QUAD 63 (and later) Delay Line Inductors

Now that you bring up the Crosby mods for the 63,
What was accomplished with these mods ?

B8EFEDCD-6095-4EEE-8AE7-9B7241B785FE.jpeg

EF7D2590-D2B0-4091-B577-6482C890A4D7.jpeg

Regards
David
 
Member
Joined 2004
Paid Member
A lot depends on where your values are. Most of the changes do provide a specific benefit. Not shown are the changes to the frame (added stiffness and mass) and the dust cover (.5 uM Mylar), screens and grill cloth which seems to be much more open. The net effect I perceive to be additional clarity and resolution. The wire upgrade would be the most controversial given the amount of work and the always contentious issues around wire. The Teflon caps make a big difference since they are in the main audio path with LOTs of voltage across them. The increased resistance on the bias supplies helps keep the charge on the diaphragm and not migrating. Given the stock binding posts Quad used lots of room for improvement. Those were the best we could find back then.

Do you have any specific questions?
 
IF you should change cables on the high voltage side the wire should be as THIN as possible reducing stray capacitances.
People who suggest thick wires I advice to do some studying or just stay away...

I can remind you that the delay inductors has miles and miles of wire thinner than a human hair. Get it? Walker himself struggled a lot to avoid stray capacitance's. The reason why internal signal wire is as thick as it is is only for mechanical and production reasons.
 
… I'm still not sure what they did in the delays that is so different from Quad. The delay line construction is different being cascaded LP filters instead of all pass networks
The difference is in what virtual source is trying to be emulated. For the Quad, it is a virtual point source placed some distance behind the diaphragm. The virtual wavefront is still curved/expanding as it passes thru the diaphragm, which the rings and uniform delay per section attempt to match. The wavefront stretches well beyond either side of the diaphragm, which is only capturing a section of the expanding wave. Often the virtual source is described as a section of a pulsating sphere, rather than section of expanding wavefront from a point source. Either way, since only a section of the source is captured, even in the idealized case, the polar response is not uniform with increasing frequency or smooth with angle. Walker, added losses to the delay lines to address reflections at the sectional termination.

Mellow’s FrontRo ESL is based on a similar virtual source concept, the difference being the source he is emulating is an oscillating sphere. A physical oscillating sphere source has uniform dipole directivity at all frequencies. In theory if the virtual source is built with many rings and delays increasing with ring number the surface of the sphere is completely captured(ie no termination issues), and it too will have uniform dipole directivity at all frequencies. This is the big differentiator from Walker’s concept, where even a theoretically perfect implementation of the section of a pulsating sphere still has directivity that is not uniform with frequency or smooth with angle. The original concept documented in JASA and Patents involved active delays and multiple amplifiers. The passive implementation in the FrontRo loudspeaker used only a few rings, resulting in the need to add some low-pass filtering to the delay networks for the outer rings to keep the polar response smooth at higher frequencies where wavelength approaches ring width.


A few links related to the topic:
FrontRo Loudspeaker website:
Electrostatic — FrontRo

Acoustics Text the attached figures were taken from:
Publications — FrontRo

JASA article on virtual oscillating sphere concept: (also contains attached figures)
https://www.frontro.co.uk/s/Dipole_speaker_with_balanced_directivity.pdf

Patent on vitual oscillating sphere concept US20120033834A1
US20120033834A1 - Apparatus With Directivity Pattern
- Google Patents


JAES Article on Passive implementation:
https://www.frontro.co.uk/s/How-do-...tic-loudspeaker-with-constant-directivity.pdf
 

Attachments

  • Quad_Concept.png
    Quad_Concept.png
    134.6 KB · Views: 207
  • Mellow_Concept.png
    Mellow_Concept.png
    77.5 KB · Views: 186
Member
Joined 2004
Paid Member
I think i follow now and I don't see why the '63 could not be modified to emulate this different approach. I looked at the patent and it would be really hard to defend in light of a Quad 63. Nothing in the claims differentiates their disclosure from a '63. But you need a patent for a startup to get funding.

They will have a big problem since another company calls their product "Frontrow". That won't be easily overcome in the US at least. And i can find nothing on pricing.
 
Now that you bring up the Crosby mods for the 63,
What was accomplished with these mods ?

Well the silly and worthless wire was melted into the panels slots making an ungodly mess of the panels, but maybe this was done by another party. I don't know I just buy the speakers without knowing their history. The workmanship I've seen from these mods is not great either, lots of soldering iron melting of the stator supports. I generally pay LESS for a set of crosby modified speakers than I would for a stock set because I have to clean up the mess. The holes I've seen cut for the silly binding posts weren't always lined up well etc.

Mechanically the new resistors and caps don't fit well and are not well supported.


In contrast I turn down the shoulders on gold plated 5-way binding posts on my lathe so I don't have to cut the metal on the quad itself.

