Hi Calvin,
In the case lowered bass output is acceptable(or there is a need to attenuate resonance peak), it's quite possible to use standard toroids for semi-full range operation.
An active high-pass filter of lets say 6/db oct@50 Hz is very beneficial.
The bass output can be comparable to most book shelf speakers depending on size of the panel.
I have done it with 6 toroids and observed no problems yet.
The amp can be driven into current clipping at highest output levels, but to me the SPL output is quite adequate.
In the picture you can see a subwoofer(which is a dipole) which I don't use
due to integration problems... I simply like bass from an ESL better, and can't back it up by any measurements.
I have to say that the system described works well so far.. And it's not a test setup, I use it quite often.
Regards,
Lukas.
In the case lowered bass output is acceptable(or there is a need to attenuate resonance peak), it's quite possible to use standard toroids for semi-full range operation.
An active high-pass filter of lets say 6/db oct@50 Hz is very beneficial.
The bass output can be comparable to most book shelf speakers depending on size of the panel.
I have done it with 6 toroids and observed no problems yet.
The amp can be driven into current clipping at highest output levels, but to me the SPL output is quite adequate.
In the picture you can see a subwoofer(which is a dipole) which I don't use
due to integration problems... I simply like bass from an ESL better, and can't back it up by any measurements.
I have to say that the system described works well so far.. And it's not a test setup, I use it quite often.
Regards,
Lukas.
Attachments
A few questions concerning your 6toroid setup:
1) What are the power transformer voltage ratings? (ie 230V input, 6V output)
2) For ESL use, how do you have the primaries(6V windings?) of the toroids wired?
All in series? Combination series-parallel?
Hi,
The transformers are rated at 230:11.5 V, 50 VA.
All 11.5V windings are in parallel, 230V series.
0.3 Ohm resistor to reduce DC current.
Effective measured turns ratio ~1:120.
The maximum input voltage before saturation starts is about 13V@50Hz if I remember correctly.
Regards,
Lukas.
The transformers are rated at 230:11.5 V, 50 VA.
All 11.5V windings are in parallel, 230V series.
0.3 Ohm resistor to reduce DC current.
Effective measured turns ratio ~1:120.
The maximum input voltage before saturation starts is about 13V@50Hz if I remember correctly.
Regards,
Lukas.
Last edited:
Thanks for the info, and glad to here it is working well for you.
If I recall correctly, you are using a segmented wire panel so the capacitive load of the ESL would be much lower than for a Jazzman style unsegmented panel like galop is considering. This reduces the requirment for exceedingly low leakage inductance. Did you happen to measure the effective leakage inductance for the 6 toroid setup?
Hi Steve,
No I haven't measured the leakage inductance. However I have a very rough(+/-20% or so) estimation of capacitance reflected to transformer's primary.
My measurements suggested transformers alone are seen like ~5uF capacitor.
With an ESL element it rises to about 7uF.
That means reactive load of transformer array is considerably higher than that of ESL element.
HF resonance point was measured to be around 21 kHz(with panel connected).
Perhaps LC resonance formula could be used to calculate leakage inductance?
In that case it would be roughly 9 uH for all 6 transformers.
By the way, 12 transformers cost about 180eur 5 years ago... Not so cheap though
.Regards,
Lukas.
more on transformers
Hi All
Calvin is correct in some of his conclusions re transformers, but I’m afraid the explanations are quite incorrect.
The voltage rating of a core is proportional to N.A where N is number of turns and A is cross sectional area. The inductance is proportional to N^2. A.
Therefore, as the core gets smaller, the number of turns must increase to give the same voltage rating. Since N^2.A > N. A, the inductance increases as the core gets smaller (for transformer of fixed voltage rating).
The important trends are:
· inductance increases as core gets smaller
· leakage inductance increases as core gets smaller.
