We have for some time (quite long) been working on a method
to repair the electrostatic elements used in the Quad ESL-63, and now we have
managed to solve all problems involved. The result is a new page with a few
products for sale and a repair service.
MT Audio Design ESL Repair Shop (closed!)
Below is a link to an article about the polyester ageing process written by Gary Jacobson. This should be interesting to those who are interested in element repair.
Ageing Quad Panels - How Old is OLD?
A link to Stewart Penketh's pages are found at the end of this page.
A complete ESL-63 repair manual is available at Geert Meddens Audio Pages.
To provide information about element repair and to provide material and tools. We have decided to be open about our findings to make it possible for almost anyone to make a successful repair. For those not interested in DIY we will also repair elements.
What is Needed
There are essentially four problems to solve:
If the information I have is correct the Mylar film (Mylar is DuPont's registered trademark for polyester film) used in the Quad ESL-63 is 3µm, very thin indeed and definitely thinner than the films used in most other electrostatic loudspeakers. I have seen different figures on the thickness, ranging from 2µm to 6µm. The used material in the films is polyester and I have bought pure polyester film, and this is labeled Hostaphan. I have bought it in 3 different thicknesses, 3µm, 4.8µm and 12µm. All the films are 65cm wide and can be bought from the 'MT Audio Design - ESL Repair Shop' page found above. Polyester is used because it is very strong and can withstand the stretching that is needed and stay that way for years.
The stretching process is not a major problem and we have built a stretching jig that makes the process repeatable. The tension must be high enough to keep the film in correct position and stay stretched and with a thin film like 3um the tension needs to be pretty high. When the polarizing voltage is applied the film tries to get closer to one of the stators (on ground potential) and the film shall be stretched enough to minimise this phenomena. We made an element with too low tension (unexpected error due to change in tension measurement) and this element played without problem when we put it into the speaker. The next day I found the speaker making a terrible buzzing sound and the neon lamp was blinking continuously, we disconnected the speaker and put it into the workbench for investigation. When we connected it again it was silent! We knew that the tension was a bit low in one element and tried blowing on it and when doing this the diaphragm moved close enough to one stator to make that noise again and with the light out there was apparent sparking. Nothing much to do except pulling the element out for inspection, we did this and there was no visible damage on the film except for some marks in the coating. From this we can learn that the tension is important, not so much for the sound as for the reliability of the element. It is difficult to describe how much tension is needed. The easiest way to learn this is to feel the tension in an original element and when you make your own elements, the tension shall not be much lower than in the original elements to play safely. To get this tension, you need to build a stretch jig. This is not too complicated if you use a wooden frame with bars applied on the inside, use bolts and wing nuts for adjustment of the tension. You must also make some kind of arrangement to fasten the film to the bars but I am sure you can figure out a method.
Cutting the Film
The polyester film is very difficult to cut and the best method is to use a hot soldering iron or a very sharp razor blade.
Making the Holes
Another problem is to find a suitable method for hole punching. A hot soldering iron melts the film nicely, you don't even have to touch the film. The holes created with this method can be made circular and without edges and other damage to the film. I have tried using a hole pipe (punch), and even if the method works, it is difficult not to damage the surrounding film.
It is also possible not to make the holes and attach the film to the spacer bolt but the stress on the film will be higher than on any other point of the diaphragm and eventually it will rip. Some kind of influence on the sound is also probable.
Glueing the Film
The glue used by Quad seems to be the source to some of the problems with ESL-63 elements, especially when the metal grilles are not used. If the elements are exposed to sunlight the resulting heat seems to make the glue loose its grip and the film can come too close to the stators, 'poop' sounds or buzzing appear and eventually the film with be destroyed (this can happen even with quite new elements). We have found a suitable glue and today there are many more options than in the early 1980's when the ESL-63 was constructed. Some search at the Internet indicate that the recommended glue types are epoxy and polyurethane based. We use a polyurethane based glue called 'U-Fix Contact' which is heavily diluted with acetone to make a thin and strong contact.
Finding a good coating is one of the most difficult problem to solve and the ESL-63 elements have a surface resistivity of 10 Mohm/squarefoot (I have read this). We have measured the surface resistivity on quite a few elements with an electrometer and a special made probe, the results have been between 2x ohms/square and the coating is mostly very evenly distributed. The electrometer measures the resistivity with low voltage and very low currents, measurements made with higher voltage would give other readings. Our aim with the measurements has not been to get an exact figure on the resistivity, more important is to find a coating that gives similar readings. We have tried some different coatings and the most commonly used types by those who make DIY ESLs are graphite and liquid soap. Graphite can be used but it is very difficult and messy. The resistivity curve looks quite a lot like a diode curve where there is almost no conductivity as long as the graphite particles don't touch each other and when the particles come close the conductivity rise very quickly and it is hard to control the resistivity. Liquid soap gives very good readings but there is a severe problem with this, the soap is designed to be degenerated by UV light for environmental reasons. We hoped that the nylon coating used in the ESL-57 should be perfect for use with ESL-63 elements but it turned out not to be. For some reason the diaphragm is not properly charged and the result is low efficiency and thin sound. If the charge is not distributed equally throughout the entire diaphragm area you don't get an equal force over the whole element and this cause the drop in efficiency. On the other hand, if the resistivity is too low you get charge escape (described further down on this page). Finally we found the coating, we call it 'Polycoat' and it does what is required.
