Hz.

However this comes at a cost with 70% to 75% or so of more self capacitance as well.

I measured about 580pf to 700pf depending on the HV winding's series connection configuration.
Using these very same figures shows the increase of capacitance in the the AN-0506 at 1000pf to 1300pf was posted here,

http://www.diyaudio.com/forums/plan...p-up-measurements-part-1-2-a.html#post2823635

Here is my data that I used for this comparison.
I also used the advertised transformation ratio of 1:40 for each core as well.



AN-0506, AS-1206(modified)

Core Permeability K=287.5

Core Area , 25mm(ht)X18mm 4.5^2cm, 36mm(ht)X22mm=7.92^2cm

Toroidal Radius(r) , 2.85cm, 3.075cm

Primary turns , 44T, 26T

Secondary turns , 880T, 520T

Secondary total turns , 1760T, 1040T

Primary inductance , 1.757mh, 1.01mh

Secondary inductance , 703.1mh , 404mh

Secondary inductance total , 2.8124h , 1.607h

Leakage inductance , 2.5mh, 6-8 mh (6.8?)

All of my measured data is within or better than +/- %5 of these calculate as a worst case figure.
In fact the only one that is off is the inductance value of the single secondary winding's at 430mh as measured by themselves.
I am not sure why this is as all of the rest of the measurements are right on with the predicted ones.

The leakage inductance was as measured from the primary winding with the secondary winding's shorted together and verified by both Visual Analyzer and RLC Bridge programs.

My earlier figures using the resonate method gave me some strange results and this may have been due to the electrostatic shield that was in place when I made them.
The primary inductance value coincides the the impedance values in my curve here,

http://www.diyaudio.com/forums/planars-exotics/161485-step-up-transformer-design-6.html#post3404300

Of approximately 4.12 ohms at 650Hz using this reactance calculator,

L or C Reactance Calculator

P.S. The layout looks great in edit mode, But when I post it, it all gets scrunched up, So I added the comma's to separate the column's. :/

 

[esltransformer]

It is not so easy to say what the permeability is.
I doubt if your transformer has a permeabilty of 287.5 unless it has a gap.

See page 62 of this pdf: http://www.wiltan.co.uk/client_files/default/wiltan_guide.pdf

Permeabilty is very complex.

Last night I found a calculator that finally helped me find an answer that I have been searching for, For the last three years!!

What is the Permeability of these cores?

On the page of this calculator it states that the permeability iron is about is about K=200.
Using the my measured data of inductance's of the winding's and knowing the turns, I was able to find out exactly the permeability of the core material that Antek is using and is about K=287.5 .

Here is that calculator,

Inductance of a Toroid

So then I plugged in my data and it all coincides the what I have measured using the AS-1206(modified).
Using the data that Bolserst had posted about the AN-0506 core I was able to interpolate what the actual inductance values should be.

I found much to my surprise that the 50 watt core does indeed have more inductance then that of the 100 watt core.
By as much as 70% to 75% more!!
The AS-1206 26 turn primary measures at about 1.01mh.
The interpolated inductance for the AN-0506 44 turn primary is 1.757mh.
This is great to know especially as it makes for a higher impedance for the ampilifer on the lower frequency side of the curve at 5.52 ohms at 500hz.

When you add a second core to double the transformation ratio the impedance will still be in about the 4 ohm range at 500hz.
As it comes at to 5.52 ohms with one core and 2.76 ohms for two cores and by the time you add another ohm 1 for HF dampening we are now at 3.76 ohms for 500Hz.

However this comes at a cost with 70% to 75% or so of more self capacitance as well.

I measured about 580pf to 700pf depending on the HV winding's series connection configuration.
Using these very same figures shows the increase of capacitance in the the AN-0506 at 1000pf to 1300pf was posted here,

http://www.diyaudio.com/forums/plan...p-up-measurements-part-1-2-a.html#post2823635

Here is my data that I used for this comparison.
I also used the advertised transformation ratio of 1:40 for each core as well.



AN-0506, AS-1206(modified)

Core Permeability K=287.5

Core Area , 25mm(ht)X18mm 4.5^2cm, 36mm(ht)X22mm=7.92^2cm

Toroidal Radius(r) , 2.85cm, 3.075cm

Primary turns , 44T, 26T

Secondary turns , 880T, 520T

Secondary total turns , 1760T, 1040T

Primary inductance , 1.757mh, 1.01mh

Secondary inductance , 703.1mh , 404mh

Secondary inductance total , 2.8124h , 1.607h

Leakage inductance , 2.5mh, 6-8 mh (6.8?)