· Winding capacitance decreases as core gets smaller
· Unloaded resonant frequency increases slowly as core get smaller
· Voltage rating is independent of core size (ac primary = 230 VAC always, or 115VAC always)
The transformer secondary behaves as an ac voltage source with an RLC filter on the output. The leakage inductance and winding resistance are in series with the output, and the winding capacitance between output and ground.
If you have a single segment ESL, the panel capacitance will be in parallel with the winding capacitance and pull the resonant frequency down. This one is easy to calculate.
If you have a segmented ESL, the panel impedance is more complicated – somewhere between a resistor and capacitor, but it too will pull the resonant frequency down. The ESL impedance is sqrt (R/jwC) where R and C are the resistance and capacitance for each segment – can provide more details if you want – not so easy to calculate and there are a couple of gotchas.
The choice of transformer is actually quite complicated. It is actually a balance between two opposing problems. We can see the effects if we take the two extreme positions.
First – a very small core with large leakage inductance and practically zero winding capacitance. When connected to the ESL with a large capacitive component to the impedance, the resonant frequency = cutoff frequency of the RLC filter will plummet – not good – no bandwidth!
Second – a very large core with large capacitance and practically zero leakage inductance. The resonant frequency will hardly fall at all when the ESL capacitance is added. There are several downsides (i) the RLC filter response will have a very very low Q, so the system behaves more like a first-order low pass filter with a low cutoff frequency, worse still, the Q and cutoff frequency will be very sensitive to resistance in the primary – speaker leads, transformer winding resistance etc (ii) even the smallest inductance in the input circuit – amplifier OP inductance and speaker leads – will pull the resonant frequency way down. These problems get worse with high step-up ratios.
So a balance is required in order to maximise the frequency response and minimise sensitivity to stray impedances in speaker leads, amplifier OP etc. I’m not sufficiently experienced with the single segment ESLs to know off the top of my head, but I would guess than any transformer between 15 VA and 150VA will be fine, especially if you only using a couple of transformers. If you start using many transformers to give decent step-up ratios and voltage rating, the balance becomes trickier I have successfully used 16 x 20 VA transformers to drive a segmented ESL and the leakage inductance is too low (6 mH each, 96 mH total). I should have used 15VA transformers with 16 mH each.
Be warned, not all commercial transformers sold for ESL’s have the balance correct for wide band systems – they can be a very expensive mistake if you are building a big ESL. It pays to do some homework.
Calvin is definitely correct about avoiding transformers with twin high-voltage windings e.g. 2 x 115VAC – these horrible things have a very high winding capacitance and a risk of breakdown.
regards
Rod
Hi All
Calvin is correct in some of his conclusions re transformers, but I’m afraid the explanations are quite incorrect.
The voltage rating of a core is proportional to N.A where N is number of turns and A is cross sectional area. The inductance is proportional to N^2. A.
Therefore, as the core gets smaller, the number of turns must increase to give the same voltage rating. Since N^2.A > N. A, the inductance increases as the core gets smaller (for transformer of fixed voltage rating).
The important trends are:
· inductance increases as core gets smaller
· leakage inductance increases as core gets smaller.
· Winding capacitance decreases as core gets smaller
· Unloaded resonant frequency increases slowly as core get smaller
· Voltage rating is independent of core size (ac primary = 230 VAC always, or 115VAC always)
The transformer secondary behaves as an ac voltage source with an RLC filter on the output. The leakage inductance and winding resistance are in series with the output, and the winding capacitance between output and ground.
If you have a single segment ESL, the panel capacitance will be in parallel with the winding capacitance and pull the resonant frequency down. This one is easy to calculate.
If you have a segmented ESL, the panel impedance is more complicated – somewhere between a resistor and capacitor, but it too will pull the resonant frequency down. The ESL impedance is sqrt (R/jwC) where R and C are the resistance and capacitance for each segment – can provide more details if you want – not so easy to calculate and there are a couple of gotchas.
The choice of transformer is actually quite complicated. It is actually a balance between two opposing problems. We can see the effects if we take the two extreme positions.