If you want to be sure that the coating is distributed evenly throughout the element you need to measure the resistivity. Unfortunately a common multimeter is not able to measure resistance as high as that of the coating used on ESL diaphragms and you either need to get a measuring device that can make it or just hope that the coating is applied properly. Normally the resistivity on ESL diaphragms is measured as surface resistivity. The surface resistivity is a bit difficult to understand but the method is described further down on this page. The reason for using this method is because when you measure between two points on a surface you get a lower resistance than when you measure on a strip at the same distance (the strip resistance is in parallel with several other slightly longer strips), and the measurements will depend on the shape of the surface. The high resistivity membrane is in fact one of the most important factors behind the success of the ESL-57. With high resistivity the membrane operates in 'constant charge mode' and this is what makes the ESL linear. The ESL-57 was as far as I know the first electrostatic loudspeaker operating in 'constant charge mode' and it is very important for a full range ESL to do this. High resistivity coating makes this possible since it prevents the charges on the membrane to move when the membrane moves closer to the stators and different parts of the film are at different distance to the plates. The resistivity shall be chosen so that the resistivity is high enough to prevent 'charge escape' at low frequencies and high levels, but low enough to make it charged in a reasonable time. We were aiming at higher resistivity than on the original ESL-63 membranes to get better low frequency response and maybe also higher possible SPL, this turned out to be wrong. The chosen resistivity on the ESL-63 elements is perfect, it does not take long for the elements to be properly charged and even with slightly lower resistivity there is no change in sound.
Why is High Resisitivty Important?
When a voltage is applied on the stators, negative on one and positive on the other, an electrostatic field is created between the stators and this causes the positive charged membrane to move towards the negative charged plate. With alterating current (AC) the electric field will switch direction and the membrane will follow this.
With some help from the picture above I will try to explain the difference between 'constant charge' (left picture) and 'constant voltage' (right picture) diaphragm behaviour. With +/- E on the stators as in our example the positive charged membrane moves towards the negative stator (red) because of the electric field that is applied between the plates and when the stator voltages are inverted the membrane also moves towards the negative stator but this time in the other direction (blue). This is the principle behind electrostatic loudspeakers. The difference between a membrane with constant charge (called CQ) and constant voltage (called CV) is that the resistivity on it is either very high (CQ) or very low (CV). The positive charge (+Q) in the example above always tries to get closer to negative voltage and if it is free to move on the diaphragm (low resistivity) it will do so because it is easier than jumping through the air gap, this will cause an uneven distribution of charges in 'constant voltage' operation and the operation becomes non-linear. The 'moving charge' phenomena occurs because the membrane is slightly curved and this is increased with higher amplitudes, and makes it very important to have a high resistivity membrane especially for full range electrostatic speakers.
Note that with constant charge the force F does not depend upon the membrane-plate distance at all!
The needed resistivity can be calculated from the ESL element capacitance and the resulting time constant.
The dielectric constant for air is K = 8.85 x 10^-12.
A Quad ESL-63 panel has an area of A=0.099 m² and the distance between the panel and the diaphragm is d=3 mm, with these figures the resulting capacitance C1,2=292 pF. The two plates are in parallel and the total capacitance is then Ctot=584pF.
Time constant = T = RC -> R=T/C
T = 1/10
The resulting resistance = R = 0.1 / 584 x 10^-12 = 1.7 x 10^8 ohm.
To be on the safe side, the time constant shall be 10 times larger than the period time for the lowest used frequency. In my example I have used 10Hz which is pretty low, but even if the loudspeaker does not produce sound at 10Hz it may well be exposed to it. With the 10 times safety margin and f low = 10Hz we need a resistance of 1.7 x 10^9 Ohm.
ESL-63 Element Construction Details
During our research for this repair project we have found some interesting facts about the ESL-63 elements and some construction details we were aware of before. Below follows a mix of newfound facts and old knowledge, but it is all interesting facts about this eminent design.
The element uses two different stators, the front stator (to the left above) holds the Mylar film, and on the rear stator damping material has been applied to make an acoustic resistance. The rear stator also contains the metal contact strips (on the long sides) used for charging the diaphragms from the 5.25kV high voltage supply.