All of my measured data is within or better than +/- %5 of these calculate as a worst case figure.
In fact the only one that is off is the inductance value of the single secondary winding's at 430mh as measured by themselves.
I am not sure why this is as all of the rest of the measurements are right on with the predicted ones.

The leakage inductance was as measured from the primary winding with the secondary winding's shorted together and verified by both Visual Analyzer and RLC Bridge programs.

My earlier figures using the resonate method gave me some strange results and this may have been due to the electrostatic shield that was in place when I made them.
The primary inductance value coincides the the impedance values in my curve here,

http://www.diyaudio.com/forums/planars-exotics/161485-step-up-transformer-design-6.html#post3404300

Of approximately 4.12 ohms at 650Hz using this reactance calculator,

L or C Reactance Calculator

P.S. The layout looks great in edit mode, But when I post it, it all gets scrunched up, So I added the comma's to separate the column's. :/

 

[bolserst]

I found another old measurement.
I did make in the past a transformer for a hybride esl, 1:115, never used it but last year i gave it to a friend.
Unfortunately no impedance measurements are left but i know for sure that it will be not very low (above 8-10Ω at least at 10kHz).

I had a couple questions concerning your measurements:

1) Do you know what the input impedance was of your measurement setup? The amplitude of the resonance peak for the "no load" and "load=200pF" should not be the same if there is no change in damping resistance between the two cases.

2) When adding the 25K resistors in series with the primary, do the measurements show voltage across the secondary winding? or voltage across the 200pF capacitor.

3) Curious if you recall number of primary turns, and core size. Based on frequency of resonance for "no load" and "load=200pF" leakage inductance would have to be quite lower, perhaps around 5uH, and winding capacitance similarly low around 90pF.



@geraldfryjr
You may recall we had discussed core permiability in the other transformer thread. It is a function of flux density which is directly proportional to input voltage, inversely proportional to core area. For most grain oriented silicon steel transformer cores, a good rule of thumb is an intial permiability of 1500 at low input voltage. The permiability increases with input voltage amplitude(flux density), to a maximum of 5000 to 8000 depending on material, and then abruptly dropping toward zero as the core saturates.

So, as esltransformer mentioned, trying to nail down a single value for primary inductance is difficult. Similarly, it is difficult to compare between transformers when core areas and primary turns will result in different flux densities in their cores.

 

[esltransformer]

For measurements i use a normal "cheap" audio amplifier (NAD 320). I always measure with a 1:10 probe on my Fluke or Tektronix scoop.
I measure between zero and each phase and if a secondairy resistor is connected i still measure on the transformer wire because i want to know what the transformer dos.

I used for this hybride transformer a SU60A c-core with 40 primary turns and 4400 secondary. Original i calculated 38 turns (that makes it a 1:115 transformer) but with 40 turn the final version is of course 1:110. A small differance.

I also found the saturation level back: 20Vrms at 200Hz.

I had a couple questions concerning your measurements:

1) Do you know what the input impedance was of your measurement setup? The amplitude of the resonance peak for the "no load" and "load=200pF" should not be the same if there is no change in damping resistance between the two cases.

2) When adding the 25K resistors in series with the primary, do the measurements show voltage across the secondary winding? or voltage across the 200pF capacitor.

3) Curious if you recall number of primary turns, and core size. Based on frequency of resonance for "no load" and "load=200pF" leakage inductance would have to be quite lower, perhaps around 5uH, and winding capacitance similarly low around 90pF.



@geraldfryjr
You may recall we had discussed core permiability in the other transformer thread. It is a function of flux density which is directly proportional to input voltage, inversely proportional to core area. For most grain oriented silicon steel transformer cores, a good rule of thumb is an intial permiability of 1500 at low input voltage. The permiability increases with input voltage amplitude(flux density), to a maximum of 5000 to 8000 depending on material, and then abruptly dropping toward zero as the core saturates.

So, as esltransformer mentioned, trying to nail down a single value for primary inductance is difficult. Similarly, it is difficult to compare between transformers when core areas and primary turns will result in different flux densities in their cores.

