I rebuilt my ESL-63s using a diaphragm stretcher made from MDF and wood. It worked well, but after multiple uses, adhesive from the tape used to secure the film built up on the wood and the tape started letting go before I could glue the diaphragm to the stator grid. I found it very difficult to clean the adhesive off the wood. Also, the opening in the center didn't match the size of the driver, so the resonance measured on the stretcher was different from the resonance measured once the diaphragm was glued to the driver. I'm rebuilding more drivers (going to turn my 63s in 989s) so I decided to try to make a better stretcher, one that would, I hope, provide the same resonance on the stretcher and driver, and wouldn't have the same problem with tape adhesive.
I found some promising aluminum extrusion in my materials pile collection. I tried sticking some of the tape to the aluminum and pulling it off, multiple times and found it didn't leave residue behind like it did on the wood stretcher. The next thing to do was model the aluminum in CAD.
 |
| The aluminum I used has this profile. It's 56.8 mm high x 42.5 mm wide. Those wide, flat areas on the top, bottom, and sides are very useful for this application. I have no idea who makes this particular stuff, and neither does Google Lens. Square or rectangular aluminum tubing would work as well and would be it would be easier to design and print corner pieces. |
I measured the drivers and found the opening to be exactly 583 x 175 mm, so I cut the aluminum a few mm longer than needed with a saw, then milled it square to exact, matching lengths. The round holes in the aluminum are sized for a 1/4" tap, so I tapped them with 1/4-20 threads.
The next step was to design and 3D print corner pieces to hold the aluminum rails. I split each corner piece into two identical parts so I could print the part that inserts into the aluminum without using any support material. The screw holes were printed 4mm in diameter and drilled out with a 1/4" drill after printing.
 |
| This is what the corner pieces look like. Support material is used inside the screw insertion slot. The slot is about 12mm high to accommodate the button head cap screws that mount the corner pieces on the aluminum rails. |
 |
| One of the corners holding two of the rails together. It's a tight fit! |
 |
| The whole frame. The 1/4" hole in the green rail is for the tire tube valve stem. I had to get a 20 x 1 3/8 tube with a 60mm long stem and Presta valve. |
 |
| Here's a driver sitting on the frame. The opening in the frame matches the driver (583 x 175 mm) within a fraction of a mm. The plastic corners fit so tightly I had to tap them in with a rubber mallet. |
The next part of the design was to make a base that would allow the stretcher to be positioned vertically for resonance testing. In the previous design, when the stretcher tilted up, one edge hit the hinge support and stopped it from tilting further. That meant that during resonance testing, the tensioned diaphragm was in contact with the hinge support. I didn't like that. It doesn't take much to puncture and tear the tensioned film, so it's best to avoid physical contact with the film.
In the new design, I added "pins" to use for the hinge and an additional one for the support/tilt stop. It will allow the stretcher to tilt up to vertical, and I'll add a cord that will prevent it tilting so much that the edge hits the hinge support.
 |
| I installed three bolts for the hinge and support/stop. |
 |
| The stretcher mounted on the stand. Yup, that's all there is to it. All that's left is to add a cord to the support/stop to prevent it from tilting too far, and some neoprene foam to the top surface of the stretcher frame. |
 |
| First test with 6 um film. Worked well, but pointed out some minor issues. The pump hose has a pressure indicator that makes it hard to attach and detach the hose without accidentally closing or opening the valve. The film ultimately split, I think due to something sharp along the bottom edge of the frame. |
After the film split during the first test I used a scotchbrite pad and ran it over all the edges of the stretcher to smooth out anything sharp that might have caused the film to split. I had no further problems with film splitting.
The pump/hose I was using on the wood stretcher didn't work so well on this one. The pressure indicator on the hose took up too much space. I ordered another small tire pump that came with a hose and no pressure indicator. That solved the pump problem.
There's one more problem I've been struggling with since I bought the roll of film back in the 80s. When the film is pulled from the roll, it generates a static charge that causes it to try to stick to anything and everything nearby (and pulls dust, cat hair, etc., from the air). When I put the film on the stretcher, it immediately curls under and tries to stick to the tape. Then I have to try to pull it free so I can position it where I want.