The picture above shows the resistive coating used by Quad,
note the uncoated areas in the corners. The surface resistivity is 10 Mohm/squarefoot
and the uncoated corners are essential because of the fastening screws in the
corners that are in contact with the grounded loudspeaker frame, if coating
was applied on the whole element, the charge on the diaphragm would be drained
by the grounded screws. 2 Mohm/cm and 5.25kV gives a resulting current of 2.6mA
at 1cm and it is not possible to keep the elements charged if this amount of
current is drawn. The hole in the middle is for the screws that hold the stators
in place and tight to each other, the film has holes around these.
The picture above shows the plastic stator frame that holds the stator (printed circuit board) in place and absolutely flat, the printed circuit boards are attached to the stator frame with glue. Some Quad owners (especially those who use Arcici stands) have reported problems with stator plates loosening from the frame, this is a serious problem and we will try to find a method to re-glue them in a good manner. If you use Arcici stands and have removed the metal grilles it is easy to bend the whole loudspeaker frame and this can harm the elements, be very careful when you move the speakers with Arcici stands. There is no reason to avoid Arcici stands as long as you are extremely careful.
Common Faults Causing Need for Element Repair
Patrice Hassen in France, an experienced Quad ESL-63 service man contacted me via email and sent these very informative pictures.
In my point of view, the main reason of defective element is in attached files. Take a look and give me your comment.
I do service on QUAD ESL63 since 20 years, so if I can help...
Take a look at the pictures (and the 'floating texts' that appear when you point at them), Patrice's comments appear further down on this page.
The reason for these pictures is to help the customers who contact me to understand what can happen inside their favorite speakers without taking the risk to touch inside the speakers. The main causes of the problems (see my pictures) are:
A common fault in ESL-63s are damaged or loose dustcovers. Normally faulty dustcovers make rattling sounds at certain bass notes.
From Quad's own dustcover repair instructions:
Damaged dustcovers on the Quad ESL-63 should be replaced immediately to prevent the ingress of dust. Because the dustcover is made from 2.5um My1ar it is extremely fragile and it is not feasible to pack and ship complete dustcover assemblies.
Repair kits are available from Quad, but if the prices are too high I can
sell 3um film suitable for dustcover repair. The width is the same as Quad's
original dustcover film (650mm) and I normally deliver in 10m rolls, enough
for more than two pairs of ESL-63s. I will try to find a suitable double sided
Measuring Surface Resistivity
Information copied (and altered) from Delta Technology's homesite.
A thin film's electrical property is generally characterized as surface resistance, Rs, which is the ratio of the dc voltage applied to two electrodes on the surface of the sample to the current between them. Since the effect of volume resistance is negligible, it is typically ignored in films as thin as these. The surface conductivity is the reciprocal (inverse) of the surface resistivity. Surface resistivity, Ps, is the surface resistance, Rs, multiplied by the ratio of specimen surface dimensions (width of two parallel electrodes defining the current path divided by the distance between them) which converts the measured resistance into a value that would be obtained if the electrodes formed two opposing sides of a square. While specific resistance is quantified in ohms, it is frequently referred to in 'ohms/square'. In this context, the size of the square is immaterial and without dimension.
The formula used:
Ps = Rs x W / L
P: the surface resisitivty (P represents the greek letter, rho)
Rs: the indicated resistance
W: the length of each of the electrodes
L: the distance between the electrodes
Current will flow from one electrode to the other. Assuming Rs for the example (first figure) is 80 ohms, and that L = 2W, we would calculate that:
Ps = Rs x W / L = 80 ohms x W / 2 W = 80 ohms / 2 = 40 ohms.
In comparison, measuring in the other direction (second figure), our ohmmeter should indicate Rs = 20 . Since we know that Ps is a ratio of the electrode length to the distance between (and that the ratio for our example is now W = 2 L ), we would correctly calculate the surface resistivity:
Ps = Rs x W / L = 20 ohms x 2 L / L = 20 ohms x 2 = 40 ohms
(Note that the dimensions for L and W cancel, regardless of the size of the sample or units of measure, hence the dimensionless "ohms per square".)
Parts for ESL Repair
MT Audio Design ESL Repair Shop
Coating for ESL-57:
Available from 'The Quad ESL', check Gary's parts page and look for 'Area 51'.
Gary Jacobson's Parts page
HiFi World supplement 52-2 (Quad ESL care and cleaning, by Mike Gerrard)
From Stewart Penketh in Canada I received an email with a question if I would be wiiling to publish some information about his Quad ESL repair service. I agreed and I put it on his own page Stewart Penketh Repair Service.
This site uses frames, but if you have found this page via a search engine and do not see a frame on the left side of this page, you can use the home button to get to the main frame of the MT Audio Design site.