First – a very small core with large leakage inductance and practically zero winding capacitance. When connected to the ESL with a large capacitive component to the impedance, the resonant frequency = cutoff frequency of the RLC filter will plummet – not good – no bandwidth!
Second – a very large core with large capacitance and practically zero leakage inductance. The resonant frequency will hardly fall at all when the ESL capacitance is added. There are several downsides (i) the RLC filter response will have a very very low Q, so the system behaves more like a first-order low pass filter with a low cutoff frequency, worse still, the Q and cutoff frequency will be very sensitive to resistance in the primary – speaker leads, transformer winding resistance etc (ii) even the smallest inductance in the input circuit – amplifier OP inductance and speaker leads – will pull the resonant frequency way down. These problems get worse with high step-up ratios.
So a balance is required in order to maximise the frequency response and minimise sensitivity to stray impedances in speaker leads, amplifier OP etc. I’m not sufficiently experienced with the single segment ESLs to know off the top of my head, but I would guess than any transformer between 15 VA and 150VA will be fine, especially if you only using a couple of transformers. If you start using many transformers to give decent step-up ratios and voltage rating, the balance becomes trickier I have successfully used 16 x 20 VA transformers to drive a segmented ESL and the leakage inductance is too low (6 mH each, 96 mH total). I should have used 15VA transformers with 16 mH each.
Be warned, not all commercial transformers sold for ESL’s have the balance correct for wide band systems – they can be a very expensive mistake if you are building a big ESL. It pays to do some homework.
Calvin is definitely correct about avoiding transformers with twin high-voltage windings e.g. 2 x 115VAC – these horrible things have a very high winding capacitance and a risk of breakdown.
regards
Rod
Hi,
Rod is right about the voltage rating, but he quotes only a proportionality, not the equation.
The equation also contains B as factor, the magnetic flux density.
Now B is a result of the magnetic flux and core size, and as such limited and not linear.
The smaller the core the higher B.
Running into the limits of B the core saturates, thereby creating loads of distortion. THD increases already well before the B-limit is reached.
So it´s good to keep B small.
Since power trannies need to be cost-effective devices, the number of turns and the core size is chosen as small as possible. So they function close at the core´s B limit. If one raises the inductance by adding turns of wire he automatically reduces the maximum allowed signal level.
If one wants to allow for higher signal levels there´s no other way as to increase the core size to keep B sufficiently low.
So, just increasing core size alone reduces THD at low frequencies.
If one adds more windings and raises B to the former level, the inductance raises quadratically with the number of turns.
To keep the voltage transformation factor both, primary winding and secondary winding need of course to be changed proportionally.
Improving transformer behaviour at the low frequency limit regarding maximum signal level and signal quality, there´s no other way than to increase core size.
jauu
Calvin
Rod is right about the voltage rating, but he quotes only a proportionality, not the equation.
The equation also contains B as factor, the magnetic flux density.
Now B is a result of the magnetic flux and core size, and as such limited and not linear.
The smaller the core the higher B.
Running into the limits of B the core saturates, thereby creating loads of distortion. THD increases already well before the B-limit is reached.
So it´s good to keep B small.
Since power trannies need to be cost-effective devices, the number of turns and the core size is chosen as small as possible. So they function close at the core´s B limit. If one raises the inductance by adding turns of wire he automatically reduces the maximum allowed signal level.
If one wants to allow for higher signal levels there´s no other way as to increase the core size to keep B sufficiently low.
So, just increasing core size alone reduces THD at low frequencies.
If one adds more windings and raises B to the former level, the inductance raises quadratically with the number of turns.
To keep the voltage transformation factor both, primary winding and secondary winding need of course to be changed proportionally.
Improving transformer behaviour at the low frequency limit regarding maximum signal level and signal quality, there´s no other way than to increase core size.
jauu
Calvin
Hi,
I do not think large amount of turns has much to do with primary inductance, but rather stray inductance and parasitic capacitance.
That's two big problems of audio transformer design.
The rise in THD is typically very sudden and begins when approaching saturation point. 10-15% below this and there can be no signs of saturation.
