Here they wrote that you can stretch mylar with a hair dryer, yes, you can, but only if the area of the panel is small and if the tension is not required too much.
When I suffered from gigantism and made the first electrostatics where the bass panel (two membranes and three stators) was 50-2000 cm in size, then a lot of tension was required, although the membrane was divided by insulators into separate sections of different areas so that the resonances were different, then I had to make a large frame out of wood and a tensioning device with spring balances.
It should be borne in mind that mylar stretches in one direction much more than in the other, for example, it does not stretch lengthwise, but it does stretch in width, accordingly, the tensioner needs to be pulled strongly only lengthwise and this especially applies to when the emitter is semi-cylindrical, it cannot be stretched in width because the film will fall between the insulators and come closer to the stators. So, I stretched such a film of 20 kg along a length of 2000 cm, the film was 12 microns thick. After this, in no case should you heat it with a hair dryer, because the heating will immediately release a lot of tension!!!
When I suffered from gigantism and made the first electrostatics where the bass panel (two membranes and three stators) was 50-2000 cm in size, then a lot of tension was required, although the membrane was divided by insulators into separate sections of different areas so that the resonances were different, then I had to make a large frame out of wood and a tensioning device with spring balances.
It should be borne in mind that mylar stretches in one direction much more than in the other, for example, it does not stretch lengthwise, but it does stretch in width, accordingly, the tensioner needs to be pulled strongly only lengthwise and this especially applies to when the emitter is semi-cylindrical, it cannot be stretched in width because the film will fall between the insulators and come closer to the stators. So, I stretched such a film of 20 kg along a length of 2000 cm, the film was 12 microns thick. After this, in no case should you heat it with a hair dryer, because the heating will immediately release a lot of tension!!!
Not true, MD and TD differ only 20% according to Dupont spec sheet.It should be borne in mind that mylar stretches in one direction much more than in the other, for example, it does not stretch lengthwise, but it does stretch in width, accordingly, the tensioner needs to be pulled strongly only lengthwise
Quad used MD on the longest side of the 63 panels
Attachments
I measured 3 original diaphragms on drivers that were in good condition before I pulled those diaphragms off for replacement (I don't trust the original glue). They read 86-87 Hz. I set all mine at about 92 Hz (one outlier was 96). I went slightly high because I wasn't sure how much it might change when I applied Licron (it drops about 1 Hz) and the factory diaphragms I was measuring were 40+ years old, and might have "relaxed" a bit over the years.it is amazing you got such high res by the way on the 3 micron ! i had a hard time to reach 75hz (it has been a while so numbers might be off a litlle, i think the originals where in this region according to my measurements, then again they where old panels as well so who knows) keep in mind they used heat while stretching on the ESL63. my thoughts on that is they cant get the desired res by over streching, since it will rip. by heating the foil the drop after a few days is far less then without heating. so they did not have to over strech as hard . there is a topic somewhere here. where i used heat and without heat and plotted the drop in res. maybe the 95 will drop to the 75 hz.. i hardly managed to strech it that hard. quite a few ripped foils 🙂 i did use the same bicycle jig 🙂
I went through about 20m of film to put diaphragms on 8 driver grids. I had some failures, mostly in handling the film. On one driver I managed to cut the diaphragm after replacement twice! The glue I use (4693H contact cement) bonds well to both the grid and the film which made it a royal PITA to redo that one, twice.
I'm not really sure how much the resonance of the panels matters, other than as an indication of sufficient tension on the film. The factory diaphragms all sat at about 86 Hz, yet response of the speaker goes down to 50 Hz or so, and there's no pronounced hump anywhere. There is a huge impedance peak at around 100Hz, maybe (but I don't think) that's related to diaphragm resonance.
The factory diaphragms have greater inertia due to extra weight of the coating and the graphite layer.
In general the original diaphragms are 9-12 micron thick in total (say 3 micron for each layer: Mylar, electrostatic coating, adhesive layer, graphite layer).
So maybe the tension should be different as well, related to the actual weight?
