Townsend Labs Sphere Hacking

[kingkorg]

It is capsule from a Dave Pearlman mic.

 

[rmaier]

Kind of amazing that your 797/Townsend simulation hangs with the Pearlman (along with your other comparisons).

 

[kingkorg]

Ok, i've spent some time with AA Polar designer, and i have to say it is stunning. The one thing it does amazing is when dual out mic is used for snare drum, and hi hat bleed needs some extra control.

Naturally cardioid rejects most of  HH bleed. But as i usually boost snare around 8K, that bleed starts to sound unbearably thin, and piercing. So what you can do with AA Polar Designer is to switch low mid pattern control to OMNI, or F8, and that way you bring back direct signal from rear diaphragm in that range  and HH gets way warmer.  In a sense, the mic becomes both snare and HH mic.

Instead of gating snare one can embrace the bleed, control it, and use it creatively.  The same goes for any other instrument, or unwanted ambience coming into vocal mic.

 

[mickdundee63]

I'm interested in trying the post production polar patterns. I knocked up a dual output microphone but it's not working, wonder if you have any tips kingkorg or anyone else?

I have a chinese u87 type mic and I bought a second amp board to stuff in there, then swapped the 3 pin xlr for a 5 pin xlr, wired up both boards to the 5 pin which terminates with two 3 pin xlrs, one for each board.  Both amps are wired to either side of a k67 capsule and they share the same backplate connection.

I'm getting output from both sides BUT each channel is giving me the sum of both sides. I.e. I get both sides of the capsule at a similar level in both outputs. I'm definitely getting both sides of the capsule because I can switch on either side (by applying or removing phantom power) and each amp only works with it's appropriate side of the capsule when it's the only amp turned on. But when both are turned on, both my xlrs are outputting both sides of the capsule.

I noticed you said in your other thread that the backplates don't need to be electrically isolated so not sure what to try adjusting. Any thoughts??

 

[mickdundee63]

This is the amps I'm using btw, if it helps:

https://detail.1688.com/offer/565714684246.html

 

[mickdundee63]

Oh and I'm getting the same signal in both xlr's, the phase cancel each other

 

[mickdundee63]

The orignal capsule in the mic was one sided, or at least only one side had a diaphragm wired. So maybe the circuit is omni by default because it wouldn't matter for a cardiod only setup? So two omni signals that are the same? But how is each side getting the omni signal when each side is only connected to one half of the capsule? I am baffled...

 

[kingkorg]

How are the capsules polarized? Are backplates polarized, or diaphragms?

I can't see what is going on with the circuit, the page you shared is all in Chinese.

 

[mickdundee63]

I will ask the vendor for a schematic. I thought that because it's a cardiod only mic there might be no polarization but you're right I probably need a schematic to get this figured out

 

[rmaier]

Sounds like the same problem I ran into using a pair of Pimped Alice boards, where the schematic called for a polarized backplate. I ended up reversing the polarity of both capsule sides and everything works as expected.

 

[mickdundee63]

Rmaier you're exactly right! Swapped the polarity and all working correctly now. You made my day. Kingkorg your correction curve is also working great despite the different circuits and capsule I used. My mic flat and then the u87 modeling with the correction curve applied before sound very similar so I'd say it's in the ballpark. Enough to get some interesting tones out of the modeling anyway. This should be fun!

 

[kingkorg]

Glad you resolved it. My circuit polarizes the diaphragm, so i guess it doesn't work the other way around.

I look forward to hear how are your thoughts on the thing.

 

[mickdundee63]

Yep yep, my friend has a Townsend sphere so see if I can do some comparisons

 

[kingkorg]

Well the best thing you can do is to measure his mic against yours and come up with exact curve for your capsule. Just one side is enough, cardioid, without Sphere VST.

Just set them at exact same spot in front of a speaker and play pink noise. Lets say 40 cm away from the speaker. Then use something like CurveEQ from Voxengo to measure the difference in responses. The curve you get is the one you should use instead of mine before the plugin.

Sadly i don't know anyone in Norway who has this mic, otherwise it would be a child's play to clone it.

If he allows you, you can as well measure the response of the circuit by injecting a signal into capsule/pcb connection spot.

 

[rmaier]

Hey mickdundee63, glad it's working. Please let us know if you have a chance to compare it with your friend's mic.