For sure source impedance should be low.
I am not trying to tell that power toroids are the best choice for full range
operation. Array of it gets large and heavy with loads of parasitic capacitance.
In my case weight was used as a counterweight for ESL stands
Still it's doable and can yield quite adequate results for everyday use.
Regards,
Lukas.
I do not think large amount of turns has much to do with primary inductance, but rather stray inductance and parasitic capacitance.
That's two big problems of audio transformer design.
The rise in THD is typically very sudden and begins when approaching saturation point. 10-15% below this and there can be no signs of saturation.
For sure source impedance should be low.
I am not trying to tell that power toroids are the best choice for full range
operation. Array of it gets large and heavy with loads of parasitic capacitance.
In my case weight was used as a counterweight for ESL stands
Still it's doable and can yield quite adequate results for everyday use.
Regards,
Lukas.
Last edited:
Well,
such a solution is not easy but it was already done somewhere in Finland.
1st of all, cheap core has quite high coercive force meaning high magnetizing current somewhere around 30-50 a/m. Magnetizing current is subtracted from the signal current - distortions, proportionate to the primary x-former resistance
Secondly permeability varies in respect to excitation voltage - more distortions.
Thirdly, high stepup ration requires a lot of turns in the windings thus capacitance.
Fourthly, high AC voltage and the air in the x-former insulator creates barrier discharge thus insulation deterioration.
What needs to be done:
1. Use the core with lower coercive force. There are materials with Hc around 1 a/m.
2. Preferably the material shall have rather linear permeability, again such materials exist.
It should be around eu 16 per kilo.
3. Now we can safely increase core cross-section 20-30 times with the distortions at least on par (certainly lower) than the ones in a cheap halo-lamp x-former. So the number of turns thus much lower capacitance and winding volume - leakage inductance goes down as well
4. Now the hard part: you have to have solid, preferably polyethylene inter-winding insulation potted in vacuum - no discharge in air free material.
EDIT: In real life usually the oil comes into the rescue... but it is hygroscopic, expands with temperature, ages and so on. On the other hand have a look on the horizontal deflection x-formers...
such a solution is not easy but it was already done somewhere in Finland.
1st of all, cheap core has quite high coercive force meaning high magnetizing current somewhere around 30-50 a/m. Magnetizing current is subtracted from the signal current - distortions, proportionate to the primary x-former resistance
Secondly permeability varies in respect to excitation voltage - more distortions.
Thirdly, high stepup ration requires a lot of turns in the windings thus capacitance.
Fourthly, high AC voltage and the air in the x-former insulator creates barrier discharge thus insulation deterioration.
What needs to be done:
1. Use the core with lower coercive force. There are materials with Hc around 1 a/m.
2. Preferably the material shall have rather linear permeability, again such materials exist.
It should be around eu 16 per kilo.
3. Now we can safely increase core cross-section 20-30 times with the distortions at least on par (certainly lower) than the ones in a cheap halo-lamp x-former. So the number of turns thus much lower capacitance and winding volume - leakage inductance goes down as well
4. Now the hard part: you have to have solid, preferably polyethylene inter-winding insulation potted in vacuum - no discharge in air free material.
EDIT: In real life usually the oil comes into the rescue... but it is hygroscopic, expands with temperature, ages and so on. On the other hand have a look on the horizontal deflection x-formers...
Last edited:
Hi,
I understand the reasoning behind trying to find core materials that are more linear.
But as long as source impedance is low the distortion of the transformer can be low too even for power toroids(~0.03% or so over the audio band).
Thats a fraction compared to a typical ESL element.
When tube amps are used as a source it's a completely different story as high source impedance would amplify nonlinear effects of transformer.
Sure not all toroids are manufactured the same and some experimentation is necessary.
Lukas.
