Nylon .vs. Graphite

I have so many questions by e-mail about coatings, and have had for some years, so I thought I would write this section of the site on a set of comparative trials that we have carried out using the original Nylon coating (CALATON CB), and the “usual” DIY favourite coating, powdered graphite.

Immediately, there will be those who (metaphorically) jump to their feet shouting - “He's (meaning me) biassed”, or “I've used graphite, and it works fine!” OK, that's why we do blinded trials and I don't take part; because I can't avoid some bias. Nor can others at times! Secondly, graphite does work “fine”, so read on, and see what's up?



Four treble panels were re-diaphragmmed. The choice of treble panels was made because it was thought by all the enthusiasts involved that if we were going to hear subtle differences (we thought) in an original Quad then we would hear them in the mids and highs and not in the bass. Perhaps there would be value in doing a comparison in the bass frequencies but we have not as yet done one.

Two panels were built using 6 micron tensilised P.E.T. coated with CALATON CB. Two other panels were built using 6 micron Hostaphan and graphite. All panels were coated on both sides. The graphite was rubbed off the film using isopropyl alcohol to obtain a surface resistivity between 105W and 107W per square. This is about as high as can be achieved with graphite coatings. Anything under 10 6 in a Quad will make the panel hummmm due to excessive current drainage from the EHT block, so that is not a value to aim for. The CALATON stock solution (10% by weight) was diluted 50/50 with pure methyl alcohol to give a final surface resistivity of of 1012W to 10 14W per square. The upper limit stated is problematical because of the limits of the instrumentation used, but is within the range of surface resistivities quoted for Nylon(s) generally. We were able to measure surface resisitivities up to 1013W per square reliably and reproducibly using a standardised SRM-100 meter.


Subjective Listening Tests

The graphite panels were installed in my resident stacked Quads, lower bunk, one each side. They were left as the “resident” system panels for a period of 3 months. Approximately 10 casual and 4 “serious” listeners, other than myself, heard these in use over this time. All listeners commented on the sharpness and general clarity of the sound, especially casual listeners. The "serious" listeners bring their own source material (CD & vinyl), but we always re-use the same source materials at consequent sessions.

The “serious” listeners described the sound as powerful, clear and forward. The Quads just being their usual stupendous selves, and so on. I was a bit relieved in one way. Grinding that bloody graphite into a 6 micron plastic film without tearing it is a real effort. In short, this was a good result for the graphite coated panels and they did perform far better than I thought they would. A good thing thing that I “imported” a few technically disinterested listeners then!

After the initial 3 month period, the “lower bank” graphite panels on each side were replaced by the Nylon coated (CALATON) 6 micron panels. “Serious” listeners were invited back to listen about a week after the change over. Things were nicely settled in and well charged by then, of course. What did they say? Lots of little things, which add up to one big thing, really - Nylon is better - by far.

My "serious" listeners with their own consistent source material all came out with similar (but distinctive) comments ... and these people were not together for these tests. Some comments...

· “”””””I can hear the 'air' around everything so much more clearly...”
· “”...micro dynamics are just perfect...”
· “”.. and I thought the other panels were clear and powerful, but they were comparatively) hard, brittle and lacking in subtlety.”
· “”Oh yeah, you can see the shape of Janet Seidel's lips....”(?!?!?) [OK]
· ”...harmonics of the violin are correct, that's all. Harp too.”
· ”...never heard the intonations of that little kid's voice on track three of Blue Moon Swamp (John Fogerty) before. Now I can hear it, and what he (obviously a 'he') is saying more clearly.”
· “”Slightly ‘warmer’ presentation I think, but not over-etched like the previous panels. These are ‘delicate’, is the word I would use. Clear, precise, natural.”
· ”The focus area is wider” [3 comments on this]

So on, and so on. The microdynamics of the nylon coated panel in an original Quad is superior to graphite by a fairly noticeable margin. If you listened to graphite in isolation you could easily live with it, let's be clear about that. The difference is like that between a transistor amplifier driving a Quad and a good Valve amplifier driving a Quad. The harmonic distortion is lower, apparently well below audibility at any listening level used (90 to 102 dB). The nylon retains its "subtlety" down to low listening levels also, far better than the graphite coated diaphragms.

The tests were “single blind” and the “serious” listeners didn't know which panels were which in the system. It was a matter of - “come on over and listen to this” - “what do you think?”