 

[esltransformer]

@geraldfryjr
You may recall we had discussed core permiability in the other transformer thread. It is a function of flux density which is directly proportional to input voltage, inversely proportional to core area. For most grain oriented silicon steel transformer cores, a good rule of thumb is an intial permiability of 1500 at low input voltage. The permiability increases with input voltage amplitude(flux density), to a maximum of 5000 to 8000 depending on material, and then abruptly dropping toward zero as the core saturates.

So, as esltransformer mentioned, trying to nail down a single value for primary inductance is difficult. Similarly, it is difficult to compare between transformers when core areas and primary turns will result in different flux densities in their cores.

Maybe EI core material has a maximum permeabilty of 8000 but a toroid with very good low loss FeSi can be up to 90000 (if you have HiB material. I use this material for my c-cores)

 

[bolserst]

For measurements i use a normal "cheap" audio amplifier (NAD 320). I always measure with a 1:10 probe on my Fluke or Tektronix scoop.
I measure between zero and each phase and if a secondairy resistor is connected i still measure on the transformer wire because i want to know what the transformer dos.

I used for this hybride transformer a SU60A c-core with 40 primary turns and 4400 secondary. Original i calculated 38 turns (that makes it a 1:115 transformer) but with 40 turn the final version is of course 1:110. A small differance.

I also found the saturation level back: 20Vrms at 200Hz.

20Vrms(200Hz) sound about right for SU60A c-core with 40 turn primary.

The best fit I could find for your transformer is:
Leakage inductance = 4uH
Winding capacitance = 110pF

The main thing I couldn't match was the Q of the unloaded resonance. Perhaps at the high frequencies there are other losses not included in the model that would damp the resonance. Adding just 0.1 ohm in series with the primary would bring the unloaded resonance peak down in line with the 200pF curve.

Estimated impedance at 10khz would be about 10ohm unloaded, or 4 ohm with 200pF load.
If used to drive a 1000pF load, impedance at 10kHz would be ~ 1 ohm.

...a toroid with very good low loss FeSi can be up to 90000 (if you have HiB material. I use this material for my c-cores)

Agreed. The values I gave were for more common M6(and similar) material.
For ESL step-up use where the primary is driven from a low impedance source, the HiB material is not nearly as advantageous as it is for tube amplifier output transformers.

 

[bolserst]

...
Transformer : AN-0506, AS-1206(modified)
Leakage inductance: 2.5mh, 6-8 mh (6.8?)
...
P.S. The layout looks great in edit mode, But when I post it, it all gets scrunched up, So I added the comma's to separate the column's. :/

Thanks for the info on the AS-1206

Couple questions on your leakage inductance measurements:
1) Should the units be uH instead of mH?
2) Are the values for a single transformer? or two wired for ESL step-up use.
3) Did you have an AN-0506 handy to compare with the AS-1206? or were you using the 2.5uH value I had posted previously for two AN-0506s.

 

[geraldfryjr]

The leakage measurements are uh as in microhenries.

Thank you for catching that!!!
Sorry for the confusion, My Bad !!!!

The values are for a single transformer.

I don't have a AN-0506 to test otherwise I would have used one for a proper comparison so I used the value you posted.

I am looking at a test right now and the values are jumping around at about 8.5uh for a 26 Turn primary.
And 2.9uh for A 13 turn primary winding.
This looks about right.
I hope a better sound card will fix this jumping around issue !!!!
I will recheck this again with a resonate test again a little later.

jer

 

[esltransformer]

This is the reason why i like graphs double logarithmic:

An externally hosted image should be here but it was not working when we last tested it.

btw: the impedance of the transformer are real measurements

 

[geraldfryjr]

I understand what you mean now!!

Since the equations are basically linear the lines will be straight whether you use a linear or log scales but not if you use both linear and log on the same graph.
I did this same mistake when I created my core saturation graphs here,

http://www.diyaudio.com/forums/planars-exotics/161485-step-up-transformer-design-5.html#post2194027

jer

 

[esltransformer]

.

Estimated impedance at 10khz would be about 10ohm unloaded, or 4 ohm with 200pF load.
If used to drive a 1000pF load, impedance at 10kHz would be ~ 1 ohm.

For a esl unit i guess 200pF would be enough if you have a step-up ratio of 1:115.
I prefer small units because i don't like this typical "sweet spot" that most esl have. For the same reason i like electrical segmentated esl units, small units with wires.

 

[bolserst]

For a esl unit i guess 200pF would be enough if you have a step-up ratio of 1:115.
I prefer small units because i don't like this typical "sweet spot" that most esl have. For the same reason i like electrical segmentated esl units, small units with wires.