There's some value to the more open grilles, thinner dust covers. Stiffening up the internal panel supports and such. But the last foot of wire after miles if thin magnet wire is just plain silly.

Sheldon
 
Last edited:
I spotted this for sale along with a Crosby sheet showing different levels of mods
availble.
The price was too high for the set and might have bought it had it been cheaper
Someone got it

It looked interesting and would have liked to hear how it changed the speakers.
It appears some serious thought might have gone into these mods though

Regards
David
 
Member
Joined 2004
Paid Member
I think the last Crosby mod to a Quad was 20 years ago. Jerry Crosby, Richard Lees and I worked on that for several years before it morphed into a "product". They were a lot of work to build. The wire was a PITA and something I was never too enthusiastic about but Richard was convinced so that became a part of the project. You had to make the slots in the stators for the wire to clear. not easy to do and risky. It was never a kit but Jerry did train his Italian distributor in the mod and ship parts to him.

Still a lot more real effort than Quad has put in in 30+ years.
 
I think that Crosby mod seems quite ok except for the wires... i did not mean to sound besserwisser, sorry for that. It would be interesting to change wires to much thinner ones to reduce stray capacitanceI have the metal grids removed and it would be interesting to model the capacitance of the speaker with and without the metal grid. And with/without grounding of the metal grid.
Anyone?
 
Member
Joined 2004
Paid Member
When I have the broken one open again I may be able to take a few measurements. I think the capacitance is really small for the wires. You could open the simulation and check the effective source impedance at the various nodes to see how much capacitance is needed to have an effect.
 
I think i follow now and I don't see why the '63 could not be modified to emulate this different approach. I looked at the patent and it would be really hard to defend in light of a Quad 63. Nothing in the claims differentiates their disclosure from a '63. But you need a patent for a startup to get funding.

They will have a big problem since another company calls their product "Frontrow". That won't be easily overcome in the US at least. And i can find nothing on pricing.
Demian,

I do seem to agree with you.
The two images showing the extension of the radius R for the Esl63 and line perpendicular to the membrane from the end of R for the FronRo are very suggestive but misleading IMO.

Fact is that both systems seem to have equidistant time delays.
That the HF polar diagram is nicer for the FrontRo has much to do with the fact that the various frequencies are less overlapping,hence less interference and the higher the frequency, the smaller the radiating area.

But as you mentioned, I don’t see a reason why the ESL63 delay line filters couldn’t be modified the same way.
Maybe Steve could give his opinion on the matter.

Hans
 
Member
Joined 2004
Paid Member
I think the polar issues with the '63 are more from the physical implementation. Certainly the physical limitations need to be understood and addresses before getting to involved in other revisions. Everything from the ribs on the stator support to the frame and the top trim will have an impact and cause diffraction. With both modern software (Comsol etc.) and manufacturing techniques a better stator and frame could be designed with some fiberglass in the right places around the frame. It would improve the focus both on and off axis (probably by just enough that us obsessives would notice). Atkinson did mention some chaotic behavior in the decay and it could very well relate to those physical issues.

Conceptually, these speakers are really large acoustic lenses and both electronic and acoustic tweaks could make for significant improvements. Estimating acoustic transmission loss of perforated filters using finite element method - ScienceDirect
 
Conceptually, these speakers are really large acoustic lenses and both electronic and acoustic tweaks could make for significant improvements. Estimating acoustic transmission loss of perforated filters using finite element method - ScienceDirect

Very interesting link and covering exactly a topic that keeps me busy.
Since I don’t want to pay $36,- for this paper, what are your impressions when projecting this on a ESL63 stator in terms of attenuation versus frequency ?

Hans
 
I think i follow now and I don't see why the '63 could not be modified to emulate this different approach.
The ESL-63 rings certainly could be used for emulating an oscillating sphere, but their sizes are not optimized for the task.

I looked at the patent and it would be really hard to defend in light of a Quad 63. Nothing in the claims differentiates their disclosure from a '63.
Sorry, I had meant to say this was a Patent Application, not an approved patent. But, looking again, I see now that revision B of the application was approved in 2014, (US8831248B2 - Apparatus with directivity pattern - Google Patents). I also notice the US patent is owned by Nokia, but the UK versions of similar patents are owned by Mellow Acoustics.

I’m not well versed in patent law, but from few dealings with legal dept. here at work I know that to be patentable your device must fit into at least one of several categories. The most common is an improvement on previous patent, something that could be argued. Some weaknesses in the approach from the Walker patent are called out in the first section. An improvement patent allows you to keep others from using your patented method, but you may still have to pay royalties to the patent on which yours is an improvement. If the method is “novel”, or not an obvious extension of existing patent it can be used without infringing the improved upon patent. At least that was how I understood it.

