Debenham capsule directivity

[molke]

As a matter of fact, I’m already happy that my crude attempt actually passes any sound, let alone with a rather flat frequency response. The pattern/directivity is ‘optional’. An electrical problem would be very welcome as it seems easier to fix than a mechanical one. So I’m really grateful for your suggestions so far.

 

[soliloqueen]

yeah! that's what's so great about single-backplate designs like this and the similar k47. They're very insensitive to problems. They don't really care about diaphragm tension or anything. It makes them great DIY capsules.

 

[molke]

Just for comparison, this is for a cardioid-only circuit. The transformer is different, but otherwise, it is pretty much the same, except for the capsule.

 

[soliloqueen]

what about the other diaphragm. same issue?

 

[molke]

That would take some time. I’m a little concerned to change the connections too much since I followed the original plan, where the connection to the diaphragm is made with the same screw that holds down the clamping ring.

 

[soliloqueen]

That would take some time. I’m a little concerned to change the connections too much since I followed the original plan, where the connection to the diaphragm is made with the same screw that holds down the clamping ring.

for future reference, standard procedure is to connect both diaphragms to wires and change it at the solder joint instead in situations like this

 

[soliloqueen]

wait, you have both diaphragms wired, right? how are you connecting it to the multipattern Schoeps, and to the other circuits?

 

[molke]

The Schoeps circuit takes the signal from the backplate and polarization to front and rear diaphragm, as shown here. [RF filter C6/C7 is now 10 nF and Q1 is J113 since I couldn’t get the BF245 to work.]
For the AKG, the diaphragm is grounded and signal/polarization is respectively from/to the backplate. [Q1 is also J113 now.]
The FET847 takes signal from the membrane and applies polarization to the backplate. [Q1 is 2SK30]
In the cardioid-only circuits, the other diaphragm is unconnected.

 

[soliloqueen]

I was just meaning, if you use the back diaphragm in cardioid do you experience the same problem

 

[molke]

I’ll try that probably tomorrow. Since the tendency is the same for all three capsules in different circuits, I don’t expect much difference though.

 

[molke]

The other side seems to be worse, but apparently shows the same behavior. The level is now normalized to the average between 100 Hz and 10 kHz (before it was 40 Hz and 15 kHz), and the back level is referenced the same as the front. The 180° response posted before was just a scaled version of the front by mistake; this should be corrected here.

 

[soliloqueen]

Interesting. Maybe it is acoustic then. All your through holes are drilled all the way through right? The cardioid pattern is created by a phase delayed sound coming through the rear and interfering destructively whenever the sound comes from beyond a certain angle. Maybe you were right that the tension is so high that it's attenuating sound from the rear somehow. I'm not really sure on the mechanics of this. I'm still somewhat new to capsule design. I'll have to make one of these specific capsules at some point to see what can go wrong.

How are you tensioning the diaphragms exactly?

 

[molke]

The 13 through holes (0.7mm) are drilled all the way through.Drilling was surprisingly the easiest part for me.

The tensioning however didn’t work at all as planned. I have a rather simple jig which is about twice as large as the capsule (my lathe can’t fit any larger). The dirty lettuce foil was just for testing. Initially, I thought I could excite the tensioned diaphragm material with a small speaker and measure/tune the resonance frequency with another microphone to about 500 Hz, but I could not get any useful reading. So, to get at least some progress, I just relied on the weight of the steel jig for a half-way consistent tension. I did also try to get silica gel to bounce (something which was mentioned here somewhere), which did not work either.

Also, the membrane material I got is unfortunately not very flat/very wrinkled to begin with, which requires probably more tension than necessary to get it somehow flat. I gave up on finding rather unwrinkled spots at some time.

 

[soliloqueen]

I wonder if this is just a fundamental issue with the design of the capsule or if the stated specs are just wrong? I would love to get someone who has actually made one of these on the line. Because the k47 from which this design is probably somewhat derived has approximately the same dimensions and it has twice as many through holes of the same size...

My intuition tells me that 13 is not enough holes of this size to let enough through to properly cancel from the rear... The interference from the rear would be diminished by the inadequate amount of through holes and cause the capsule to be omni...?

 

[soliloqueen]

I just assumed because this capsule is so well known and I've seen it made a couple times that my immediate feelings about the amount of venting between the two sides must be wrong so I ignored that feeling and focused on trying to troubleshoot other things but maybe my initial thought was correct

Would you be willing to widen the holes to 1 mm and see what happens? Maybe they had runout without realizing it and their specifications for the capsule are based on the drill bit size. That would also make the blind holes about 1.2ish mm but you don't need to do that too. Just widen the through holes and see what happens if it's feasible and you're willing to do it.

 

[molke]

Maybe you’re right. The measured responses for different patterns (Fig. 10) at different angles (Fig. 9) at the end of the second article still seem to suggest there’s something wrong with my execution. I don’t know if anyone here who build it did measure off-axis response because on-axis seems fine.

Doesn’t the K47 capsule have just 30 through holes and also a lot more blind holes?

Would you be willing to widen the holes to 1 mm and see what happens?

Probably not with the semi-working omni capsules for now, but there’s still the rather thin left-over one.

As large capsules are readily available for DIY, I would be more interested in small diaphragm designs. I just chose this one because it was well documented and recommended by several people as a starting point.

 

[soliloqueen]

Yeah that was a typo I corrected it. Sorry about that!

Maybe it is something to do with the diaphragm tension. I'm not entirely sure but it seems to me like something is interrupting the sound from getting in through the back. What that thing is, would take a fair bit of troubleshooting.

I was actually offered an MXL small diaphragm capsule to model but I got so busy that I never sat down and did it. Maybe I should do that as it is kind of rare

 

[molke]

As it turns out, I will have to redo one of those capsules anyway. I never should have trusted the manufacturer in that isopropanol would be appropriate for cleaning acrylic (maybe it is, but not when there’s any tension at the same time). That’s the effect of a quick wipe – even the capsule holder just snapped.

A model (even an approximate one) for the MXL capsule would be great. All I’ve found so far are some pictures and generic drawings of the M50 and some AKG capsules.

 

[soliloqueen]

Yeah only soap and if you're really in a bad place hexane for cleaning acrylic iirc. I'm so sorry you were misled like that

I ran into the same problem when googling it. It literally says non-resistant to alcohols on the official data sheet... This article is straight up wrong. Dangerously so.

 

[molke]

Surprisingly, after taking the ‘cleaned’ capsule apart, it does hold together very well (until it doesn’t).

It’s probably a shot in the dark, but instead of guessing, I tried to plot the effect of changes in compliance (to 0° and 180° response at 30 cm) according to the admittedly rather simple model given in a Shure paper on “Unique Directional Properties of Dual-Diaphragm Microphones”. Starting point are the values given there for C_d (diaphragm compliance), C_a, and R_a (compliance and resistance of the acoustic phase shift network).

According to this model, increasing (or decreasing) diaphragm compliance by 20 % (middle row) won’t change much.
Obviously, increasing R_a by 20 % (blue curves) is going to reduce attenuation, but not that much.
The largest effect seems to be an increased C_a (too much cavity volume?), so I will try to reduce the depth of the clearance ring considerably (which now accounts for almost half of the volume of the blind holes) when redoing the acrylic body.