Esl 63 what type of coil for the delay line?

[hilbren]

Im thinking about building a pear of ESl 63
I already found the schematic, but each time i find the schematic de coil values are missing.

Does anybody know what type of coils these are?

Greetings,
Hilbren

 

[bear]

I don't know if I have that info...

There was a schematic published in the Loudspeaker & Something Handbook, a British author... (John Woram??)

You can likely discover the values by figuring out the required delay and calculating the time constant... to my eye they look
fairly large value air coil type... they may have an iron core there,
dunno. And, it may be set up as a single large "tapped" delay line,which is what I would expect to see.

Also there may be one or more articles that talk about the delay time...

Which I think you can calculate because the delay, iirc, is set to be equal to the speed of sound over the distance from the center to the edge of the diaphragm - perhaps with some minor empirical tweaking to get the speed of the transverse wave through the mylar dialed in...

It's supposed to approximate a spherical source.

_-_-bear

Quite frankly I don't think I'd try to build one unless ur going to do
something way bigger or different - its easier and cheaper just to buy a pair if ur interested in that sort of design...

 

[hilbren]

I've got the stepup trafo's and i've got mylar....
i want to build something like this one http://esl.hifi.nl/project04.htm

 

[bear]

Sure, nice project.

Done by an advanced builder, with a Quad delay line to start with.

My suggestion is to start by building the easiest and simplest ESL you can manage and make sure you can get it to work first. Then move up in complexity and size...

This project is quite complex and requires the 100% perfect implementation of a great number of precise and critical tasks and constructions.

_-_-bear

 

[hilbren]

ow I forgot to tell, I already build some esl's:
http://home.hccnet.nl/g.c.van.doorn/hilbren/electrostatic.htm

 

[george a]

IIRC
The delay lines for the ESL 63 were described in Elektor magazinr about the time the speakers were originally released.
Also memory seems a bit hazy but i think each concentric ring on the stators were each individually (signal) delayed to approximate the spherical source and the coils feeding each ring were made from very (hair) thin wire but the length of wire in each was extraordinary , (thousands of metres???) to approximate acoustic delay through air eg"
1mS through air at 330m/s = 0.33m
how long would a piece of wire be to impart 1mS delay???

approx 300km ?
Delays required between each ring was however a lot less than this so wire length was also a lot shorter..


I could be wrong though..

Cheers

 

[hilbren]

I already did a little research, but the delay line is based on the fact that a LC circuit has a 90° phase shift on the resonance frequency.

So If you take a LC circuit with a resonance frequency 11khz you will get a delay (phase shift) of 22,7 uSec and that is ~8mm @11khz

And if the frequency gets higher de phase shift will get 180°

All lower frequencies don’t have any delay and leave the circuit without any phase shifting…

 

[george a]

and furthermore

from Stereophile mag review http://www.stereophile.com/loudspeakerreviews/416/index6.html

"The second innovation was the speaker's unique radiating element, which used driver plates that employed a printed circuit board of annular rings, like the ripples formed when a stone is dropped into a lake. These rings were fed by delay lines (employing some 11 miles of wire!) which allowed the flat diaphragm to radiate the sound first at the center and last at the periphery, as if it were a radiating sphere—the ideal shape for approximating sound emanating from a point source with an apparent location 12" behind the panels. "

Can you post schematic???

Cheers

 

[hilbren]

11miles=17.71Km
speed of electriciti is 299 792 458 m / s http://www.google.nl/search?hl=nl&q=speed+of+light&btnG=Google+zoeken&lr=

that is (17,71km)/(299 792 km/sec)
that is 56,7uSec wich is not enough for the complete delayline

 

[george a]

a quick calc tells me approx 87mm delay distance (acoustic) required for stator ring approx 600mm diam. max (for example )
what wire length gives electrical delay equal to 87mm for a sound wave???

Cheers

 

[hilbren]

your calculation sounds good, its almost the same as the total delay line of a quad esl63.
but i dont think its done with the extreem long wire. i think its done with LC circuit and not on the fact that eletricity needs time to travel trough an extreem long wire...

ive got some theroies from Peter Baxandall i will post some screenshots from his book to explane the theorie...

 

[AndrewJ]

For a thorough analysis of the ESL63, read the chapter on electrostatic louspeakers by Peter Baxandall in the book Loudspeaker and headphone Handbook, edited by John Borwick. (ISBN 0 240 51371 1)
Peter Baxandall was a good friend of Peter Walker and had a full mind dump from him for writing the chapter, along with contibuting much understanding from his own experiments. There is no better treatise available on the subject.
The theoretical and practical work that went into devising the ESL63 was extensive, as you will see from the book. It was an astonishing undertaking to develop and shows just how much more PJW contributed to the science of electrostatics than anyone else.
I admire your wish to try to experiment with your own solution, bet be prepared for a lot of work!!!

Good luck

Andrew

 

[bear]

Yeah... Borwick... Loudspeaker & Headphone Handbook.

That was the book I was trying to recall...

There was a schematic there, but I doubt that it had values for the inductors. The hair thin wire is probably because of the number of turns required to get the requisite L... with the signal being essentially low current and high voltage, I guess he was able to get away with fairly thin wire keeping the coil size modest.

