Some progress. All membranes use identical coils, 30 um self adhesive aluminum with ~ 30 um acrylic glue.
Ordered from top to down:
The kapton with the glue is of course the heaviest. The 10+15 composite is very much lighter, even though it in theory has thicker aluminum. It is so light I get suspicious if it might be thinner than we think it is. But if so then that is just better for me, it will be interesting to see how it compares.
The plain mylar is of course the lightest, even with the border. And keep in mind that on the top end, the wavelength is short enough that the coils act independently of the suspension stiffener so it should have by far the most top end.
Before I corrugate I plan to make a clone of the just 12 um mylar as a control such that I can bake 1 in the oven and not the other and compare the stiffness afterwards.
Ordered from top to down:
- 12 um kapton + 12 um silicone adesive + 12 um aluminum. This is pretty much a remake of my previous best. I decided to remake it since I will revert to the original corrugation pitch based on @solhaga s findings in his thread so to ensure I compare apples to apples I might as well make another one.
- A 10 um mylar + 15 um aluminum composite based on some samples I got from @solhaga (Thank you! ❤️)
- Just 12 um mylar (+ a 3 mm 30 um self adhesiveu aluminum border at the suspension). After measuring the weight I also added extra stiffening to the top and bottom to make it easier to mount and reduce the risk of ruining the membrane when soldering the wires.
The kapton with the glue is of course the heaviest. The 10+15 composite is very much lighter, even though it in theory has thicker aluminum. It is so light I get suspicious if it might be thinner than we think it is. But if so then that is just better for me, it will be interesting to see how it compares.
The plain mylar is of course the lightest, even with the border. And keep in mind that on the top end, the wavelength is short enough that the coils act independently of the suspension stiffener so it should have by far the most top end.
Before I corrugate I plan to make a clone of the just 12 um mylar as a control such that I can bake 1 in the oven and not the other and compare the stiffness afterwards.
Last edited:
Yup, although I will use a less fancy temperature controller... manual adjustments + a separate probe 🙂
I think it should be close enough, although if I need more consistent temperatures I can always use a pizza stone but it takes a while to get up to temperature, but when there it should be very consistent during the whole baking time.
The 30 um aluminum has a 30 um thick acrylic adhesive on it, which based on my googling should be able to handle 150 degrees long term (days - weeks), and short bursts of 230 degrees.
So I think the adhesive will be fine, probably even if I went higher like 150.
Some progress, but no measurements yet, that will have to wait until tomorrow.
I corrugated all the foils and then baked one of the plain mylar ones. When I had the oven on 135 degrees celcius with the fan on then it fluctuated between 130 - 145 degrees celcius, close enough for me. I baked it for 20 minutes but to be honest I don't feel much of a difference.
It will be interesting to see if it performs any differently but my gut feeling is that it will perform identically to the control membrane. Maybe it wasn't the mylar that was softening and getting reshaped @solhaga but rather your adhesive which made your membrane stiffer?
Also, this time I split the solder pads to an inner circle and then an outer ring, to reduce the risk of the adhesive releasing from the foil and so far it worked great.
Although for the final driver I am considering skipping wires altogether. My rough idea is to say haev the above setup but instead of a solid aluminum circle in the middle of the pad, to make it slightly bigger and then have a hole. Then order a kapton-copper flexible pcb with a pad under the hole and then first glue the flexible PCB to the underside of the membrane, and then bridge the aluminum membrane to the flexible PCB pads. Thus there would be less strain on the actual membrane and less risk of damaging the membrane.
Also, when I mounted some membranes I realized that I am really pushing it to the edge with my membrane width, there is not a lot of wiggle room before touching the orange edges of the driver. I am considering reducing the membrane width by 2 mm to have 1 mm more margin on each side.
I corrugated all the foils and then baked one of the plain mylar ones. When I had the oven on 135 degrees celcius with the fan on then it fluctuated between 130 - 145 degrees celcius, close enough for me. I baked it for 20 minutes but to be honest I don't feel much of a difference.
It will be interesting to see if it performs any differently but my gut feeling is that it will perform identically to the control membrane. Maybe it wasn't the mylar that was softening and getting reshaped @solhaga but rather your adhesive which made your membrane stiffer?
Also, this time I split the solder pads to an inner circle and then an outer ring, to reduce the risk of the adhesive releasing from the foil and so far it worked great.
Although for the final driver I am considering skipping wires altogether. My rough idea is to say haev the above setup but instead of a solid aluminum circle in the middle of the pad, to make it slightly bigger and then have a hole. Then order a kapton-copper flexible pcb with a pad under the hole and then first glue the flexible PCB to the underside of the membrane, and then bridge the aluminum membrane to the flexible PCB pads. Thus there would be less strain on the actual membrane and less risk of damaging the membrane.
Also, when I mounted some membranes I realized that I am really pushing it to the edge with my membrane width, there is not a lot of wiggle room before touching the orange edges of the driver. I am considering reducing the membrane width by 2 mm to have 1 mm more margin on each side.
