[Title change] Mic capsule voltage oscillator - how to set the voltage increase?

The u87Ai and at least a couple clones of it (including the Golden Age Project one) multiply the voltage inside the microphone to hit 60V and 120V (backplate voltage). EDIT: Derp, those are +-60V

Can anyone know how this would be done? I'm not familiar with schemes that multiple DC voltage without chips.

Is it worth even considering since the voltage increase doesn't give that much more sensitivity?

Does it add much noise?

This diagram might be useful:

http://goo.gl/ZEuohE

It's a variation on a Hartley oscillator using 2 mutually coupled inductors (Instead of the usual center tapped inductor).
Oscillates at ~ 2 mHz. You can find this type of polarizing circuit in some older Schoeps microphones. (e.g. CMC-5)
The diodes and caps provide level shifting and rectification to create a voltage multiplier.
It's a simple and effective circuit.  Very weak output current though so - it's not meant to drive ANY resistive load - just the capsule itself.  Some el-cheapo China mics violate that rule in order to get a multi-pattern mic.  Bad idea! That's another topic though.

Cheers.

ZAPNSPARK

zapnspark said:

This diagram might be useful:

http://goo.gl/ZEuohE

It's a variation on a Hartley oscillator using 2 mutually coupled inductors (Instead of the usual center tapped inductor).
Oscillates at ~ 2 mHz. You can find this type of polarizing circuit in some older Schoeps microphones. (e.g. CMC-5)
The diodes and caps provide level shifting and rectification to create a voltage multiplier.
It's a simple and effective circuit.  Very weak output current though so - it's not meant to drive ANY resistive load - just the capsule itself.  Some el-cheapo China mics violate that rule in order to get a multi-pattern mic.  Bad idea! That's another topic though.

Cheers.

ZAPNSPARK

Click to expand...

Thanks!

http://store.studio939.com/product/u87-ai-booster-conversion-pcb

Pip said:

http://store.studio939.com/product/u87-ai-booster-conversion-pcb

Click to expand...

Oh, thanks, I missed that. That's perfect.

zapnspark said:

This diagram might be useful:

http://goo.gl/ZEuohE

It's a variation on a Hartley oscillator using 2 mutually coupled inductors (Instead of the usual center tapped inductor).
Oscillates at ~ 2 mHz. You can find this type of polarizing circuit in some older Schoeps microphones. (e.g. CMC-5)
The diodes and caps provide level shifting and rectification to create a voltage multiplier.
It's a simple and effective circuit.  Very weak output current though so - it's not meant to drive ANY resistive load - just the capsule itself.  Some el-cheapo China mics violate that rule in order to get a multi-pattern mic.  Bad idea! That's another topic though.

Cheers.

ZAPNSPARK

Click to expand...

Bumping this thread to ask a few questions. I tried reading the Wikipedia article on Hartley Oscillators, but I couldn't quite match up what it was telling me with this circuit.

AudioImprov has several posts with a very similar oscillator:
http://audioimprov.com/AudioImprov/Mics/Entries/2013/8/8_Mic-Parts_RK-47_files/RK-47%20mic%20schem002.jpg

1) The source voltage on the linked oscillator is 6.8V, and 9.1V in the AudioImprov circuit. If I *just* want to double the voltage, to go from say 30V to 60V, what would I change in this single-transistor oscillator? I obviously don't want to end up with enough voltage to collapse the capsule. I know there's the u87Ai's method but it's got a lot of extraneous parts and the simplicity (plus variability) of the single-transistor method is really appealing to me. Or more to the point, how do I translate the LC filter/oscillating frequency to a DC voltage?

2) You said it's not meant to drive any resistive loads and that it's a bad idea to use it for multi-pattern mics. I don't see another resistor after the "capsule polarization" point, but R1 is 10M. If I go waaaay bigger, like 100M or 1G, is it in much better shape for multipattern uses? Audio improv actually goes lower with the resistance there but I don't know how it works out in practice. (I've had decent enough performance with only 1.3M to the + voltage in other mics as long as it was really well filtered.) Is there something else that makes the U87Ai's method that much more stable? It's certainly not el cheapo. Were you talking about methods other than decoupling and rectifying it twice?

3) Someone in another thread about hartley oscillators said that "zeners are noisy." I've encountered this of course but haven't used zeners in microphones. Is there any way to eliminate the zeners from this (perhaps with a bias trimmer somewhere), or is just filtering caps in parallel with and RC filters after them always going to be the more efficient method?

Ehrm... Nope, the U87Ai circuitry produces +60v and -60v

I've yet to see or hear of a phantom-powered mic that has an oscillator outputting over 100v...

midwayfair said:

The u87Ai and at least a couple clones of it (including the Golden Age Project one) multiply the voltage inside the microphone to hit 60V and 120V (backplate voltage).