My preference is for segmented wire ESL units with lower capacitive load as well.
However, this thread was started for transformers used with "jazzman" style large flat stators which have capacitive load of 1000pF - 1500pF.
So, I added the comment on what would happen if the load was 1000pF instead of 200pF.

This is the reason why i like graphs double logarithmic:

This reminded my of old joke from college days...In lab class if your experimental data wasn't following expected linear trends, just plot it on LOG-LOG scales.
Almost all experimental data looks better behaved when plotted LOG-LOG.

Seriously though, plotting impedance for circuits with reactive components in LOG-LOG format can often provide better insight into which components are dominating the impedance, and relative level of damping at resonance. Attached below is re-plot of transformer data from post#28.

I believe STAX used to plot impedance for their ESLs with LOG-LOG scale.

 

[esltransformer]

I think the Antek 0506or 1206 are both not very suitable transformers for the "Jazzman" esl unit, unless you have a very powerfull and expensive amplifier (capable of driving almost 0 Ohm). But even if you have such amplifier it is not a good design goal. The load is way to much and a lower step-up would be much better.

The Anteks are nice transformers for very less money but a proper designed esl-transformer is much better in any case (except price....?)

 

[tyu]

So The people like jazzman an others that are useing the Anteks tranfourmers on there ESL... have no other tranfourmer to comepar too? So are you saying there just lost in the Price? Cost is so littel that thay just THINK it good?
Anyway thanks for all the work an grafs but i gess most here have been misled.
I gess i have been wating for one of you guys to say if these sound Good?

 

[esltransformer]

Compared to most normal speakers esl sounds always special .
If a good design is used a esl would be as any normal speaker easy to drive (and sounds even better )

So The people like jazzman an others that are useing the Anteks tranfourmers on there ESL... have no other tranfourmer to comepar too? So are you saying there just lost in the Price? Cost is so littel that thay just THINK it good?
Anyway thanks for all the work an grafs but i gess most here have been misled.
I gess i have been wating for one of you guys to say if these sound Good?

 

[geraldfryjr]

Yes, I have to agree.
The draw back of using power toroid transformer is not having a high enough primary inductance to keep the impedance at a higher safer level in order to let the amplifier have its highest voltage swing at the lower frequency's.

ESL's don't use very much power at all for the lowest frequency's but do require a high voltage swing.
Due to the low inductance of the primary a majority of the power in this range is dissipated by the transformers core itself and wasted as heat.

Of course on the other end of the frequency spectrum you have the extra stray transformer capacitance issue that also lowers the impedance in relation to the step up ratio.
But with the typical rise in frequency response of an ESL when compensated also helps to greatly cuts down on the current requirements of the amplifier as well.

I can get about 10Vp to 15Vp out of my cheapy amp before it starts to struggle at about 20Vp into a 1:256 ratio and a panel of about 50pf
The panels capacitance is only a fraction of what the transformer has.
At this level it is very loud and by using a bigger amp I start running in to stator coating issues although I have done it with some success.

Since this thread is about Hybrid ESL systems there are many reasons to consider the use of smaller sized panels as far as width is concerned.
And/or a wider segmented panel if you want a little more low end "oomph" out the them.
Besides creating a easier load for the amplifier such panels also provide for a wider horizontal dispersion and this is very desirable especially if you have a rather large listening area.

Another great question that comes along is how much voltage is needed to run this things.
My answer to that is how loud do you want them to go?!!!
A larger panel will fair better from the advantage of having more surface area and will need less voltage swing to reach a desirable SPL.

Increasing the bias voltage helps considerably as well.
I have found that every doubling of the bias voltage increases the sensitivity by about +6db and this reduces the demand on the amplifier by the same amount.

Now, I have no experience building larger panels (yet), But ,I do know that they can really get up there in the SPL's if driven properly as do my smaller desktop models.
Even my small ones can scare you out of the room with +108db or so at 1 meter!!!
But it does take a lot of voltage to get to this kind of level.
This is not so bad using a larger panel with more surface area.

When the idea of using the power toroids came about they weren't readily available here in the states until recent years.
But they were the first source of good quality iron at a cheap and affordable price in many years.

I know this, Because, I searched high and low for good iron back in 2003 when I made my first panels and toroid's didn't come into the picture until some 4 or 5 years later due to availability.

I bought 10 cores for cheap for some amplifier projects in 2007 and it was 2.5 years after that was when I discovered that they have been tested for use as ESL step up transformers.

