My DIY electrostatic headphones

[GvTT]

Hello everybody,
 
I have had my mill running quite a bit this weekend:
 

Finishing the outer edge of the stators with a 4mm miller.
The mill is too small to machine the whole stator in one clamping, so I had to turn the parts when they are done on one side.
 
And this is the result:

 
The next thing will be to remove some of the copper on the stators to reduce capacity. Then I will smooth everything and paint the stators. A workshop cleaning will be necessary too. The I finally come to the Mylar.
 
The inner tire stretcher is already build.
 
What do you use to control the pressure? (a normal bicycle pump? an additional release valve?, ...???)
 
Greetings
Georg 

 

[davidsh]

Just want to say that removing the surplus copper from the stator will reduce the capacitance, though not significantly. The capacitance of the cable is mostly significantly greater than that of the electrostat itself. So don't bother if it is of much trouble 

 

[chinsettawong]

Hi Georg,

Those parts look great.

About inner tyre pump, I use a small handheld pump. Slowly observe the diaphragm. My elongation target is only around 0.5-1 mm. You might want to try a few times until you get it just right.

About etching out the unneeded copper, if you prefer the sound of Stax SR007 to SR009, then you don't need to do any etching. But if you like SR009 better, then you do. You can try to experiment and see it for yourself which way you like better.

 

[davidsh]

Why is that going to make much of a difference, except for slightly higher capacitance?

 

[chinsettawong]

Yes, it does make quite a lot of difference to my ears.

 

[davidsh]

I guess, I would prefer a logical explanation. I don't like to trust other people's ears when I don't have to, but it isn't like I'm going to make my own 'stats any time soon.
 
I'm just saying that from a theoretical stand point it might be better to look into optimizing the cable, if capacitance is of concern. Of course, the resistance of the cable won't matter if within reasonable limits.
For a Stax headphone the driver capacitance amounts to ~1/5th of the total capacitance.
 
An easy way to lower the capacitance is to make your cable shorter.
 
Whether to be concerned also depends on the amp that you are using and how loud you listen as you might experience problems in the high frequencies if the amp can't swing the current needed. I think the result merely is that the frequency response becomes rolled off in the treble, though I am not certain.

 

[GvTT]

Hi,
 
I roughly calculated a cable:
 
3m cable, 1.2mm copper diameter, 2x 6mm apart (bias in the middle), PVC as insulator gives me
  Inductance: 5,5μH
  Capacitance: 1.8nF
 
My stators (2x 0.5mm spacer):
  unetched: 57.4pF
  rim and outer area etched: 28.4pF
 
As DAVIDSH mentioned, the cable dominates.
 
I tried to modulate this as a lossy line and a capacitor in Spice. There I get corner frequencies of 100kHz. So this is out of scope as explanation.
 
The calculation clearly shows that the capacity between the stators is significantly reduced by etching them. Especially with my stators, that have most of the copper on the outside.
 
I did not consider the edge of the spacer the edge of etching (I like this sentence but a line about 5mm away from the spacer.
This is reasoned by the fact that a force on the edge of the membrane will contribute very little to it's displacement.
 
So, if the cable is dominant, what causes the difference in sound? (I take wachara's observations as serious as davidsh's comments)
And, even more important, how would a good cable look? (Could a sense line adjust the amp to correct the cable?)
 
Unfortunately I have not had the chance to listened to an SR007 or SR009.
 
Greetings
Georg

 

[edform]

   
This is an electret and not an electrostat and therefore will have LF excursion problems. If you see the FR file in their site, it shows a steep roll off under 1KHz IIRC.

It is not an electret. It uses a rather high bias voltage - in the order of 1600volts I believe - and is a pure, single-sided electrostatic device.
 
Ed Form

 

[davidsh]

  Hi,
 
I roughly calculated a cable:
 
3m cable, 1.2mm copper diameter, 2x 6mm apart (bias in the middle), PVC as insulator gives me
  Inductance: 5,5μH
  Capacitance: 1.8nF
 
My stators (2x 0.5mm spacer):
  unetched: 57.4pF
  rim and outer area etched: 28.4pF
 
As DAVIDSH mentioned, the cable dominates.
 
I tried to modulate this as a lossy line and a capacitor in Spice. There I get corner frequencies of 100kHz. So this is out of scope as explanation.
 
The calculation clearly shows that the capacity between the stators is significantly reduced by etching them. Especially with my stators, that have most of the copper on the outside.
 
I did not consider the edge of the spacer the edge of etching (I like this sentence but a line about 5mm away from the spacer.
This is reasoned by the fact that a force on the edge of the membrane will contribute very little to it's displacement.
 
So, if the cable is dominant, what causes the difference in sound? (I take wachara's observations as serious as davidsh's comments)
And, even more important, how would a good cable look? (Could a sense line adjust the amp to correct the cable?)
 
Unfortunately I have not had the chance to listened to an SR007 or SR009.
 
Greetings
Georg

I am no electronic experts, but I do know a little about physics. The biggest problem with high capacitance in your 'stat is that it requires the amp to deliver significant current at high voltage swings, high frequencies. I have not done the calculations, but for most amps driving a single Stax headphone, playing music, I honestly do not think current delivery will be a general issue, though it certainly could be in some cases.
 