Most important is that the diaphragm stays in the middle when bias voltage is applied and not sticks to one stator side or become unstable.
Quad used a factor of 3.5x the negative stiffness according to Baxandall.
On this page he mentions: "5 Hz which is well below resonance". In case 90+ Hz panel resonance that is an understatement to say the least.
In general the original diaphragms are 9-12 micron thick in total (say 3 micron for each layer: Mylar, electrostatic coating, adhesive layer, graphite layer).
So maybe the tension should be different as well, related to the actual weight?
Most important is that the diaphragm stays in the middle when bias voltage is applied and not sticks to one stator side or become unstable.
Quad used a factor of 3.5x the negative stiffness according to Baxandall.
On this page he mentions: "5 Hz which is well below resonance". In case 90+ Hz panel resonance that is an understatement to say the least.
Attachments
I'm not sure what you're trying to say.
I measured the resonance of three factory diaphragms and got a consistent 86 Hz, so that's about where I set mine- I went a few Hz higher. I used 3 um film for the diaphragms. None of the diaphragms stick to the stators when they are powered up, they make absolutely no noise, and the little neon lamp that flashes when the diaphragm voltage drops below the 5.25 kV bias supply voltage, flashes once every 90 seconds or so indicating very low leakage. I'd call it a successful restoration.
I measured the resonance of three factory diaphragms and got a consistent 86 Hz, so that's about where I set mine- I went a few Hz higher. I used 3 um film for the diaphragms. None of the diaphragms stick to the stators when they are powered up, they make absolutely no noise, and the little neon lamp that flashes when the diaphragm voltage drops below the 5.25 kV bias supply voltage, flashes once every 90 seconds or so indicating very low leakage. I'd call it a successful restoration.
In general the original diaphragms are 9-12 micron thick in total (say 3 micron for each layer: Mylar, electrostatic coating, adhesive layer, graphite layer).
During the rebuild of my ESL63s I indeed discovered that these membranes are some kind of sandwich. I had the impression that there were two equally strong membranes somehow bonded together. So I would guess that there were two sheets of Mylar? Is your statement an assumption, or is it rock-steady knowledge? And if the latter aplies, where do you have this knowledge from (Nota bene: This is a real question, not a hidden contestment).
Background and a playfield for fun: You can strip this sandwich apart e.g. by carefully melting/scratching at the surface with a welding iron. One side I painted squares, the other side I painted some circles before doing so.
Interesting! I dug one of the factory diaphragms out of the trash and tried to get layers to separate but couldn't do it. My speakers were made in 83. When were your speakers made? Maybe they made some change to later models?
It is knowledge, but not rock-steady, there is always room for doubt. Not two layers of Mylar, just one.I had the impression that there were two equally strong membranes somehow bonded together. So I would guess that there were two sheets of Mylar? Is your statement an assumption, or is it rock-steady knowledge?
Order of layers starting from the stator PCB:
1. Electrostatic coating layer
2. Mylar
3. Bonding layer
4. Graphite layer
Order of layers starting from the stator PCB:
1. Electrostatic coating layer
2. Mylar
3. Bonding layer
4. Graphite layer
Am I missing something? If the electrostatic coating layer was at the stator side of the membrane, then there would be no direct contact from the outer aluminium strips of the matching grid to this inner layer meant to be charged?
Instead, I assumed that the graphite powder layer would provide the electrostatic coating. In the fotos one can see that this graphite layer is applyied straight to the upper and lower border in order to establish contact to the aluminium stripe electrodes or the matching grid. While the side rims and the corners are spared from the graphite coating not to cause load losses to the supporting frame structure. So for my understanding, it's the graphite which provides the coating for the electrostatic load.
Furthermore, when you rebuild and coat the 63's, then you certainly not will coat the stator side or the mylar. Or would you? I did not.
And indeed, there is a mistery third layer immediately visible under some magnification, the one you may name the "bonding" layer. But what would this layer bond, then. As seen in the foto, this layer does not spead to the upper and lower rim of the structure. So it's not intended to bond the graphite to the membrane (the graphite instead does expand to the rim in this model of understanding). What would this "bonding" then be good for? Damping the mylar somehow? Or just adding some mass? Mistery ...
