My ESL attempt

Status
Not open for further replies.
Wound up a couple of new toroids today.

300VA core instead of 80VA and the level above 6k is up abt. 5-10dB! 😀

ELESH24.jpg


Do'nt like the peak at 125Hz, but plan to let the woofer handle everyting up to abt. 300Hz.

/R
 
That graph looks quite normal, except for the HF roll off around 6kHz. A +15dB peak at the resonance frequency and from there on a rising frequency response which is normal for a voltage driven ESL.

The most likely candidate for the roll off is still the toriod. You got a nice improvement from that new transformer, so we're probably looking in the right direction. You should try to minimise the leakage inductance (together with the capacitance of the panel it will give a second order low pass filter). Probably by interleaving the windings a lot, but I'm not a transformer expert.

The little peak at 18kHz could be the resonance of the leakage inductance and the capacitance of the panel, in which case the roll off at 6kHz is caused by something else, perhaps too much capacitance in the transformer? The combined capacitance could be too much for the amp? Just thinking out loud here.

Perhaps you could measure the basic properties of the transformer and panel (capacitance, inductance etc)?
 
Hi,

I didn´t deburr the sheet metal. The roundings stem from the fact that I spray laquered each side of the panel 8 times! I wasted a complete 5L can of laquer to coat just 4 sheets -- too bad that holes take the laquer soo badly :xeye:

The fequency response looks nearly typical, except three points.
The first is that the peak of the response in the upper midrange area usually takes place around 1-4kHz.
The second is that the falling response above 6kHz is stronger than it should.
The third is that the red curve is....hey what does show anyway?? It can´t be acoustic phase response since the bump at fo is missing and it can´t be the impedance response plot because that looks completely different too.

Btw: 1:80 transformation factor? As I understand You made Your own tranny? If not, 1:80 is a very uncommon relationship. With a standard 6V/230V tranny-pair You could expect a factor of just below 1:70. A 300VA core is very oversized for this task, 30VA would already be ok powerwise. I´d rather stick to a smaller core. I have very good results with 80VA-120VA cores and the Amplimo toroids uses a 100VA core afaik.
My assumption is, that the winding resistances might be a bit high, especially on the primary side. The panel´s impedance and the winding resistance form a frequency dependant voltage divider. Or did You measure with a series resistance between amp and panel?
The bandwidth of the toroid-panel-system seems to be quite ok with 18kHz, since its no easy task to design a tranny that offers a bandwith ~18kHz with such a capacitive load and such high voltage levels. A measurement of the impedance would tell a lot more about the tranny´s qualities.

Ohh, don´t wanna be nasty, but one Q: How long took the placing of the mic to get that response? 😉 Especially flat panels exhibit a high degree of directivity. To find the sweet point where things look ok -not to speak from great- is not easy. On axis it might even look horrible and just a few fegrees off axis the HT-response sucks. Together with the strangeness of the red curve´s behaviour I assume not only probs with the tranny-panel-system but within the measurement setup too.

jauu
Calvin
 
The trannies are wound for 5V/230V.
I quickly measured my old ones (80VA) this morning. They measure 0.9mH on the primary and 0.1mH with the secondarys shorted - do'nt remember the secondary reading.
The panel measures 0.7nF.
The roll off at high frequencys is probably aided by the measuring system that is all but perfect!
Placment off the mic was quite rough, but I get similar curves so I guess they at least point in the right direction. No series resistance.
I'll try and make a impedance reading, just have to get the hardware first!
The red trace is for the right chanel - just forgot to uncheck it in the program - ignore!
 
OK, 0.1mH leakage and 0.7 nF capacitance.

The capacitance transformed to the primary with 1:80 stepup is 7e-10 * 80 * 80 = 4.48e-6 (F). Together with 0.1mH leakage inductance and the well known f0 = 1/(2*pi*sqrt(LC)) we get:

fo = 7523 Hz

Which is a bit higher than expected from looking at the measured frequency response, but close enough for me. With a panel capacitance like that you need *really* small leakage inductance to get any high frequency performance.

A seperate (smaller!) tweeter panel would make the job a lot easier. You also want this to get any kind of acceptable dispersion.
 
Hi,

my panels feature higher capacitances and I never came across a standard toroid tranny that was so bad that it gave such low resonance frequencies around 8kHz. Usually the resonance was well outside the audible band and even with my large panels (150x40cm, C>2nF) this resonance just dropped to 18.5kHz. It would need a really badly wound EI-core to have the resonance as low as 8kHz.
I assume, that as long as the measurement can´t be regarded reliable in the HT-range and as long as there is no impedance measurement, we won´t solve the prob.

jauu
Calvin
 
Hmm... yes the figures don't add up, do they? I thought about it and found the mistake: I should not transfer the capacitance to the primary for this calculation. Sorry for the confusion.

The correct calculation would add the panel capacitance and the internal capacitance of the transformer secondary and combine that with the leakage inductance. Since we don't know the internal capacitance of the secondary the calculation can not be done at this point.
 
Hi,

to give some comparable values I might cite from the Amplimo/Plitron datasheets.

Ratio: 1:75
secondary Inductance (max.): 1700mH
effect. sec. leakage ind.: 26mH
prim. resistance: 0.138Ohms
sec. resistance: 248Ohms
effect. sec. internal Capacitance: 320pF

2nd-order resonance freq (with 1nF load): 27.157kHz

based on the formula fo=1/2pi*sqrLC with C as capacitance value of the panel + eff. sec. internal Capacitance (remark: not the cap-value reflected to the primary side!)

jauu
Calvin
 
Hi,

I´m a bit puzzled. The reason is, that the resonance frequency in the Amplimo-Datasheets seem to be calculated with simple capacitance values, while normally I´d use the reflected (to the primary side) values (as it is stated in RonWagner´s Book et al). Too, the leakage inductance values seem to be a bit high. One could expect values of 1/10 of the datasheet values.
On the basis of reflected values the Amplimo would should show a resonance frequency of just 11.45kHz with a 1nF load.
My own measurements with those trannies were much better and correlated with the datasheet values which stated a -3dB-freq of >20kHz.
??????

jauu
Calvin
 
The toroids are hooked up att the moment, but I'll measure them this evening!

The transformers are single primary and single secondary.

Using two toroids - primarys in parallell, secondarys in series.

I'm using insulated standard "AC house wire" , wound for 5V on the "primary".

/R
 
Motivation hav been down the last days, slipping tape, arcing and to top it all off - I got the flu!

The stators have now been coated with 5 layers off laqeur, and I'm going to use PVC spacers instead of tape.

I measured my toroids, and I'm a bit confused!

Toroid1: Prim 1mH Leak 0.01mH
Sek 2.07mH Leak 50mH

Toroid2: Prim 2.05mH Leak 0.01mH
Sek 4.25mH Leak 50mH

If I do'nt remeber things wrong, they were the exact same type off toroids (220V/27V).

Testing them with 220V AC, I get 5.05V & 5.1V on the "prinary" with no load!

/R
 
Status
Not open for further replies.