If you are interested in knowing the leakage inductance and total winding capacitance of your 6 transformer setup it can be calculated from a few resonance measurements. Basically, measure the resonance frequency for the unloaded transformer, and for the addition of some known capacitances. A spread in resonance frequencies is desireable, so in your case, perhaps 500pF, 1000pF, and 1500pF. Or, whatever capacitors you happen to have on hand. I can do the calculations for you if you would like.
Details on the technique here:
http://www.diyaudio.com/forums/plan...ruct-cube-louver-acoustat-14.html#post2181730
I posted graphical results for the Antek toroids here:
http://www.diyaudio.com/forums/plan...p-up-measurements-part-1-2-a.html#post2823887
Unfortunately Finnish web page does not exist anymore, where all of the above was implemented. ESL were huge as well: 2sq meters or so.
Second of all we are talking about full range i.e. 30-50 Hz minimum frequency and x-former is running at full swing.
90% efficiency of 100W X-former means few tenths of Ohm primary resistance and approximately .5A of core related current (hysteretic/magnetizing), if we put losses as 2.5W pri 2.5w sec 5w core. And it's not near 0.03%... not mentioning the "tube" source
Surely at higher frequencies B drops substantially, thus a core related distortions.
Is this the setup you were referring to?
Do It Yourself - Electrostatic Speakers - Project: Full range trafo for ESLs by Arto Kolinummi
Do It Yourself - Electrostatic Speakers - Project: Full Range ESL by Arto Kolinummi
jer
Do It Yourself - Electrostatic Speakers - Project: Full range trafo for ESLs by Arto Kolinummi
Do It Yourself - Electrostatic Speakers - Project: Full Range ESL by Arto Kolinummi
jer
In my earlier experiments of stacking cores back in 2010 I had found that the more iron I used the low frequency THD did get less.
I had adjusted the number of primary turns as I added each core for the same volts per turn factor as well in order to keep everything the same except for the added iron.
I was using a frequency range of 10hz to 30hz at only about 2v to 4v peak on a average on the primary while I monitored the signal on the scope coming off of the secondary winding.
My overall ratio was around 1:160 I think (or maybe 1:128).
With 4 of my 210watt cores I was actually able to get a decent 15hz sine waveform and a fairly clean one at 30Hz.
With just one core it was quite distorted and non-symmetrical.
I was not pushing the voltage to cause any saturation of the core as the whole test was to see if the added iron does reduce the overall low frequency THD characteristics of transformer setup.
I determined that it did lower it, Although I don't have any THD data from those times as I didn't have a sound card setup to do such measurements at the time.
Here is a more recent THD test of the Antek (1206) core that Charlie had sent me,
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
Posted is the a HOLMimpulse chart.
Disregard the rising THD at the high end of the scale and any other abnormalities in the middle of the sweeps in the REW tests as I found out later that those were caused by my motherboard sound chip acting up at random times.
Under normal conditions the THD was quite low and almost reaching the limits of my sound card at about .005%THD.
Under full power conditions with a Crown DC300A the reading I was getting were basically that of the amp itself and showed no significant sign of added THD.
However at frequency's below 1Khz there was a slight rise in THD when it got just to the edge of saturation.
I have not determined if this was due to the core or the amp it self but it was nothing to be alarmed about as it was still quite low ( < .08% or so).
Even at the onset and into full saturation at 300HZ and 115Vp-p the THD as measured from the secondary output was still only a surprisingly low .5% according to Visual Analyzer.
At this point it is the current demand from the amplifier is quite high as expected and is shown in the last graphs in the above link.
If this core had double the amount of turns its low frequency impedance would be twice as high and in a much safer range for the amplifier as it was dipping below 2 to 1 ohms at that point.
I had posted a preliminary impedance sweep test in an earlier thread.
The ESL does not present any load on the amplifier at this point and all of the power is wasted as heat within the core ( and the amp) itself.
It is the Primary inductance that determines the load on the amp at low frequency's.
A good compromise would be to design the transformers primary to have a slightly higher impedance say 16 to 32 ohms or so as this would help to keep the total turn count low thus a low leakage inductance as well.