Or is this "bonding" layer then the one I previously peeled off? I can't beleive it, because these layers were so similar in strength and structure.
@ Mark Rehorst: Sorry, I currently am away from the serial numbers of my Quads. This will change in some days. I will remember then your quest for date-of-birth
Horror, could Quad really not cope with the problem of making diaphragms for electrostats? It seems that only Stax and Martin Logan did it correctly, because they introduced the conductive metal (graphite) at the stage of making the Mylar. The whole problem is in the dosage, and therefore in the final conductivity and resistance on the Mylar surface.During the rebuild of my ESL63s I indeed discovered that these membranes are some kind of sandwich. I had the impression that there were two equally strong membranes somehow bonded together. So I would guess that there were two sheets of Mylar? Is your statement an assumption, or is it rock-steady knowledge? And if the latter aplies, where do you have this knowledge from (Nota bene: This is a real question, not a hidden contestment).
Background and a playfield for fun: You can strip this sandwich apart e.g. by carefully melting/scratching at the surface with a welding iron. One side I painted squares, the other side I painted some circles before doing so.
View attachment 1447492
View attachment 1447493
View attachment 1447494
View attachment 1447496
View attachment 1447497
It's not rocket science. Licron Crystal works fine.
I am finished:
Yes, they sound great!
I am finished:
Yes, they sound great!
A small tip on a soldering iron set to 265C. Steady the hand holding the soldering iron with your other hand. You touch the tip of the iron against the post and circle around the post, keeping it in contact until you get all the way around. Then remove the little disc of diaphragm film with tweezers. It's pretty easy, but if you mess it up you have to install a new diaphragm...
Last edited:
The diaphragms worked OK. They sound a little complicated, but the real problem was they didn't use the right glue to hold the diaphragm. The stuff they used doesn't bond to the film, so they relied on the shear strength of the very weak bond to hold the diaphragms. Here's me peeling some of the old diaphragm off one of the drivers:Horror, could Quad really not cope with the problem of making diaphragms for electrostats? It seems that only Stax and Martin Logan did it correctly, because they introduced the conductive metal (graphite) at the stage of making the Mylar. The whole problem is in the dosage, and therefore in the final conductivity and resistance on the Mylar surface.
Here's me pulling off one of the diaphragms I glued with 4693H contact cement.
I have some drivers with diaphragms I installed in the 90s (IRIC) using 4693H that are still holding fast. The Licron I sprayed them with is still reading 10^8 Ohms/square, too.
A small tip on a soldering iron ...
A perfect hole, and thank you for the info.
This was the tech way I ended with also. My variant: To thermically "spare" the central distance holder, I used the central hole of a big washer to lead the tip. While using the iron tip melting approach, I observed some potentially unwanted production of tiny melted filaments. I had the impression that filaments building is (amongst other) a matter of iron tip temperature. And the slimmer the tip, the lesser the melted mylar quantity and so the lesser the filaments production. You also may clean the tip after 180° of melting to minimize the amount of melted mylar on the tip. One opening would then require two or three takes. Trial and error on a sample bit of mylar helps improving the technique.
These filaments can worst-case produce a faint, but distinct micro-rattling and high-pitched sound when the diaphragm is moving. Therefore, I did an end control on all these melted openings with a binocular microscope. Beware: You won't see some filaments with "bare eyes". I have the optical impression that there might be such a very small filament at 2h30m in your photo. And it seems to me that this filament is expanding in 3h direction also: You may see it projected over the 3rd hole as seen from the central pillar of the stator. But this also might be another artefact, not related to a filament.
If there were filaments, then they easily could be melted/corrected away. Or broken away when cautiosly stripped a bit with a tweezer. While doing so, I was enlighted how stable the mylar foil is: You can pull on the filaments with a quite wide safety margin before the mylar would tear. None of my membranes were torn this way.