 

[mickdundee63]

I think he might be reluctant to let me under the hood for warranty reasons but I will ask him and I will definitely try the eq matching

 

[mks]

Thanks for your work on this. Just to learn a little more, I wanted to ask:
- is this curve applied to the raw capsule output WITHOUT a high frequency compensation circuit?
- if so (when compared to the frequency response of a K67 with a U87 compensation circuit), the curve looks like it is an attempt to further flatten the frequency response that the capsule would output. Does that seem correct to you?
- have you tested the frequency response of your mic with this curve to see how flat it is? it would be interesting to see if Townsend's effort is to start with as much of a flat response as possible. That certainly would simplify things I'd imagine.

Thanks

 

[kingkorg]

Well in analog domain it is quite difficult to create an eq curve to flatten out k67 and fit it inside the mic, and operate with phantom.

Original Sphere with built in filter is not flat. It is k67 + the curve i provided. Actually, very close to u87 sound, but not quite the same.

The curves of each emulation in software include complex correction curve that aditionaly flatten the response, and apply mic curve that is modeled.  It is the same thing with Slate mic, it is not flat even they claim so. That one has edge terminated capsule from the same manufacturer. Slate's curves include also correction before modeling.

The analog filtering found in the mic is a safety net. Protection key if you will. That prevents you to stick in say Behringer B2 which has the same exact capsule and find out it sounds exactly the same minus stereo outputs.

With that filter the mic sounds more ballanced, and usefull without software.

Analog filter allows to fine tune mics if capsule response changes during production life.

It behaves much like de-emphasis network in u87 - adresses some of the noise.

It serves as a part of modeling for free field models found in software. Explained in Townsend's pattent document.

 

[mks]

Thanks kingkorg. Thanks for clarifying that this curve is for a raw capsule output and not already u87 filtered. I assumed since that would quite the aggressive compound adjustment

I'm a little confused though with some of the points in this thread. Particularly mention of the diffuse field model as comparison. Pardon my simpleton-ness I'm also not presuming to understand the internal workings of their DSP.  It seems to make a mic that would behave in a similar way with respect to emulations, we'd want to have the frequency response of the hardware to be similar to the original. In this case, we can see what that is by analyzing the "Linear" model since that would show us the deviation from Linear. Rather than the diffuse model which would have an expected additional tweak to simulate if the mic had been designed as a diffused field device. The question is does a K67 797 capsule with the standard U87 filter curve look like (the inverse of) the Linear model, or is it different? It seems another alternative would be the dual capsule kit by microphone-parts.com that apparently has a different curve aiming to yield a more flat output. However, if that's true,  then that would bring us further from the expected input (which is clearly not flat based on the "Linear" model analysis). Just curious which hardware configurations would bring us the closest. And yes, it's entirely possible they've engineered a skewed filter circuit as a form of copy protection.

It would also be interesting to see you test your capsule + software filter + "Linear" model to see how close to flat the output got. Rather than comparing to other emulations and hardware (which has their own inconsistencies). No worries if you don't have time, but this is all really interesting to me

p.s. image attached of curve applied to linearize the mics natural frequency response (ie.obviously the inverse of the frequency response)

 

[kingkorg]

I did this in the first place. Here is the image of my sphere mic in flat mode, including correction curve i provided against calibrated Beyerdynamic measurement mic.  Blue is Beyerdynamic. Naturally it has slightly less low end, due to proximity effect of Sphere which is in cardioid.

Why not measure Sphere in omni mode and compare? Well, in that case omni forming is way different than true omni capsule of Beyerdynamic, it is achieved in different manner, and it makes a lot of trouble. So it's just better to take this route. You have to trust me on this. It has to do with capsule physics.  Omni capsule and k67 set in omni are nothing alike.  Two membranes separated by almost a cm + backplate = phase issues, against one membrane with no phasing issues.

I dont see the point in using Microphone Parts capsule, 797 capsule is superior in every aspect, readily available and inexpensive. Yes it is slightly  brighter, but in this case it is a good thing. Better effect of emphasis/de emphasis  noise reduction. Better response above 15K.

I've taken apart somewhere around 40 various mics with this capsule, Studio Projects, Behringer, Rode, new capsules from factory, all different ages - 20 years!!! They all measured within 1db in frequency response, and couple of pF in difference.

I am sure all of this is the reason for Townsend and Slate to go with 797 audio.