That amount of un driven foil...i know it is the idea to provide shading.. but i wonder by now if this is a good idea. you are dragging along much more weight on useless foil then actual coill 🙁 2 traces ? around 5-10 % is actually driving the rest is dragged along.
listen to a light coil at the top end vs as heavy coil. it might change your idea of making it even more heavy. in my older designs it was clear 30 micron alu on mylar was no match for 15 or less on mylar. i cant imagine how it is with 30 while 80-90% is not actively driven. if you got the time could you try one with regular foil 30 but allot of coil or less thick if you can? it might change your approach. (having as much coil on the foil as well matters, even if it is not in the perfect field it will get rid of breakups etc)
listen to a light coil at the top end vs as heavy coil. it might change your idea of making it even more heavy. in my older designs it was clear 30 micron alu on mylar was no match for 15 or less on mylar. i cant imagine how it is with 30 while 80-90% is not actively driven. if you got the time could you try one with regular foil 30 but allot of coil or less thick if you can? it might change your approach. (having as much coil on the foil as well matters, even if it is not in the perfect field it will get rid of breakups etc)
Last edited:
You have pretty accurately described my own reaction when I held it 😆 , it is nowhere near as stiff as the solid membranes.
And no, I don't think it is a good idea, it makes more crinkly noises than I feel comfortable with but for now I mainly want to test if the baking / tempering of the mylar makes any difference. The interesting experiment, will be when I get my hands on twice as thick mylar (23 um instead of the current 12 um), in that case it might be fine even with all that undriven foil, or at least that is what I want to find out 🙂
Yes, I have tried that. The maximum efficiency is at 7-9 mm wide traces, where 9 mm is practically equivalent to covering the whole membrane in driven coils since the magnet-to-magnet-distance is 10 mm.
It works well enough, the main downside is that with such a setup, we can't have the thin 3 mm coils that have the best magnetic field performance, so we get a lot more distortion on the low end. Or rather the total amount of harmonic distortion is roughly the same, but there are a lot more higher order distortion products so much less usable than the 3 mm wide coils.
When I build and test the full scale driver, then one thing I want to try is to skip the 3 mm wide traces and instead go with wide traces. It will have higher distortion sure, but will it be a problem? One possibility is that yes, when pushing xmax it will distort a lot but maybe with such a large driver I might hit my target SPL without getting close to pushing the xmax so it might not be a problem after all.
Last edited:
hmm never encounter that problem . so if gap is 7 mm you can easily use 2 x 3mm traces. (i myself go over the magnets and leave as little as free mylar open on the middle of the magnet) it might be a a discussion about having proper top end vs trying to go as low as possible with less distortion (priorities are top end the lower you go the less we as human hear them) although if they will be large in the end (at least 1.2 meter or something i think , thats all not relevant since they will be insane low. so if you test a 30cm one and top end sucks it only becomes worse the longer you make them. compared to low en distortion in 5th order 7th etc you cant hear , and becomes less and less when growing larger in length. i would not put that much effort in the low end, or calculate it back when using 4 times the size. .... it will be almost 4 times lower, while the top end becomes worse. i guess its a balance but since they already drop of at 15khz. i would shift a little to top end , the lows are far lower then you make the huge build anyway. you cant make the top end better by adding more length
Are they they all the same alu thickness, i mean you did take into account impedance ? i use often traces up to near the middle of the magnet (leaving 1-2mm without coil, dont want to cross it ofcourse) (yes its not driven as hard there compared to having a few traces in the middle of the best field. but if you look at FEM its not an insane loss., the gap is strong and the edge of the magnet is also strong.
I did not notices any increase in distortion when doing so i did notice FR been far more smooth. in the upper range. but the lower range will have 4 times the surface area when your done compares to the top end.
did you put 2 x 5mm in the 9 mm gap ? or 1 trace 5mm ?
if magnet to magnet distance is 10 mm,,, covering 9mm in between is not the whole foil 🙂 most efficient yes 🙂 (it depends on how wide the magnet is, i leave 1-2 on the middle of the magnet free of coil, this will make the FR smoother, and you lose some output, but not insane)
Exactly.
In the end everything will be a compromise. And while I want good low end end performance, I don't want to sacrifice the top end too much. I would love to be able to cross to the midtweeter planar @ 400 hz, which based on the distortion performance of 1/6 of the full driver looks to be doable. I could cross higher though, I could push it up to 600 hz which reduces the required xmax by a lot, which in turn means that those tall order distortion products with a wider membrane are not a problem anymore.
Here are some distortion measurements of 3 mm vs 6 mm.
Yes, the 6 mm coil, when pushed to its xmax, it distorts a lot in a bad way. But those measurements are with a 370 hz LR4 roll off. And with a single driver, the full driver will after shading be roughly 4x this.