This can be accomplished by using a bigger core (as mentioned) while keeping the same amount of turns.
The higher impedance will also allow the amplifier to swing its full voltage capability and this would increase the performance of the ESL panel at the lower frequency's in the critical midrange of 300hz to 3000hz.
The nice thing about using several cores is the the leakage inductance is added as each core is added.
If you were to double the amount of turns on a single core instead using two of them of the same size in series the leakage inductance would then Quadruple instead of double unless you use a bigger core.
This is caused by the winding having a mutual inductance from being on the same core.
If they don't have a mutual inductance going on then they are simply added together.
Affordable cores can be found if one wanted to take the time to wind them at Alpha Core, Just start with the largest one that you can afford.
They even have C types and exotic materials as well if you want to pay the extra cost for them.
Silicon Steel Toroidal Cores - In Stock
I posted a while back a paper on the performance of such exotic materials and good ole' silicon steel was the best performer for the lowest frequency's and the others were superb for the highest frequency's.
At the highest frequency's silicon steel was okay but the exotics were much much better and a good compromise was 80/20 (20%nickel) for low cost and 50/50 for a good well rounded full range transformer.
I have no idea as to how much such cores cost at this time.
Once I get some more time I will go over my measuring technique article (that I did start) and verify my findings and measuring methods for all to understand once the winter starts to set in.
I learned a lot earlier this year, and, after about 2 years of it, it all finally clicked in.
FWIW
jer
The measurement in the chart is calibrated (Top Blue line) and are of 20Vp-p, 10Vp-p and 5Vp-p Sweeps into the primary (12v configuration) of the Antek AS-1206 and measured (captured) from the secondary side through a 1000:1 resistor divider of about 20megohms.
The second Blu line is the phase.
The measurement was made with no extra capacitance load applied and is that of just the transformer alone.
I had adjusted the number of primary turns as I added each core for the same volts per turn factor as well in order to keep everything the same except for the added iron.
I was using a frequency range of 10hz to 30hz at only about 2v to 4v peak on a average on the primary while I monitored the signal on the scope coming off of the secondary winding.
My overall ratio was around 1:160 I think (or maybe 1:128).
With 4 of my 210watt cores I was actually able to get a decent 15hz sine waveform and a fairly clean one at 30Hz.
With just one core it was quite distorted and non-symmetrical.
I was not pushing the voltage to cause any saturation of the core as the whole test was to see if the added iron does reduce the overall low frequency THD characteristics of transformer setup.
I determined that it did lower it, Although I don't have any THD data from those times as I didn't have a sound card setup to do such measurements at the time.
Here is a more recent THD test of the Antek (1206) core that Charlie had sent me,
A TEST JIG FOR FINDING ESL STEP-UP TRANSFORMER PARAMETERS
Posted is the a HOLMimpulse chart.
Disregard the rising THD at the high end of the scale and any other abnormalities in the middle of the sweeps in the REW tests as I found out later that those were caused by my motherboard sound chip acting up at random times.
Under normal conditions the THD was quite low and almost reaching the limits of my sound card at about .005%THD.
Under full power conditions with a Crown DC300A the reading I was getting were basically that of the amp itself and showed no significant sign of added THD.
However at frequency's below 1Khz there was a slight rise in THD when it got just to the edge of saturation.
I have not determined if this was due to the core or the amp it self but it was nothing to be alarmed about as it was still quite low ( < .08% or so).
Even at the onset and into full saturation at 300HZ and 115Vp-p the THD as measured from the secondary output was still only a surprisingly low .5% according to Visual Analyzer.
At this point it is the current demand from the amplifier is quite high as expected and is shown in the last graphs in the above link.
If this core had double the amount of turns its low frequency impedance would be twice as high and in a much safer range for the amplifier as it was dipping below 2 to 1 ohms at that point.
I had posted a preliminary impedance sweep test in an earlier thread.
The ESL does not present any load on the amplifier at this point and all of the power is wasted as heat within the core ( and the amp) itself.