Speaking of Walker and Patents, I always wondered why the Kellogg patent from 1934 wasn’t cited as prior art. It describes using LC transmission lines to drive segmented ESLs and the advantages of easier amplifier load and improved dispersion. Funny that Kellogg is well known for his contributions to dynamic drivers, but not ESLs.
US1983377A - Production of sound
- Google Patents


They will have a big problem since another company calls their product "Frontrow". That won't be easily overcome in the US at least.
Yup, that certainly could be a problem for marketing in the US if they pursue it.
 
…The two images showing the extension of the radius R for the Esl63 and line perpendicular to the membrane from the end of R for the FronRo are very suggestive but misleading IMO. Fact is that both systems seem to have equidistant time delays.
The time delays for the 2 approaches are not the same. I will try to clarify the differences with additional figures.

Walker envisioned the spherical wave propagating from a point source towards an ESL diaphragm. If he sectioned off the ESL into ring stators he could delay and reduce the amplitude of the pressure generated by the ring to match that of the propagating wave as it passed thru the plain of the diaphragm. To capture the whole wave you would need an infinitely large diaphragm.
Attachment #1: illustrates delay for the 6th ring
For each ring -
Delay = sqrt(b^2 + y^2) – b
Attenuation Factor = b / sqrt(b^2 + y^2)


Mellow envisioned an ESL diaphragm sectioned off into rings that could be delayed and driven at uniform pressure to emulate acoustic radiation from the surface of an oscillating sphere. The ESL need only be as large as the oscillating sphere.
Attachment #2: illustrates delay for the 6th ring
For each ring -
Delay = a - sqrt(a^2 - w^2)….using variables from Figure in Attachment #2
Delay = b - sqrt(b^2 - y^2)….using variables from Figure in Attachment #1
Attenuation Factor = 1.0


=========================================================
Years ago I made many analytical comparisons of the two techniques. Here is one comparing polar performance for a 80cm diameter ESL divided into 100 ring segments using the two techniques. No losses or low passing added, to clarify the differences in the two theoretical concepts.

Attachment #3 shows the virtual point source placed 30cm behind the ESL. Recall a point source provides flat response at all off-axis angles. The 80cm diameter section of the point source wavefront does provide flatish response out to a reasonable off-axis angle, but is marred by ripples. If you could build one with theoretically ideal delays and drive levels, it would be provide you with this less than ideal acoustic behavior. In the ESL-63, Walker utilized lossy delay line as well as segment 7 to smooth out the ripples and polar response trend. Never ceases to amaze me what he was able to accomplish without the aid of modern computer simulation.

Attachment #4 shows the oscillating sphere provides the near uniform dipole polar response as advertised. If you could build one with ideal delays and uniform drive, it would provide you with a very nice smoothly trending acoustic behavior. Easier said than done though, with those large and increasing delays.

NOTE: Plots are shown in ARTA, but these were based on theoretically calculated impulse responses, not measurements.

Just to be clear, I am not saying that the practical implementation of either of these concepts is by default superior to the other. I was just trying to point out a relatively recent theoretical design concept that seems to hold at least as much promise as Walker’s brilliant idea.
 

Attachments

  • Virtual_Point.png
    Virtual_Point.png
    52.6 KB · Views: 198
  • Virtual_Sphere.png
    Virtual_Sphere.png
    33.7 KB · Views: 198
  • Virtual_Point_Polar.png
    Virtual_Point_Polar.png
    240 KB · Views: 195
  • Virtual_Sphere_Polar.png
    Virtual_Sphere_Polar.png
    218 KB · Views: 196
Last edited:
Steve,

You always keep amazing me how much time you take to explain things and how many different tools you use to support your explanations.
Thank you for that.
It's obvious that you are years ahead of me on this subject of acoustics.

But I still have this little voice in the back of my head that tries to tell me that I'm not yet convinced and both concepts are far from flawless.
Before shooting from the hip, I need some more time to react properly.

Hans
 
Fascinating stuff, all; esp the total sim showing the effect of the input 1R5//220uF network. I'd thought that would be about the effect, but for totally the wrong reasons.

In the ESL-63, Walker utilized lossy delay line as well as segment 7 to smooth out the ripples and polar response trend. Never ceases to amaze me what he was able to accomplish without the aid of modern computer simulation.
Oh, Walker had a really quite incredible computer: his name was Peter Baxandall.
 
From what I read the reason for using the pulsating sphere was that in emulating a point source as the microphone was in effect a point receptor, the aim being to reproduce what was seen by the microphone, I am not sure that an oscillating sphere does this.
I could be wrong, but what you are describing sounds more like it might be related to the reciprocity argument Walker used in his AES paper to derive what is often called “The Walker Equation”. It states simply that the far-field pressure response of a dipole ESL is directly proportional to the summation of currents thru all the ESL elements. A point source greatly simplifies the derivation, but the resulting equation holds for any distribution of ESL currents; completely un-segmented, emulating a pulsating sphere, or emulating an oscillating sphere. I use it all the time when designing segmented line source ESLs that use resistive ladder networks rather than inductors.