And there is an argument that that 2nd order harmonics are more pleasant than 3rd, so if we ignore the huge tall order distortion peak at 500 hz and below, the 6 mm wide traced membrane actually performs better. But also, say we cross at 600 hz in addition to a full driver with 4x the size, then it is entirely possible that all those distortion products will be pushed far down enough into the noise floor to be completely insignificant.
But you did had 2 x 3 mm vs 1 x 6mm ? i dont know what width your magnets are ? but lets say they are 5mm wide and you got a gap of 10 mm between magnets (witch is big by the way for neo's) that means for what i use it would be 13mm usable gap give or take
Second order will raise in any case, even for push pull. well less then single ended. but even then its not annoying at all.
i dont know what your goal is for spl 🙂 but if you got 4 times the amount of foil... i cant imagine its a huge problem. pretty sure it will play loud and clean. besides that it will add up in the low end and not in the top so you got room to spare.
Second order will raise in any case, even for push pull. well less then single ended. but even then its not annoying at all.
i dont know what your goal is for spl 🙂 but if you got 4 times the amount of foil... i cant imagine its a huge problem. pretty sure it will play loud and clean. besides that it will add up in the low end and not in the top so you got room to spare.
even in bass panels using only less then the gap looks worse then using more foil. driving it more uniform.. in your measurements we are taking 1-1.5db dB ? i would not focus on 1 dB to be fair. you can get that back easily with a lighter foil.
then again i never was or am a fan of getting as much efficiency as possible 🙂 most often it is loud enough to mate with another driver. so then i rather have it smoother
then again i never was or am a fan of getting as much efficiency as possible 🙂 most often it is loud enough to mate with another driver. so then i rather have it smoother
The wider coils had more number of traces to make the impedance more similar, but in the end when measuring I normalized the SPL and used an amp that could handle all the different impedances without adding any significant distortion so the end performance is all due to the coils.
But they did had the same amount of aluminum so they weighed the same, the difference was in how much of the aluminum was driven vs undriven.
I also added the 3 mm to the example since this is basically the main reason I started playing around with different aluminum thicknesses for the driven vs undriven parts of the membrane. If I have the same alu thickness with 3 mm coil then the majority of the weight is in undriven alu, so it is crazy inefficient. But if I double the alu thickness of the coil, then the end efficiency turns out to be pretty much the same since while it is slightly heavier, the magnetic field is much stronger in the 3mm region so it evens out.
Then of course came the next step which was thinking, hey, what if I just have a thinner but solid sheet of alu on the other side of the kapton / mylar? It is way stiffer since there are no gaps without aluminum. And then this evolves into trying out alu-mylar composites which are even lighter
I think I had 3 x 2 mm for the 6 mm wide coils, and then probably 4 x 2.25 mm for the 9 mm.
The magnets are 5 mm wide, and the gap in between the magnets is also 5 mm. So pretty close to the dimensions of the Neo3 except that my xmax is bigger and the Neo3 has wider magnets and a smaller gap.
So in my driver the absolute maximum possible coil width would be 10 mm, assuming I have 0 margin so in practice 9 mm was close enough. But my experiments showed that there was no benefit in driving more than 7 mm of coil width. The extra width does not add efficiency and does not improve distortion either. Although both have bad distortion compared to the 3 mm wide coils.
My rough target is to be able to hit 106 dB @ 1m with the target crossover roll off in place, without significant odd or tall order distortion products.
So basically at least as good as my current SB65 cone CBT, but I won't complain if it is better 🙂... and based on my current measurements, the DIY planar looks to far exceed the cone drivers in performance.
yeayh i just wondered where they are all lets say 3 ohm, and not a few been 2 ohm since that would already explain the difference 🙂
Ah ok so 10 mm is then magnet center to center 🙂 that makes sense !
Neo 3 has 5 mm and 5 mm space in between as far as i know , but a very tiny xmax.. 0.5 or less. thats why they all went to **** when the coating left the magnets. i have multiple neo 3's because i wondered. most of them dead.
thats a really good distortion measurement, the only one i have that reaches low (at these high volumes) that is planar is this one with the weakest magnets possible . so expensive does not mean everything i guess 🙂 @ 100dB above 700hz (my amp back in that day would not put out more... a random suround amp from onkyo) clearly res is at 200 hz 🙂 push pull rubber magnets 🙂 weakest of all magnets
Last edited:
i think it might be hard to beat a dynamic driver in the lower end though 🙁 unless you got major surface area 🙂
It looks like there was no corrugation support foil present when you baked the membrane, only the membrane itself?
I make a sandwich of 40µm aluminium sheets with the membrane in between wound up on a steel cylinder.
The whole thing is then baked:
Here's what a bare (no adhesive) 3M 74 PET film (13µm) looks like after baking at 140°C for 10 minutes:
It's springy and can hold a moderate weight without straightening.
So it behaves like a rigid beam.
Last edited: