Diy Push-Pull symetrical undamped F8 capsule

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

kingkorg

Well-known member
Joined
Apr 15, 2017
Messages
2,913
Location
Norway
I got the idea from playing with original Sennheiser capsule.

https://groupdiy.com/threads/sennheiser-mkh800-mkh80-capsule-ks80.83049/
I am not aware of any true F8 single diaphragm large externally polarized capsule.

@jarvis said he was up to something similar so i thought i could make a video about modifying cheapo Ali RK12. @rogs was also after alternative push-pull capsule for his RF circuit.

https://groupdiy.com/threads/diy-rf-condenser-mics.71586/
It can also be used with regular polarization circuit.


The end result is as expected undamped (large sad face curve) frequency response that needs to be corrected with internal or external EQ.
20230813_150937.jpg20230813_150943.jpg20230813_150951.jpg20230813_150956.jpg

Here's a quick audio test in a Schoeps circuit, it is EQed in DAW to counter the sad face curve. Just one backplate of the two is polarized with 60v, diaphragm is at 0v. The other backplate can also be polarized to get the full effect and double the signal level. The signal noise ratio would improve.
View attachment Push pull capsule.mp3
This is by no means detailed DIY video, just a proof of concept that worked out great. Hopefully someone gets inspired by it. You can certainly see me struggling with this thing, improvising, trying to film while speaking non-native language...

When i'm done, i'll put some transparent sounding (probably silk screen) fabric in front of backplates to help with dust, and plosives.

 
Ah I get it now , single diaphragm , sandwiched between the two backplates .

Thanks again for going to all the trouble of producing the video for us ,
it looks all to easy to slip on the flat head screws and go straight through the diaphragm .
 
Ah I get it now , single diaphragm , sandwiched between the two backplates .

Thanks again for going to all the trouble of producing the video for us ,
it looks all to easy to slip on the flat head screws and go straight through the diaphragm .
Yup, happened to me couple of times.
 
I just had a thought watching the video ,
If you lit the capsule from underneath it could help the line up and placement of the upper sections .
 
I describe the design of such a beast in my MicBuilders Files
Log In
You have to join.

... but I take my hat off to you kingkorg 😯

If you don't enlarge the holes, there will be more resistance and the response will be flatter .. I hope my single remaining brain cell remembers correctly :)

Hmmm. I see that one side will have a lot more spacing so not quite symmetrical. And the bigger cavity on that side may need bigger holes to avoid a resonance in the audio range.

Duu.uh! I need to watch the video more carefully. Equal capacitance on both sides means spacing is equal.

Have you measured back & front responses? You are getting this beach bum excited about capsule design again
 
Last edited:
I describe the design of such a beast in my MicBuilders Files
Log In
You have to join.

... but I take my hat off to you kingkorg 😯

If you don't enlarge the holes, there will be more resistance and the response will be flatter .. I hope my single remaining brain cell remembers correctly :)

Hmmm. I see that one side will have a lot more spacing so not quite symmetrical. And the bigger cavity on that side may need bigger holes to avoid a resonance in the audio range.

Duu.uh! I need to watch the video more carefully. Equal capacitance on both sides means spacing is equal.

Have you measured back & front responses? You are getting this beach bum excited about capsule design again
The responses are identical on both sides, but capacitance is different by couple of pF because one side is metalized and faced towards the backplate while the other side has couple of uM of foil in-between.

The responses are very, very sad face shaped.

I could have done it the other way where the sides now facing outwards go towards the diaphragm. This way i would get more metal surface facing the diaphragm, more output, and more damping. This way i have all the blind holes facing the diaphragm. As this was first time experiment i didn't want to bother with additional lapping to smooth out the surface where i widened the through holes. The sides that now face outwards. Hope it makes sense. Thanks for that link!
 
Kinkorg,

I remember seeing somewhere a cutout of Schoeps ingenious CM 66 3 pattern capsule with all mechanical switching. There was a different diagram for each pattern, IIRC. Try to find it—it will give you a very good idea how they achieve a good diaphragm damping with relatively open area to avoid resonances.

Best, M
 
Last edited:
Kinkorg,

I remember seeing somewhere a cutout of Schoeps ingenious CM 66 3 pattern capsule with all mechanical switching. There was a different diagram for each pattern, IIRC. Try to find it—it will give you a very good idea how they achieve a good diaphragm damping with relatively open area to avoid resonances.

Best, M
Thanks a milion!
 
Having less damping the capsule should have less noise caused by resistance.

Would it be good to make the sides facing the inwards concave?
You could do this using the method hand grinding/lapping telescope mirrors
 
Last edited:
Having less damping the capsule should have less noise caused by resistance.

Would it be good to make the sides facing the inwards concave?
You could do this using the method hand grinding/lapping telescope mirrors
Interesting, haven't thought about this.
 
Interesting, haven't thought about this.
Another option might be to make the backplates thinner (which I think would create less comb filtering of the sound passing through the less-deep holes), and add slightly thicker spacers between the backplates and the diaphragms. That would dampen the diaphragm less, but obviously it would change the capacitance as well...so I don't know how it would sound.
 
The responses are identical on both sides, but capacitance is different by couple of pF because one side is metalized and faced towards the backplate while the other side has couple of uM of foil in-between.

The responses are very, very sad face shaped.

I could have done it the other way where the sides now facing outwards go towards the diaphragm. This way i would get more metal surface facing the diaphragm, more output, and more damping. This way i have all the blind holes facing the diaphragm. As this was first time experiment i didn't want to bother with additional lapping to smooth out the surface where i widened the through holes. The sides that now face outwards. Hope it makes sense. Thanks for that link!
It looks like double-sided metalized film might be available and would solve this issue: 6micron Both side metallized PET FILM China Manufacturer
 
Another option might be to make the backplates thinner (which I think would create less comb filtering of the sound passing through the less-deep holes), and add slightly thicker spacers between the backplates and the diaphragms. That would dampen the diaphragm less, but obviously it would change the capacitance as well...so I don't know how it would sound.
Not really, because the close proximity, and being already thin enough, the comb filtering and resonances are above the audible range.

4mm - thickness of the half backplate would affect the frequencies of about 85Khz and above.

It looks like double-sided metalized film might be available and would solve this issue: 6micron Both side metallized PET FILM China Manufacturer
Thanks for the link. Ideally we would want the metal layer to be sandwiched between two layers of mylar. Don't know if this is even possible. Mylar provides insulation if the diaphragm hits the backplate, also helps a little bit with potential arcing. For Rog's RF circuit we actually need this kind of capacitance mismatch. In some other circuit, like traditionally polarized, perfectly matched sides would be desirable.
 
Having less damping the capsule should have less noise caused by resistance.

Would it be good to make the sides facing the inwards concave?
You could do this using the method hand grinding/lapping telescope mirrors
You do this by increasing the spacing.
Finding accurate different thickness shims probably easier than lapping a concave surface :eek:
We did use a lap for fine tuning our capsules.
 
Last edited:
For Rog's RF circuit we actually need this kind of capacitance mismatch. In some other circuit, like traditionally polarized, perfectly matched sides would be desirable.
eg Great Guru Baxandall's circuit. Not 'Perfect' matching but similar.

I think the REAL advantage of Rog's circuit is that he found 'easy to source' inductors to suit. GG Baxandall thought nothing of designing custom RF inductors or audio transformers ... as he goes back to the invention of RDF before the Battle of Britain

I'm still hoping to be lent a complete LDC mike with Rog's circuit to investigate its noise performance in detail :)
 
When you design such a capsule there are two main challenges to consider. Both of them are linear and predictable, so it should be quite straightforward to control them:

1. Obviously such capsule is way underdamped for linear response, so you will need to correct its response. The peak will be around the frequency of the diaphragm tuning resonance. Since the system is resistance controlled below that frequency you will get stiffness controlled response and above--mass controlled. The parameters are linear, so for the de-emphasis even a simple low shelf boost should do the job adequately. Of course, with DSP you could fine tune it very nicely.

In comparison to condenser cardioid capsules one of the main advantages here is a lack of chamber resonances with their peaks and other anomalies. Also, phase behavior should be quite uniform. Very good indications will be this capsule taking EQ very nicely.

2. The top frequency response is limited by the front-to-back distance traveled by the sound wave--effect very well known from ribbon microphones. Many years ago on one of the forums I wrote a little primer on its details, so will just re-post it here. For some reason the Italics are not working, so I am putting it into quotes:

"In order to understand how the ribbon microphone works, how its flat response is formed, and how this response is different from capacitor mic, it is important to know three concepts: 1) Resistance Controlled system, 2) Mass Controlled system, and 3) Forces on the diaphragm.

1) The best example of the resistance Controlled system is a condenser microphone capsule, where the diaphragm is tuned in the middle of the band (usually somewhere between 900Hz to 1500Hz, depending on a mic). That exhibits a huge (sometimes up to 60db) peak. Naturally, for the flat response we need to damp this peak, which is done by means of air cushion trapped between the diaphragm and back-plate. The amount of viscosity of this cushion is regulated by the certain size holes (or sometimes grooves) in the back-plate.

2) On the contrary, the ribbon microphones are tuned into the lowest frequency of the band, so actually their natural response FALLS with 6db8 rate.

So, where is the flat response in ribbons coming from?

In the "native" fig8 pattern the sound wave strikes the front of the diaphragm and creates acoustic pressure p1. Since the back of the ribbon is exposed, the same sound wave flows around the ribbon and pole pieces/magnet structure some distance (called acoustic path d) and creates some acoustical pressure p2 at the back of the ribbon. This results in pressure difference p1-p2, which in fact, is a driving force to move the ribbon.

For example, why there is a null at the 90 degrees polar response? Because the sound wave reaches both, front and back at the same time, so there is no pressure difference.

The interesting feature of this driving force is that it doubles with every octave, so the acoustical response of the ribbon actually RAISES with 6db8 rate. Now, remember that the Mass Controlled system naturally has a falling response? When we combine those two that gives us an overall flat response.

Now let's see what's going on on the extremes of the bandwidth.

1) Low end:

Obviously, the tuning frequency of the ribbon would determine the lowest response. However, in the real system there always will be a slit between the ribbon itself and magnet/pole, so because of the viscosity of the air in that slit below some certain point the system turns into the STIFFNESS Controlled one (the one, which defines true omni operation) and the response below that point rapidly falls--that's why it is not practical to tune the ribbon much lower that point.

2) Top end:

As we talked earlier, there is a distance d, which represents the path between front and back and obviously this distance can be translated into the wavelength.
As we talked, the driving force p1-p2 increases with each octave, but only to the point where the d represents 1/2 wavelength, because when the d reaches the full wavelength our driving force p1-p2 becomes ZERO. That is, in the condition when the wavelength of the signal becomes equal to the distance of the acoustic path d obviously the p2 will become equal to p1 (because it is 360 degree shift) and the ribbon won't be moving.


That's why if we know the acoustical path d and ribbon dimensions, it is very easy to calculate the top frequency response.

It is important to notice, the top response will also be somewhat affected by diffractions caused by magnet system cavity, but there are special graphs, which help to correct the calculations to the very high degree of accuracy."

Hope this helps,

Best, M
 
Last edited:
.... For Rog's RF circuit we actually need this kind of capacitance mismatch. In some other circuit, like traditionally polarized, perfectly matched sides would be desirable......
That is true for cardioid and omni - but not for figure of 8............
The best matched capsule I have here is one of Ari's 'Flat K47' .. That measures as 64pF each side - within 1pF.
I fitted it into one of my MultiPattern OPIC builds, and the calibration pot VR1 was set pretty much centrally, for the same amplitude output from each side.

So I tried the same capsule with a 'figure of 8' RF circuit. With this pattern, the bridge can be fully balanced, simply becasue the capsule forms 2 legs of the bridge, instead of 1. When the pressure from an audio stimulus forces one side to have slightly larger capacitance, the other side is decreased by a similar amount..... So the 'centre point' of the bridge moves rather like the 'wiper' of a capacitive potentiometer.

With a single sided output, one leg of the bridge is a fixed capacitor, so doesn't change in value.... If the bridge is balanced - or very close - there is the danger that the changing capsule capacitance value will pass through the 'centre point' of the bridge..... If that occurs, the polarity of the output signal can suddenly reverse - producing some very strange results!
That's why the bridge needs to be unbalanced enough so that cannot happen.

For figure of 8 that cannot occur of course - the value of both legs change simultaneously.

So -- my figure of 8 mic using a 'flat 47' should naturally have been pretty much a balanced bridge...... Except that it wasn't!
It did work quite well -- but there was still more RF carrier present than I had expected from a fully balanced brdige.

Reading Baxandall's notes yet again, he makes a very significant comment about the inductor in his balanced bridge version.....
"The output winding of the oscillator is bifilar, so as to obtain very tight coupling between the two halves and thus to ensure that the voltages at the
two ends will be very accurately in antiphase"


When I first encountered an alternative coil former, it was obviously necessary to 'sacrifice' a Spectrum coil to see exactly how it was wound. And that reveals that, not only is it not bifilar wound, but the centre tapped secondary is wound with 10 turns on each of 2 plastic former 'shelves'.
The primary (4 turns) is then wound on top -- but only around one of those segments.

So, this inductor is not finely 'balanced' in that sense....... It works well enough by selecting the value of the bridge capacitor to create an almost balanced bridge, but where a fully balanced bridge is required - like in the case of a figure of 8 - it's not quite as simple as I'd hoped.

At the moment, I have returned to trying one of my high sensitive edge terminated 3micron capsules, which have a capacitive vlaue of around 90pF.
I'm trying a differently wound inductor, with a 6 turn primary wound onto the bobbin first - with 3 turns on each of two coil former 'shelves'.
The secondary is then wound over those windings, with 10 turns overwound onto to each of the same 2 'shelves'.
So the coil is physically wound to be more 'balanced'.

A preliminary test suggests a pretty quiet, high sensitivity mic..... But I'm still checking out the figures.....

I think there may be some futher 'mileage' in inductor experiments, using these formers....

One thing I'm still on the fence about is going fully balanced, like Baxandall suggested. Using an infinite impdence detector was not something that Baxandall considered (bit tricky - without any access to FETS of course! :) ). So he used bipolar transistors as his 'switches'.
Whether this approach would be any quieter than using a JFET self biased around Vp I wouldn't know?
What would be necessary to try out Baxandall's 'balanced bridge' approach would be more complex inductors -- which rather takes us back to where we came in with this project ! :)
 
Last edited:

Latest posts

Back
Top