One of the two fotos below is from one of my very first melting approaches: Filaments a gogo.
Last edited:
I had some hairs form in the process, but I pulled them off with tweezers. I use 5x loupes when I do this stuff so I can see what I'm doing. I keep the temperature of the iron as low as possible to reduce the production of the hairs. It seems like 265C is about the right temperature.
I like the idea of using a washer as a guide....
I like the idea of using a washer as a guide....
Oh nice i only see now you used a ball to hit the resonance. nice ! i used a second driver to do so. but this is a much simpler nice solution 🙂I've seen photos of the factory jig before. Crazy!
Here's my diaphragm tensioning jig. It's an improved version of one I invented back in the 80s when I started making ESL. It has a bicycle tire tube around the perimeter of the jig. There's double stick tape on the underside. You just wrap the film over it, stick it to the tape and put some air in the tube. All the wrinkles disappear and you can get the film as tight as you want it.
I can measure and tune the resonance by adjusting the air pressure in the tube, while the diaphragm is on the jig. I excite the resonance by thumping the diaphragm with a steel ball hanging from a thread. The RTA in REW plots the spectrum of the signal from the UMIK-1 mic.
View attachment 1447029
After I tune the resonance I glue the grid to the film and then test resonance again using the same technique. There's a slight difference is size between my jig and the ESL-63 grid and the result is that the diaphragm resonance on the grid is always about 10Hz higher than on the jig, so I just set the resonance on the jig about 10 Hz below the target and it comes out fine.
I wanted to duplicate the factory diaphragm resonance, so I tried a few different techniques. I started very simple by making marks on a factory diaphragm that were 500 mm apart. Then I cut the diaphragm free of the driver gird and measured the spacing between the marks. Then I was going to put marks at that spacing on the untensioned diaphragm and stretch it until the marks were 500 mm apart. The problem is that very small changes in the spacing result in very large changes in the resonance, so that didn't work well.
I also considered using a speaker mounted under the driver and sweeping a tone to look for the resonance, but the speaker I tried it with had a broad resonance at about the same frequency as the ESL, and I figured trying to separate them would be a problem.
The tight diaphragm is like a drum head, so I decided to try the thumping technique to excite the resonance. You can get different results if you thump and place the mic at different spots on the speaker. I suspect that the thump sends out ripples that reflect off the edges of the driver/jig and bounce back setting up standing waves on the surface, some of which will be at different resonance frequencies based on the dimensions of the driver.
At one point I set up my AWG to produce a thump at 86 Hz to duplicate the sound of the thump, thinking I might be able to tune the resonance by ear, but there's no way to do that! The thumps are very brief and it's very difficult to compare them audibly. Maybe someone who tunes pianos for a living could do it, but I can't.
I tried recording the sound of the thumps using a digital recorder then brought the file into Audacity and used the FFT to get the spectrum. This looked very promising. I looked at REW and ordered a UMIK-1 mic. Once it arrived I tried the RTA in REW and could see the spectrum of the thumps in real time ( and the resonance obtained that way agreed with what I saw in Audacity) and that made it possible to tune the resonance while the diaphragm was on the stretcher. I would stretch the film to the target resonance, then apply glue to it and the grid and wait 20 minutes for the glue to set, then I'd check the resonance of the film one last time to make sure it hadn't changed while the glue was setting (sometimes the tape on the stretcher comes loose and that reduces the tension on the film and the resonance), and then stick the grid down on the film. It would bond instantly, but I always gave it a couple hours before I cut the grid free of the stretcher.
I standardized on putting both the mic and thumper as close to the center of the speaker as possible, and it seems to work reliably. The stretcher has the center of the open area marked so I use those marks to position the mic and thumper and when I test a driver I just use the center screw post as the reference point. I tested factory diaphragms this way and saw 86 Hz resonance, so I aimed to duplicate that (actually to go a little higher).
It seems to have worked well, because the speakers sound really good.