I decided to try to make a proper dispenser for the film that would kill the static charge generated when the film comes off the roll. I designed a couple conical end pieces with F608 skate wheel bearings and added two grounded steel tubes for the film to pass over as it comes off the roll. 8mm bolts go through the bearings and extend through the printed mounting brackets that are screwed to a piece of wood.
 |
| One end of the film dispenser. The other end is a mirror of this one. There are two conical plugs that fit the ends of the film roll, each of which has two F608 skate wheel bearings and an 8mm bolt. The ends of the bolts pass through the end supports that are screwed to a piece of wood. The film comes off the top of the roll, goes down under the bottom steel tube, wraps between the tubes, then pulls over the top of the upper steel tube so both sides of the film contact the steel. |
The bad news is this method doesn't actually take the charge off the film. It simply provides a new means of charging the film by sliding the plastic over the steel tubes. The good news is that it doesn't matter. I placed the film dispenser on the work table just behind the stretcher and I worked out a technique for attaching the film to the stretcher that works fine even with the charge on the film. Getting rid of the static charge on the film would require some sort of ion generator that would spray the film with ions and neutralize the charge as it comes off the roll.
Now I pull the film off the roll right over, and attach it to the two front corners of the stretcher. Then I cut the film free of the roll and attach it to the stretcher at the back corners, pulling wrinkles out of the film as much as possible as I do it. Then I attach the film to the short sides of the stretcher and finally to the long sides. The new stretcher allows me to see the film as it attaches to the tape.
 |
| Film coming off the roll and getting attached to the stretcher. I've added neoprene foam (black) around the edge of the opening in the top of the stretcher. |
Once the film is attached, I put some air in the tube and tilt the stretcher up to check the resonance and adjust the air pressure as needed to get to the target value (about 80-85 Hz).
Next I wipe the film and driver grid with IPA, then apply the 4693H contact cement to both, wait 15-20 minutes, check and adjust resonance again, and stick them together. I have found that with this stretcher, the final diaphragm resonance will be about 6 Hz higher on the driver grid than it is on the stretcher.
 |
| Resonance testing. I have added a couple marks on the stretcher frame (not visible here) indicating where the center of the diaphragm is to make positioning the microphone as accurate as possible. You can see the thin, black cord that is used to limit the tilt-up. |
I pile some weight on top of the driver grid sitting on the stretcher and wait a few hours before letting the air out of the tube and cutting the driver free. Then I use a soldering iron set to about 260C to make the holes in the diaphragm around the center posts. When making the holes, I wear 5x loupes so I can see clearly, and keep the tip of the iron in contact with the post as I circle around it. Sometimes this creates fine plastic hairs that I remove using the soldering iron.
After letting the drivers sit for a couple days, I clean the excess glue off all the edges, tape off all the long edges, spray and wipe the diaphragms with IPA, then position the 3D printed masks in the center holes and on the ends, and spray with Licron Crystal. I make 4 passes, alternating L to R and R to L, then turn the driver around and do it again. I set the driver aside to dry while I spray the next one. I get reliable 10^8-10^9 Ohms/square resistivity using this technique.
 |
| Four drivers with newly recemented stators and new diaphragms, with long sides taped for spraying with conductive coating - Licron Crystal. |
 |
| 3D printed masks to block the conductive coating in selected areas of the driver. |
I test the resonance after spraying on the Licron Crystal coating. I've been able to get pretty consistent results- here are resonance plots of 4 drivers I recently tested. Note- the lowest frequency peak is the one I use to set the resonance. The the drivers have multiple resonances and the mic picks up other sounds in the room (the AC running, garbage trucks outside, neighbors vacuuming, etc.), so it's safe to ignore the other stuff.
Final testing is done using electronics from an ESL-63 speaker- I have alligator clips on all the wires that go to the driver, and simply connect an amplifier and apply a signal. What I'm mostly looking for here is any hissing or whining noises coming from the driver when the 5.2 kV bias is applied, and any odd sound that a damaged driver might produce.
 |
| The two white boxes contain all the electronics from an ESL-63 speaker. I just connect the driver using the alligator clips, power it up and give it an audio signal from an amplifier. |
The two halves of the early model drivers I have are held together by the three center screws and the four corner screws that hold the driver in the frame in the speaker. Newer ESL-63 drivers came with small metal clips to hold the two long sides of the drivers together. I think those clips are a good idea, so I ordered some 1/2" binder clips because they should fit in the 13 mm wide spaces in the driver grid. The 1/2" clips are actually 15mm wide (only in 'murica folks!) so I had to grind them all down to fit.
When I am ready to add more bass panels to the ESL-63s I'll write another blog post.
