Quad ESL Cap mods

Note: Working on Quad speakers is inherently very dangerous due to their high voltages, although you can do these mods at your own risk (and could result in death), I recommend that you send your speakers to someone that has experience in working with these speakers. This way you get to enjoy the music.

It took me a awhile to figure out a capacitor configuration to replace the stock 220uF 50V electrolytic, and the low voltage signal resistors. I have used Mundorf caps, and have been very pleased with their price/performance ratio. The MCap dimensions given at Mundorf's web site are incorrect as they listed a length of 66 mm for their 56uF-100uF caps, the 100uF are 72 mm. The stack height is an issue (66 mm max length) and the diameters are limiting factors of what will fit in the Quad base. Most other high end caps just won't fit and I am not sure of the benefit with such a large capacitor value. The MCap 100uF would not fit, unless you use the 250V version, per Mundorf, they say their 400V version sounds better then the 250V. I finally settled on the MCap (400V) 3 x 68uF + 15uF and a MCap Silver Oil 2.7uF, this configuration just fits stacked vertically in front of the step up transformer on the side opposite the power transformer. This also keeps all the signal wiring away from the AC. I used some 4 mm acrylic sheet, drilled/milled and glued the capacitors on one side, I hardwired everything with Cardas wire and it bolts to two of the studs that hold the upper electrostatic panel. Thin circuit boards may have been easier. With the MCap Silver Oil cap., one could go up to 3.9uF given it is the same dimension as the 2.2 or 2.7, I just went for the one that was available. MCap's have very tight tolerances, so both banks of caps measured at 221uF, they had unmeasurable ESR & DA. I replaced the stock 3.3 wire wound resistors with Caddock MP821 3.3 ohm resistors on the main board, they fit perfectly in the stock holes when you bolt two to the same heat sink (back to back and I needed to offset the heat sink I used). They are rated at 20W each as opposed to the stock 7W. I attached them first to the modified heat sink and then mounted them to the circuit board. I used a Duelund 1.5 ohm 10W resistor at the input, seems to smooth out the sound and costs only slightly more than the Caddock. I replaced the stock 1.5uF cap 250V cap. with a 400v 1.5uF Mcap bypassed with a ERO1837 10nF. Removed the stock wiring and pulled the PTC (FS2) . I left the stock electrolytic/1.5 ohm resister on the board (no really reason to pull them with the PTC/input wiring removed). Replaced the brass binding posts with some decent Cardas CCBP S posts. I wired the Duelund resistor (from the + binding post) and capacitor bank directly to the IN1B terminal (they are long). The negative binding post was connected directly to the OUT1 terminal (moved the bottom plate ground lead to the chassis ground circuit). This made for very short direct signal runs, O resistance in the runs and was easier to wire.

In the stock configuration, I did find some grounding issues (increased resistance between the chassis/HV transformer grounds and the circuit board ground/- terminal), this was because ground connections where screwed together. So I soldered the black step up transformer ground taps to the ground tab screwed in at the power transformer and continued the same wire and soldered it directly to the board OUT2 ground plane with 14g silver Teflon wire. The resistance in the ground circuit dropped to 0 ohms. Before this the step up transformer ground taps were grounded mechanically at the power transformer and the screw was not very tight. As mentioned the ground for the metal base cover was soldered to the ground tab that connects to the step up transformer near the new capacitor bank.

I lined the complete base cavity and cover plate with thin sound deading material, this significantly reduced the cavity resonance. I changed out all the 8 mounting washers that bolt the electrostatic speaker panel to the base with larger/thicker steel washers, and added lock washers to the bolts that did not have a ground tabs attached. This made a significant difference in the speaker flex between the base and the panel. Ideally I should have probably used washers with an even larger area or a small fender washer. The stock washers are no larger than the nut and were dug into the plastic (i.e. loose, especially those without a lock washer).

Fired it all up last night and frankly my wife and I were floored at the difference. I have had the 988's for several years, initially listened to them infrequently and then they were in storage for several years. I never was really satisfied with the sound, great mids, but lacking at the extremes. I replaced the amps, replaced the preamp, tried different wiring and even tried supper tweeters to no avail. Never could figure out what I was missing until now. Well, with the mods they are completely different speakers. Much better base and articulation, vocals sound like vocals, better air at the upper end and improved balance across the music spectrum. Since they only have a few hours on them, I look forward to them getting better as they break in. I am using Cary 805 anniversary amps that are good for 50W, I also have a Mac275 that I want to try. The changes made a significant difference, but I cannot really say the contribution of each change. Great addition for the 988/989 speakers, not sure of the space in the 2805/2905 Quads but the mods bring these speakers to whole new level.

Mksj:
I believe that the main reason why you are delighted with the sound quality after modding the speakers is the difference in distortion, clarity and speed between crummy electrolytic caps and high quality film caps.
I still can not believe that QUAD had the nerve to use lytic caps in series with the signal...Quad 988s are not exactly budget products.
A cap with unmeasureable ESR and DA is electrically pretty close to a straight piece of wire (within its frequency pass-band).

Also, your ground mods were a smart move. Many designers tend to think of ground as a zero volts, zero current wire, supposedly not important at all.
In practice, because of ground wiring resistance, voltage and current ride on the ground wiring and mess up the signal return path.
Grounding layout, wiring and connections are very important and are very rarely implemented correctly.

ENJOY!

Agree with you, replacement of the Quad ESL speaker 220uF electrolytic capacitor probably accounts for the majority of the improvement. Probably replacing the electrolytic cap/resistor and eliminating the PTC gives the most bang for the buck and the simplest to install. You would not need to pull up the circuit board, just remove the stock wiring and snip out the PTC. I am surprised that even in the latest Quad 2805 and 2905 speakers that the electrolytic is not at a minimum bypassed with a decent film capacitor given their increased price. But then I have had Magnepan speakers through the years, and they are no better in the crossover department. Hardwiring the ground system and using a modified star configuration is easy to do and improves the safety.

Greetings, may I ask if anyone has tried Solen 220uF film caps bypassed with something, or perhaps those Obligato 100uF 500V caps?

I'm running into a problem with my 989s recently and I suspect its protection behavior might be the reason why my power amp fried. Stereophile wrote that the inputs become SHORTED when a 989 goes into protection. Can any fellow member pleaseclarify that?

Also I notice that when the speakers are not connected to an amp, I have one speaker whose terminal measures almost 0 ohms using a DC Ohmmeter and the other starting from 0 and slowly going to infinity as if measuring across a cap.
So which speaker is behaving normally? Thanks.

Hi, Johnsonwu. My Quad ESL 988 speakers measure around 2.8 ohms across the speaker terminals. The Quad ESL 988, 989 and the Quad ESL 2805, 2905 use the same circuit. If your resistance measures 0 and then goes to infinity, it probably is just seeing the capacitor and you have cooked (read open) the resistor that parallels the 220uF input capacitor. In either case they should not measure 0, this would suggest that your other resistor may be shorted. The protection circuitry measures the noise radiation that occurs with the onset of ionization. This triggers the triac clamping circuit to prevent arcing. The circuit operates by limiting the input, and when that fails, by short-circuiting the input. When the triac T1 shorts the amplifier see's just the 1.5 ohm resistor in the input filter. This leads me to believe that at least in one speaker the resistor is now open (resistance goes from 0 to infinity). The R15 input resistor on the Quad schematic is 1.5 ohms 5w, so it probably cannot handle a sustained overload. My understanding is that the combination of 220uF capacitor and 1.5 ohm resistor is a form of input filter, most of the mid and high frequencies pass through the capacitor, as the frequency goes lower it attenuates the signal and more voltage (LF signal) is passing through the resistor. If the resistor is open you would probably have no base. My other concern is that you may have damaged the triac if you are reading 0 ohms in the other speaker (it is not seeing the resistance of the transformers and 3.3 ohm resistors).

As far as capacitors, I have used Solen Caps in the past, but they are just OK. I have used the Mundorf's and also Obbligato films in my speaker crossovers and have been very happy with them. Obbligato's are inexpensive, but do not fit in the Quad. The 220uF Solen cap would cost about $60 and the 3x68uf + 15uf Mundorf MKP is about $100, with either one I would bypass with a high quality cap. So I do not think there is a big savings going with the Solen (not sure it would fit). Mundorf MKP series also has a 250V version of these caps that is a little less expensive (and smaller) than the 400V series, but were talking maybe 10% less cost. I tend to like the Mundorf Silver Foil Oil for a crossover bypasses, as the best bang for the buck without going into the stratosphere (V-Cap or Duelund). I also have heard that for crossovers caps, a smaller bank of equal sized capacitors sound better than one large capacitor. Ideally serial crossover capacitors should be 10uF or smaller, but the 3x68uF is the smallest I could go inside the Quad base. Bypass caps should be 1-2% of the total value. This is important if the capacitor is in series, if it is in parallel (i.e. across the - to +, or shunt) then size or number doesn't matter, use a single cap.

Regarding the resistors in speakers, most of the information I came by recommended the Duelund as the best sounding and then the Caddock MP820 / MP821 series. The stock Quad resistors look like they cost 5 cents. By increasing the wattage of the stock resistors, you would have a larger overload margin. The Duelund 1.5 ohm across the input capacitor is rated at 10 watts (stock is 5 watt and the Duelund can take much higher short term voltage because it is a chunk of carbon not wire), The Caddock resistors are films and are rated at 20 watts with a proper heat sink vs the stock of 7 watts. Some people use Mills resistors, but I think in this application they are a step down in sound reproduction.

I got most of my parts from Part ConneXion when they were running there 20% sale. If you are not adventurous, I would suggest that you consider sending your speakers to Kent McCollum at Electrostatic Solutions. I had spoken to Kent about this mod for the 988/989 series speakers and the 2805/2905. I believe he was going to fabricate an upgrade circuit board/kit with Mundorf capacitors. He has quite a bit of history on the different web forums, is very knowledgeable about Quad speakers and does great work.