This leads me to were I am now and researching everything I can about them.
Now that I have a very good understanding about them I can now consider a ground up step up transformer design.

TYU is right, A good DIY recipe is badly needed and is what I have been striving for all this time.

There is no reason why we should have to spend $400 for a device that cost only $90 when it first came about 10 years ago that uses less than $40 worth of materials because they are the only ones that makes such a device (Plitron)!!

Yes, toroid cores are tedious to wind but they are readily available and very affordable now.
And the best we to combat all of the above issues is to use a rather large core.
Such a core can be had for less than $50 or even less depending on the size you choose to use and how low of a frequency you want to design for.

This was not the case not so long ago even with EI cores especially of good suitable material for use with audio device.
Unfortunately with all of this happiness going on, the cost of materials has gone up considerably but it is still affordable and doesn't cost +$400 per core!!!

Do they sound good, You ask?!!!

Yes, They sound very good as far as the iron is concerned, as long as you stay away from core saturation at the lowest frequency of about 240Hz 300HZ at 40Vrms input.

At this high of a level I have measured THD's of about .3% to .5% until saturation and then it shoots through the roof.
Remember this is with my Crown DC300a at full output.
Then the THD's drop off considerably and was very low above 1Khz.

I have a few charts on this but I need to switch to a higher quality sound card in order to get some definite measurements.

At a 20Vrms level input the THD's were very very low and I could not tell if there were any added by the transformer at all in the critical audio band (midrange and up).

Listening to only one of my desktop model ESL's at only a 5v to 10v peak signal form my cheapy amp I just sit in awe and can't believe the detail that I am hearing and still get to 99db SPL range.
More than enough loudness and oh so clean and detailed as they should be so that I haven't even bothered to hook them up in stereo yet.

Even though I am not running Antek cores on my setup, Antek core material exhibits pretty much the same quality and characteristics as my budget cores from Parts Express do and I will have more on this later.

jer

 

[bolserst]

I think the Antek 0506or 1206 are both not very suitable transformers for the "Jazzman" esl unit, unless you have a very powerful and expensive amplifier (capable of driving almost 0 Ohm). But even if you have such amplifier it is not a good design goal. The load is way too much and a lower step-up would be much better.

The low load at f > 10kHz is in large part due to the high panel capacitance. If you want HF bandwidth to extend past 20Khz with this size unsegmented panel you need a transformer with very low leakage inductance, which the Antek 0506 has. You also need suitable power handling capability at the crossover frequency, which the Antek 0506 has. This is why it is recommended for the jazzman panels. There is no way around these requirements. Even if you designed a special transformer you would still need to have the same low leakage inductance value for upper bandwidth limit to extend past 20Khz.

The load seen by the amplifier is a separate issue and, as you noted, is reduced if a lower step-up ratio is used. The Antek 0506 have dual 6V windings which are used as the primaries, so the wiring configuration could easily be changed to reduce the step-up ratio from 74:1 down to 37:1. The impedance would then be > 4 ohm at 10kHz…a much easier load. Of course your amplifier would then need to put out 2x the voltage(4x the power) to reach the same SPL. Or, looking at it another way, your achievable peak SPL with a given amplifier would be reduced by 6dB. This would be a useful experiment for somebody to try to see what changes in sound quality and dynamic SPL capability they notice. CharlieM?

I guess i have been waiting for one of you guys to say if these sound Good?

All who have tried them agree they sound good, with active or passive crossovers.
The issue that is being brought up is, would they allow your amplifier to sound better if configured for a lower step-up ratio.

As a point of reference, pretty much all the ML ESLs have impedance that drops to 1 ohm or less at f > 10Khz.
This is just what is physically required if you want to match sensitivity for typical dynamic woofer with unsegmented panels.

 

[tyu]

An Dont take what i am saying the roungway....It know ones place to tell me are anyone what for them to use...your all doing great work for all to see...
For years all i have heard..An read....is the Acoustat stepup tranfourmers were junk!
An we can get better sound on the Acoustat panels with better iorn...
An the ML Tranfourmer are low cost junk..an we can do better...
An i have had the Simmit here an the setup troi Tranfourmer in the Over $10k looks just like 1 size wize the Antek...an some say these are great sounding speakers...

Funny thing is i have had SoundLab,Acoustats,ML, Quad,KLH,JanZen,ESL an with work thay all sound great with stock EI cores ....I like all i gess just wont better.....thanks for all your time...