The best way to reduce cable capacitance is to seperate the conductors from each other as much as possible. And keep the cable short. It doesn't have to be costly if you are a little creative, and it isn't rocket science, though there will be other effects contributing to the capacitance in the low pF range, I'd guess.
 
1.8 nF is a high capacitance for a 'stat.
 
The typical maximum current of most 'stat amps is some 1~10mA to my knowledge. With a capacitance of 1.8 nF you'd be using a maximum current of ~1.4 mA at Vpeak = 300V at 20kHz to put it in context. Mostly, you will not need such high voltage swings in the high frequencies for reproducing music.
 
An example where the high capacitance could be problematic is if there's significant ultrasonic content, noise etc, for whatever reason in the music you play (aliasing, high-res music files containing HF-noise etc.). Then the sound quality may suffer, though I am not sure how the amp would react (perhaps somebody else might know). I would guess it'd clip somehow.
 
If you have an oscilloscope around it is possible to measure the capacitance of your 'stat, though that takes further explanation. Of course there are instruments for measuring this as well.
 
High frequency roll-off will be of no concern as the corner frequency will be very high in general, as you noted yourself.
 
I hope that clarifies a couple of things.

 

[davidsh]

 

   
This is an electret and not an electrostat and therefore will have LF excursion problems. If you see the FR file in their site, it shows a steep roll off under 1KHz IIRC.

It is not an electret. It uses a rather high bias voltage - in the order of 1600volts I believe - and is a pure, single-sided electrostatic device.
 
Ed Form

How is that so with such a small seperation between stator-diaphragm? One would expect problems with stability, corona discharge, collapsing diaphragms and all that lovely stuff because of the very high electric field. Must yield a high sensitivity, however.

 

[ayaflo]

 
It is not an electret. It uses a rather high bias voltage - in the order of 1600volts I believe - and is a pure, single-sided electrostatic device.
 
Ed Form

 
Ah apologies, I was under the impression its an electret. Either way per their website its 450V bias.

 

[GvTT]

...

About inner tyre pump, I use a small handheld pump. Slowly observe the diaphragm. My elongation target is only around 0.5-1 mm. You might want to try a few times until you get it just right.
...

 
Hello Wachara,
 
if you are talking about 0.5 to 1mm elongation while stretching, what is the reference length you are referring to?
 
It make a big difference if I stretch 200mm by 1mm  (0,5%) or 20mm (5%) please provide some more detail.
 
----------------------------------------------
 
I will do some test and I intend to measure the free air resonance before gluing.
 
Greetings
Georg

 

[chinsettawong]

Hi Georg,
 
Please refer to the dimension of the inner tire diaphragm stretcher that I use.
 

 
I draw a few lines on the wood frame.  Before I pump in the air, I draw a line on the diaphragm directly on top of one of the lines I draw on the wood frame.  Then I pump in the air and observe how much the diaphragm elongates.  Anyway, this method can only serve as a very rough guideline.  I still have to do and redo many times to get the just right tension.
 
Wachara C.

 

[khbaur330162]

Could you devise a method for vibrating the diaphragm at a certain frequency and monitoring its resonance frequency? If your desired tension has a resonant frequency of (x) Hertz when excited by a vibration of (y) Hertz you only need to monitor the current resonance frequency (z) of the Mylar when excited by (y) and match (z) to (x) with respect to the aforementioned relationship. When this relationship is calculated and a methodology is perfected via the use of a relatively large tensioning table you can create many, many diaphragms of nearly the exact same tension all at the same time.

 

[GvTT]

Hello khbaur,
 
now you are whistling my tone!
 
I mentioned before in this thread that the tension of the diaphragm is directly indicated by it resonance frequency (eigenfrequency). Any impulse on the stretched membrane will activate a vibration in it's eigenferquencies, on a stretcher there will me more than one.
 
You could also use a loudspeaker with a tuneable output to make the membrane swing in resonance. But in this case you would need to monitor the amplitude (maybe with a laser?
 
BUT: this would only help you, if you intended to make many of the speakers. For setting up just one (as in DIY) I would not consider this relevant, as the size and shape of the diaphragm will be different for all the different designs.
 
We first need to define what we expect of  tension of the diaphragm:
- prevent the diaphragm to stick to a stator (number one for me) -> the speaker is statically stable
- an interaction of the resonance frequency with the lower audible spectrum while listening? (e.g. bass increase while driving the speaker in a half closed surrounding? -> way beyond my thoughts)
 
So I suggest to test the diaphragm while being on the stretcher (before gluing) for static stability.
 
I have not had an idea how to do this in a safe and reliable manor. My thoughts go along the lines of applying an increased bias (maybe 780V [580 bias + 200V amplitude]) to speaker components only pressed (but not glued) to the membrane. But is this, what we want? Unconditionally stable? (I guess, that would be good, considering aging, different amps, air pressure, taking phones off, etc.)
 
Once you know what you are aiming fore (stability wise) you could use the eigenfrequency as a reference, but only for this one specific speaker design.
 
Greetings
Georg