OK, you knew it all along, right? Well, maybe you did, and maybe I did too. Whether I ever sell another gramme of soluble nylon from this site or not (more for me); or whether you source your own somewhere else, my earnest advice is - get it!! The way in which nylon arrests the charge within its structure and prevents it moving around on the surface of the diaphragm would seem to have unique implications and benefits in this speaker, at least. The most obvious and talked about advantage is the high resistance that ensures dynamic stability, but the most ignored effect is the dramatic reduction in harmonic distortions. We would, from rough measurements (having no anechoic chamber handy), put the distortions with nylon below 0.1% at all frequencies above 500Hz and a maximum of 0.3% below that. This is absolutely astounding. There is a measurable rise in distortion with graphite (~2 - ~2.5% which is very, very good in a speaker), but that is not all we are hearing I am sure!


Theory and Causation?

If we return to Frederick V. Hunt's original monograph which has no reference to actual loudspeakers, we can find some evidence for these subjective observations as noted above. Consider the following diagram, taken straight from Electroacoustics, F.V. Hunt, p. 188, fig. 6.8,


[Carlo V. Bocciarelli of the Philco Corporation suggested this circuit to Hunt and argued qualitatively the advantages of “constant-charge” operation. The analysis given by Hunt showed the performance to be better than expected - as outlined below. The circuit itself, though not its unusual virtues, was proposed in various patents and other publications dating back at least to H. Riegger’s disclosure in German Patent No. 398, 195 filed 10 March, 1920 and issued July 1924]

La, Ra, Lb, Lo and Rb are included for generality. There is probably not much actual inductance (Lo ) in the diaphragm coating! The resistance Ro in the common leg of the circuit serves as the high resistance usually called "protective". The impeadance in this branch is clearly common to the two circuits.

We know, as it is the most widely quoted result of Hunt's work (Walker, Williamson, Sanders, et alia), that if the time constants RoCa and RoCb are large enough then this is "constant charge" operation and the force on the diaphragm will be independent of position in the inter-electrode (stator) space. This yields the extremely well known theoretical (and practical!) linearity of the push-pull electrostatic loudspeaker. This relationship is commonly quoted as follows:



f >> 1/ (2RoCa,b)


using the notations of the diagram above. All very well, and good, nicely done, and so on. There are those who think a large charging resistance (as Ro is called) will be “fine” in series with a membrane coating with even a few hundred kW of resistance, or so. Well, they do work, as demonstrated by the graphite panels. They work pretty well too. However, they are farfrom what is possible, and even further from what was achieved in the original Quad ESL. Why have so few designers and constructors of the modern era not caught on to this? Surely, they have read Hunt's work, and Williamson, and Walker, and Kellogg, Lee, Rice and a dozen others on the subject? What seems to be the problem?

I believe, for one thing, they see the superb linearity achievable and think - good, here's a neat, cheap(!) easy way to make the speaker perform linearly - a 50 cent resistor in the 22 - 44 MW range and we're awaaaaay - Hey! Hey! They interpret the circuit (diagram 6.8, p. 188, Hunt) literally and “mechanically” as a schematic and see no further. If they did read further and look a bit more closely at Hunt's very, very comprehensive analysis, they would begin to realise why the original Quad ESL can do what it does, and why their speakers don't quite ever get to the same standard of excellence - good as they may be.

The modern “mistake” made by nearly all manufacturers and re-furbishers alike (in the main) is to see only this fabulously linear driving mechanism and such a simple, cheap way to achieve it. Either that, or the treble panel hums because of excessive current drain on the EHT supply and this is a way to quickly shut it up. You can see them drooling out there - free drinks wouldn't move quicker!

Treating the resistance (R o) and the capacitance(s) Ca, Cb as lumped (integrated) quantities, as if they were discrete components from the electronics shop is the error they almost all make. Hunt did not assume this, or ignore it. Neither did Williamson and Walker. Hunt went on to produce an elegant analysis (within the limitations of the time - no cheap machine computation was available), of the harmonic distortions in such a speaker, and these are significantly affected by the very physical nature of that resistance (Ro). To achieve the ‘ultimate’ result Ro must be very high (yes, to get it linear and reduce arcing - not prevent it) and it must be distributed on the diaphragm. {Just as an aside - High value charging resistors do not protect against arcing. There is already enough charge in the device to blast a hole in the diaphragm. What they do is prevent the arc from continuing at high current.}

Now, let's get down to the nitty-gritty and look in detail at those two currents flowing in the membrane coating Ia and Ib . “What?!?”, I hear you say. This is static electricity you idiot, what is this with currents? Relax, there will always be minute currents. This is because the circuit leaks charge in numerous places and the EHT makes up that charge - OK? These currents are there, however miniscule, and they do affect the sound of any speaker made in this fashion. Especially once you start that diaphragm moving about and creating an attractive (or repulsive) force on those charges. They will not stay still, there is some current flow, and you can't beat it - but you can come close!