It is the Primary inductance that determines the load on the amp at low frequency's.
A good compromise would be to design the transformers primary to have a slightly higher impedance say 16 to 32 ohms or so as this would help to keep the total turn count low thus a low leakage inductance as well.
This can be accomplished by using a bigger core (as mentioned) while keeping the same amount of turns.
The higher impedance will also allow the amplifier to swing its full voltage capability and this would increase the performance of the ESL panel at the lower frequency's in the critical midrange of 300hz to 3000hz.
The nice thing about using several cores is the the leakage inductance is added as each core is added.
If you were to double the amount of turns on a single core instead using two of them of the same size in series the leakage inductance would then Quadruple instead of double unless you use a bigger core.
This is caused by the winding having a mutual inductance from being on the same core.
If they don't have a mutual inductance going on then they are simply added together.
Affordable cores can be found if one wanted to take the time to wind them at Alpha Core, Just start with the largest one that you can afford.
They even have C types and exotic materials as well if you want to pay the extra cost for them.
Silicon Steel Toroidal Cores - In Stock
I posted a while back a paper on the performance of such exotic materials and good ole' silicon steel was the best performer for the lowest frequency's and the others were superb for the highest frequency's.
At the highest frequency's silicon steel was okay but the exotics were much much better and a good compromise was 80/20 (20%nickel) for low cost and 50/50 for a good well rounded full range transformer.
I have no idea as to how much such cores cost at this time.
Once I get some more time I will go over my measuring technique article (that I did start) and verify my findings and measuring methods for all to understand once the winter starts to set in.
I learned a lot earlier this year, and, after about 2 years of it, it all finally clicked in.
FWIW
jer
The measurement in the chart is calibrated (Top Blue line) and are of 20Vp-p, 10Vp-p and 5Vp-p Sweeps into the primary (12v configuration) of the Antek AS-1206 and measured (captured) from the secondary side through a 1000:1 resistor divider of about 20megohms.
The second Blu line is the phase.
The measurement was made with no extra capacitance load applied and is that of just the transformer alone.
Attachments
Last edited:
Is this the setup you were referring to?
Do It Yourself - Electrostatic Speakers - Project: Full range trafo for ESLs by Arto Kolinummi
Do It Yourself - Electrostatic Speakers - Project: Full Range ESL by Arto Kolinummi
jer
Yep but... it's not the original site. I have even guesstimated core types/size, turns numbers... some posts ago.
Last edited:
Hi,
I would be interesting how the relationship between core losses and distortion is extrapolated?
As far as I know non-linear effects of the core comes in parallel with transformer's promary(with winding resistance and source impedance in series).
Therefore source as well as primary winding impedance is a very important factor.
I have measured this in practice too; changing the value of series resistor to primary winding has a considerable effect on THD values(higher R => larger THD).
And please understand I have not put 0.03% THD from the air; it's based on real world measurements; within range of about 100-10 kHz and with significant input voltage(I do not remember exact values but it's something like~10Vrms). Perhaps going down an octave would increase the distortion somewhat; still it's a lot less than what can be expected from ESL element in this frequency range.
Regards,
Lukas.
Last edited:
Agreed.
As long as all impedances in series with the primary winding are small, the current distortion from non-linear core effects does not produce large amounts of distortion in the secondary voltage. Well, with the caveat that the amplifier must be able to produce the required current without generating additional disortion of its own.
But, as soon as you put a large impedance in series with the primary(like a capacitor for a passive crossover) then core properties become very important for minimizing distortion.
the poor OP got run over with an avalanche of interesting but way too technical information. The poor fellow was just trying to figure out how to make HV for the bias and how that was different than the driving signal... geez.
_-_-
@alexberg, hope ur kidding? if what you were saying were true, there would be the need for only *one* amplifier, and the wide range we now have would all sound identical. Is that the case?
_-_-
@alexberg, hope ur kidding? if what you were saying were true, there would be the need for only *one* amplifier, and the wide range we now have would all sound identical. Is that the case?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.