I also considered using a speaker mounted under the driver and sweeping a tone to look for the resonance, but the speaker I tried it with had a broad resonance at about the same frequency as the ESL, and I figured trying to separate them would be a problem.
The tight diaphragm is like a drum head, so I decided to try the thumping technique to excite the resonance. You can get different results if you thump and place the mic at different spots on the speaker. I suspect that the thump sends out ripples that reflect off the edges of the driver/jig and bounce back setting up standing waves on the surface, some of which will be at different resonance frequencies based on the dimensions of the driver.
At one point I set up my AWG to produce a thump at 86 Hz to duplicate the sound of the thump, thinking I might be able to tune the resonance by ear, but there's no way to do that! The thumps are very brief and it's very difficult to compare them audibly. Maybe someone who tunes pianos for a living could do it, but I can't.
I tried recording the sound of the thumps using a digital recorder then brought the file into Audacity and used the FFT to get the spectrum. This looked very promising. I looked at REW and ordered a UMIK-1 mic. Once it arrived I tried the RTA in REW and could see the spectrum of the thumps in real time ( and the resonance obtained that way agreed with what I saw in Audacity) and that made it possible to tune the resonance while the diaphragm was on the stretcher. I would stretch the film to the target resonance, then apply glue to it and the grid and wait 20 minutes for the glue to set, then I'd check the resonance of the film one last time to make sure it hadn't changed while the glue was setting (sometimes the tape on the stretcher comes loose and that reduces the tension on the film and the resonance), and then stick the grid down on the film. It would bond instantly, but I always gave it a couple hours before I cut the grid free of the stretcher.
I standardized on putting both the mic and thumper as close to the center of the speaker as possible, and it seems to work reliably. The stretcher has the center of the open area marked so I use those marks to position the mic and thumper and when I test a driver I just use the center screw post as the reference point. I tested factory diaphragms this way and saw 86 Hz resonance, so I aimed to duplicate that (actually to go a little higher).
It seems to have worked well, because the speakers sound really good.
Last edited:
Smart. Measure resonance while tensioning. It's like applying the icing on the rebuilder's cake.
As for passively stimulate the membrane's resonance: What about make explode an air-filled balloon behind the diaphragm? I used this balloon explosion technique for multiway reflex measurements within a room. Drawback: Every exploding balloon shows a different energy spectrum. But I imagine that for a membrane's resonance stimulation & measurement this spread might be of secondary relevance.
As for passively stimulate the membrane's resonance: What about make explode an air-filled balloon behind the diaphragm? I used this balloon explosion technique for multiway reflex measurements within a room. Drawback: Every exploding balloon shows a different energy spectrum. But I imagine that for a membrane's resonance stimulation & measurement this spread might be of secondary relevance.
You only get one shot at it with a balloon, so it's not a real-time friendly tuning technique. With the thumper I can keep thumping the thing once every second or so and average the results of multiple thumps. It works very well, and it's very easy to do, extremely low cost (the UMIK-1 mic isn't really necessary- you can probably use the mic in the laptop that is running REW), and I can rerun the test in seconds after putting a few more stokes of the pump into the tube.
When I was tensioning diaphragms I would put enough air in the tube to take out the wrinkles and add a bit more, thumping the diaphragm with my finger, until it sounded like about 60Hz resonance, then I'd tilt the stretcher up and start using the thumper to tune it. 5 or 6 strokes of my pump usually moved the resonance up by a couple Hz. Once it reached the target, I'd disconnect the pump, tilt the stretcher back down, wipe the diaphragm with IPA, and apply glue.
When I was tensioning diaphragms I would put enough air in the tube to take out the wrinkles and add a bit more, thumping the diaphragm with my finger, until it sounded like about 60Hz resonance, then I'd tilt the stretcher up and start using the thumper to tune it. 5 or 6 strokes of my pump usually moved the resonance up by a couple Hz. Once it reached the target, I'd disconnect the pump, tilt the stretcher back down, wipe the diaphragm with IPA, and apply glue.
- Home
- Loudspeakers
- Planars & Exotics
- QUAD ESL-63 rebuild