In addition to improving the linearity of the speaker, second harmonic and other (in this case) dependent harmonics are reduced substantially by this resistance. From Hunt (p.209) we have the third harmonic velocity ratio expressed as:


v3/v1 = Ia2 x 1/2Eo x z1/z 3 x Za[1 + 2(Ta2 / Zaz1) / ( 1 - 2/3(Ta2 / Zaz3)

where z1 and z3 represent the total mechanical impeadances, including the radiation loading at the frequencies of the 3rd (v3 ) and fundamental (v1) harmonics.

We see here a linear dependence of the 3rd harmonic velocity ratio on the second harmonic current. Anything that can reduce this second harmonic current not only reduces second harmonic distortions, but third and higher harmonics as well. Therefore, the designer increases Ro to any desired extent to reduce distortion - you just can't 'lump' that resistance all into one component, that's all. Hunt describes this on page 210 as follows:

Ia2 = E1/ 2Za[1 - 2(Ta2 /Zaz1)]2 x E1/E2 x 1/( 1 + j t) x Ta2/Zaz1

where t = 4wRoCa. This is more than 10 times the time constant ratio which is always focussed upon. It means that Ro has more than 10 times the effect in reducing harmonic distortions (of all kinds!) than it does in ensuring dynamic stability and linearity in the device. Surely, this is even better news for the “use a high value resistor and a low resistance coating brigade”? Not really, but it does explain why even rather haphazard execution in an electrostatic speaker design (or a repair) by DIYers can stillsound remarkably good!

So, the second harmonic current and its dependants are 1/(1 + j4wRoC a) or about 4p times less sensitive to variations in the resistance as are the stability and linearity factors. This resistance is required at all points on the vibrating membrane, if possible, or the speaker will have lower Ro at some points on the membrane. The upshot of which is - various higher and lower Ia2 and Ib2 can flow resulting in subtle variations in the distortion performance across the membrane.Furthermore, if each side of the membrane does not have a similar resistance (coated both sides) then there will be an imbalance between Ia and Ib on each side and this will contribute to any non-linearities in the speaker, but it will also affect the distortion performance, point-to-point. This is not unlike the distortion caused by cone-break up (variation in mechanical resistance) that we witness in moving coil speakers, although I hasten to add - at a much, much lower level here, even at its worst. Fine you say, no harmonic crises, and good linearity for average resistance - let's go home. Hang on though, it's OK when the resistance varies at a given point by (say) an order of magnitude upwards, but when it varies downwards the distortion levels increase! There is no way in the original Quad that any coating could achieve such a constancy by purpose-designing a coating material (not in the 1950's anyway), so they went for the highest value of resistance they could get in a practical substance (nylon) which had good distortion reducing performance even at its lowest values.

Back to Nylon coatings then. The nylon coating produces such a high resistance, even at its lowest ebb (10 12W/sq) that distortions are kept to a very low level even when the coating is at its thickest and least resistant. The distortions are vanishingly low at the highest resistance spots in the nylon coating of course. Even when (still) bouncing up and down by a factor of about 12.56 these distortions cannot be heard at most frequencies. In fact I defy you to hear them at any frequency! Nylon has the added advantage (over graphite) that it confines electrons to the surface of the polyester (Mylar) diaphragm which tends to improve the uniformity of the charge distribution in spite of a less than “uniform” application to the surface of the membrane. This is a triboelectric effect known to Hunt, Williamson, Walker, et al., and was once considered trivial and basic electric theory. It hasn't changed or gone away. People have forgotten all about it!

If all of this still seems to be a bit of a mystery, I can assure you that it is. Charge transport and trapping in polymeric dielectrics (insulators like nylon) is still being investigated quite vigorously to this day. The This email address is being protected from spambots. You need JavaScript enabled to view it. at Monash University is developing experimental and computational techniques that will yield detailed information on the spatial and activation energy distribution of trapped space charge in thick and thin(!) dielectrics. These investigations are particularly relevant to power cable insulators (high voltage!), but then, pure science is such an unruly creature, isn't it. It can serve different people in different and highly useful, unexpected ways!