Renovating the EHT power supplies in a pair of Quad ESL57 loudspeakers.
The Quad Electrostatic Loudspeaker (ESL) is a classic British innovation that was launched back in 1957 to widespread critical acclaim. Creator and driving force behind Quad, Peter Walker, had managed to produce a loudspeaker that sounded — and still sounds — like no other, was as close to perfect as anyone had achieved and for many went on to serve as a golden reference.
However, as with a classic car, the ESL needs care and attention from time to time and is nowhere near as simple as a moving coil loudspeaker. There are high voltage power supplies and all the fun that comes with circuits carrying 6kV, plus thin membranes that are easily damaged.
This post takes a look at examining a fault whereby one loudspeaker was producing barely any output, tracing this to a PSU issue, and then rebuilding both power supplies.
Warning: do not attempt this type of repair unless you are qualified to and comfortable working with high voltages! This is not a how-to guide or intended as a comprehensive set of instructions, but rather instead a simple summary of the steps taken.
Theory of operation
Image Source: Wikipedia
Whereas moving coil loudspeakers generate sound using a cone driven by a varying magnetic field, electrostatic loudspeakers achieve this via a membrane in a varying electrostatic field. The result is a typically much larger area generating sound, which at the same time has far lower inertia due to being only a few microns in thickness. In addition to having a fast transient response, the design is said to act like a dipole antenna and couple to the room much better than a loudspeaker cone.
The original Quad design — later dubbed the ESL57 when a newer version came out named the ESL63 — is comprised of a treble panel flanked by two bass panels, with the membranes being constructed from 6 and 12 micron tensilised film respectively. The membrane is charged via an EHT power supply, which supplies 1.5kV to the treble panel and 6kV to the bass panels.
The moving membrane is sandwiched between two stators and these are connected to the output of a step-up transformer, with the signal from an amplifier being attached to the input. The rapidly varying high voltage causing the membrane to be attracted and repelled from the stators.
In principle, the ESL is quite simple, but the reality is very different and all sorts of challenges come about due to, for example, the panels essentially being a capacitor and this combined with step-up transformer inductance and other variables, meaning that the whole thing starts to act like a filter. A sub-optimal design may at best exhibit poor frequency response, or at worst panels may arc and burn holes in the membrane, or worse still present a fatally damaging load to the amplifier.
So while the electrostatic loudspeaker had been around for some time, it wasn’t until Quad invested a significant amount into R&D that led to a product which was both reliable and sounded great.
Inspecting the speakers
Evidence of corona discharge.
Even with an excellent design such as the ESL57 it is still possible to burn holes in the bass panel membrane due to arcing from over-driving the loudspeakers. The panels also have extremely thin dust covers — they have to be so as to not colour/impede the sound — stretched tight over the frame in front and behind the stator and driven membrane, which are fragile and easily damaged.
Inspecting the panels it appeared that, thankfully, none had suffered damage from arcing.
Upon discovering that a “clamp board” had been retrofitted, which in the case of over-driving clamps (shorts) the input signal to protect the panels, this was perhaps not that surprising.
There was some evidence of white marking from what is assumed to be corona discharge, but this should hopefully not prove too much of an issue.
The problem at hand
Next, the mains transformer and EHT multiplier assemblies were removed. As can be typical with high voltage equipment, the multiplier PCBs were fairly grubby on account of attracting dust.
The secondary on each mains transformer was putting out over 600 volts and so these measured up fine. However, using a DMM and high voltage probe to measure the output at the bass terminal of the multiplier circuits, one registered a little over 3kV and with the other closer to 1kV.
At this point, it should be noted that the probe impedance may not have been quite high enough for such a low current power supply and electrostatic voltmeters seem to be favoured for the task. However, I had it on good authority that I should be measuring at least circa 3.5kV, so could be pretty certain that one if not both of the power supplies had degraded.
Old for new
Having lugged both loudspeakers from home and into the workshop, then removing the many screws affixing the rear covers and de-soldered the PSUs, it was decided to replace all the capacitors and diodes in the multiplier circuits for both speakers, just to be on the safe side.
The leads were snipped and one-by-one the old caps and diodes de-soldered. These were replaced with new 3kV rated diodesand 10nf 3kV rated ceramic capacitors .
You can also see in the above image an old circular Bulgin mains connector, which is almost certainly not up to today’s safety standards, in addition to which it is now near impossible to source mating plugs and old ones are constructed from Bakelite or similar and have a habit of cracking.
It’s a tough decision when you come across something like this with classic equipment, whether to modify or leave original. In this case, it was decided that the pros outweighed the cons and replacement modern IEC — also Bulgin! — connectorswere ordered. These had the benefit of mounting holes in close enough to the same position to work.
However, the main circular hole needed filing out very slightly in order to accommodate the new rectangular panel mount socket. In an effort to ensure as neat a job as possible a simple template was laser cut from MDF and secured to the metalwork during filing.
Prior to reassembling the EHT PCBs were given a coating to help protect against arcing and corona discharge. Small pieces of heatshrink were first placed over the terminals to prevent these from being coated also and causing problems when it came to soldering cables to them.
Once reassembled the PSUs were tested once again and the voltage measured on the bass terminal this time registered at just under 5.5kV. This was also just a minute or so after switch-on and it’s possible that if left on for longer, it may rise a little. In any case, a significant improvement!
Next, the PSUs were fitted back into the speaker cabinets and once fully reassembled testing was carried out using a little amplifier PCB that was close-to-hand. Even with just a couple of watts drive they both generated a respectable sound level and I have a feeling that once hooked up to the Quad II valves amplifiers, they’re going to sound even better than when I got them some years ago.
At this point, I’d like to thank One Thing Audio, who specialise in Quad ESL repair and upon discovering their label inside the cabinets, were happy to look up the service history of the ESLs. Turns out they had been refurbished in 2003, which included rebuilding the treble panels, as evidenced by the fact the many rivets had been replaced by M3 nuts and bolts plus threadlock.
I was advised that the treble panels should be good for a little while longer, but the bass panels are original and the partially conducting coating on their membranes could well be degrading. So I’ll listen out for a drop in bass output and if this occurs, options include attempting a rebuild myself — various vendors sell membrane material and coatings — or having them rebuild them.
As noted at the start of this post you should not attempt this sort of repair unless you are qualified at working with high voltages; mains, 6kV output from the PSUs and the high voltage audio output from the step-up transformer all present significant risk!