I've had then open on the bottom... to change out the cheap binding posts for some better ones... didn't recall anything remarkable looking about the coils at the time.

_-_-bear

 

[hilbren]

I did some caculations that explane the lenght of the wire, i came out with 4,5 H thats a BIG coil

 

[rcavictim]

You cannot use a wound coil of a long piece of ordinary wire as a delay line, it will be an inductor in that configuration. The problem is the mutual magnetic coupling between all the turns. A large inductor will only cause all the high frequencies to be lost.

You can however make a delay line with a long piece of coaxial cable. It is acceptable to roll this cable in a coil to make it more compact. You will still have a HF slope rolloff issue which is normal for coaxial cable and is a linear relationship to the length of the run. This can be compensated for in most cases with a suitable pre-emphasis network to artificially boost the highs prior to insertion into the coaxial run. DC resistance of the coaxial cable center conductor will not be trivial and it's effect on the system will depend solely on the current flow which is a function of the voltage supplied by the amplifier and the impedance of the loudspeaker. Simple ohms law calculation. If the cable run is the same impedance as the speaker then half of the power will disappear as heat in the coaxial cable. An electrostatic driver is an extremely high Z so losses in the coaxial cable will be greatly reduced compared to trying to run an 8 ohm dynamic driver.

 

[hilbren]

yes but the speed of elektricity in copper is 230 000 000 m/sec

so if i want a delayline with 100us delay (total), i need 2,3km of coax

but it is an option, I only need a very compact coax cable

another idea based on the coax...:
what if i wind a coil with for example 1000 winding to the left and the another 1000 to the right. the the coil isnt inductive.... but the delay still exists...?? would that work?

 

[bear]

You cannot use a wound coil of a long piece of ordinary wire as a delay line, it will be an inductor in that configuration. The problem is the mutual magnetic coupling between all the turns. A large inductor will only cause all the high frequencies to be lost.

You can however make a delay line with a long piece of coaxial cable. It is acceptable to roll this cable in a coil to make it more compact <snip>

Obviously, this is incorrect.

If it were so, there would be no Quad 63.

Peter Walker explained how this works with ESLs back in the 1950s... and yes it uses inductors.

There is a configuration known as an All Pass filter... it provides
delay without HF loss.

_-_-bear

 

[rcavictim]

Obviously, this is incorrect.

If it were so, there would be no Quad 63.

Peter Walker explained how this works with ESLs back in the 1950s... and yes it uses inductors.

There is a configuration known as an All Pass filter... it provides
delay without HF loss.

_-_-bear

I heard the delay line in the '63 uses some 12-14 kM or miles of wire.

You can use inductors, but not one single large one to make a delay line. It has to be a long string of L, then a C to ground, L, C to gnd, L, C to gnd, etc, etc, etc. That is the description of a long piece of coaxial cable! HF loss compensation will be required. Sometimes known as 'tilt'. The delay seen is some fraction of the speed of light based on the velocity of propagation in the cable (VP). Usually this is around 66% C in solid dielectric polyethylene coaxial cables.

What exactly was incorrect?

 

[bear]

I heard the delay line in the '63 uses some 12-14 kM or miles of wire.

Yes, but this is merely a coincidence - the requirement being to get sufficient L in order to have the requisite time constant given the value of C (the cells) being low.

Then length of the wire is a "factoid" that is used for marketing hype and discussion - nothing per se to do with the electrical delay.

You can use inductors, but not one single large one to make a delay line. It has to be a long string of L, then a C to ground, L, C to gnd, L, C to gnd, etc, etc, etc. That is the description of a long piece of coaxial cable! HF loss compensation will be required. Sometimes known as 'tilt'. The delay seen is some fraction of the speed of light based on the velocity of propagation in the cable (VP). Usually this is around 66% C in solid dielectric polyethylene coaxial cables.

It seems that you have not seen the schematic for the 63?

What exactly was incorrect?

That the length of the wire was what delayed the signal.
This delay line in the 63 is not the case of a coaxial cable made delay line.


_-_-bear

 

[rcavictim]

Bear,

No I haven't seen the 63 schematic, but I know a little about delay lines, and it is the velocity of propogation that sets the delay. I will reiterate that a single L and C will not do the trick, all you get from that is a low pass filter.

You don't have to have an actual length of coaxial cable though. You can simulate it elecrically in other physical forms. One example. Make what superficially looks like a large bobbin would air core choke. This would be a layered air coil but each layer would have a copper foil shield above and below the turns of wire on that layer. The turns would be spaced on each layer to reduce capacitive crosstalk to adjacent turns.

A copper wire with enamel insulation sandwiched between two copper foil ground planes is electrically an actual 'transmission line'. A form of this is also manufactured commercially called tri-plate line or printed on circuit boards this way for microwave use. What you would see in schematic presentation of such a device would be a series L followed by a bypass C to ground (the foil shield), then another series L followed by a bypass C, etc, etc, etc.

Quad or anyone else reputable isn't going to advertise 14 kM of wire in the crossover delay network unless it is true. Your comment about coincidence and factoid nonsense for better play on advertising is IMO your own mistaken misbelief.