Click to expand...

Re: oscillator arrangements: https://youtu.be/WTJhIVIGvSU?t=736 - something like that miiiiight be what's inside my NT2-A. I recall seeing a SOIC14 or SOIC16 chip inside my Shure KSM27 as well, with a host of diodes & caps around it.

midwayfair said:

AudioImprov has several posts with a very similar oscillator:
http://audioimprov.com/AudioImprov/Mics/Entries/2013/8/8_Mic-Parts_RK-47_files/RK-47%20mic%20schem002.jpg

1) The source voltage on the linked oscillator is 6.8V, and 9.1V in the AudioImprov circuit. If I *just* want to double the voltage, to go from say 30V to 60V, what would I change in this single-transistor oscillator? I obviously don't want to end up with enough voltage to collapse the capsule. I know there's the u87Ai's method but it's got a lot of extraneous parts and the simplicity (plus variability) of the single-transistor method is really appealing to me. Or more to the point, how do I translate the LC filter/oscillating frequency to a DC voltage?

2) You said it's not meant to drive any resistive loads and that it's a bad idea to use it for multi-pattern mics. I don't see another resistor after the "capsule polarization" point, but R1 is 10M. If I go waaaay bigger, like 100M or 1G, is it in much better shape for multipattern uses? Audio improv actually goes lower with the resistance there but I don't know how it works out in practice. (I've had decent enough performance with only 1.3M to the + voltage in other mics as long as it was really well filtered.) Is there something else that makes the U87Ai's method that much more stable? It's certainly not el cheapo. Were you talking about methods other than decoupling and rectifying it twice?

Click to expand...

Answer to 1) - it's right in that linked picture What does it say to the right of that variable resistor that's above the 2N5551 in the middle?

2) That series resistor, in conjunction with the capacitors before and after it, forms an RC low-pass filter (to clean up the bias voltage going to the capsule)

Khron said:

Ehrm... Nope, the U87Ai circuitry produces +60v and -60v

I've yet to see or hear of a phantom-powered mic that has an oscillator outputting over 100v...

Click to expand...

Here you go my friend

Works really good in better circuit!

Yes, this is a very common circuit used in several Chinese microphones!

[Posting before I read the last two replies]

Khron said:

Ehrm... Nope, the U87Ai circuitry produces +60v and -60v

Click to expand...

Oh, yeah. I know that now. My original post was stupid.

Answer to 1) - it's right in that linked picture What does it say to the right of that variable resistor that's above the 2N5551 in the middle?

Click to expand...

I totally understand how to adjust the voltage of his exact setup. However, I was wondering what I change in the oscillator itself to change how much it increases the voltage by.

The reason is that I'd prefer to be able to port the same oscillator to different designs that don't already have a convenient 9V (or 6V) point, and maybe also don't have room on the layout to add one (humor me).

So if I had to use, say, a 30V or even 48V point as the input voltage, I would need a regulator to drop the voltage to the appropriate level (any reason I shouldn't use a 9V regulator? ... It's cheaper and smaller than a diode+transistor+resistors+extra filter cap but I never see them ...) on the oscillator board itself.

So let's say I want to avoid that step. Can I just change, say, one of the inductors or capacitors when I plop it into a different mic and know roughly how much voltage I will get out based on the oscillating frequency and the voltage I choose to use as the supply voltage for the oscillator?

Again, my disconnect is knowing how the LC filters and frequency affect the voltage increase, not necessarily that I don't get the basic principal of how the oscillator works.

ln76d said:

Here you go my friend

Works really good in better circuit!

Click to expand...

I recognize that one from the CAD circuit. Figure 8 is set to 30V, cardiod is the full 60V, right?

The figure 8/omni modes I made for my Aurycle build had similar switching - though that one was still only 48V.

I should change the title of the post.

Ever considered increasing the value of that series "input" resistor? ;D That's even simpler & cheaper than any (other) sort of regulator or whatever, isn't it?

I couldn't tell you how this works either, but i'm reasonably sure it's quite directly related to the values of the two inductors (and/or the ratio?), and/or the coupling between them.

midwayfair said:

The reason is that I'd prefer to be able to port the same oscillator to different designs that don't already have a convenient 9V (or 6V) point, and maybe also don't have room on the layout to add one (humor me).

[...]
So let's say I want to avoid that step. Can I just change, say, one of the inductors or capacitors when I plop it into a different mic and know roughly how much voltage I will get out based on the oscillating frequency and the voltage I choose to use as the supply voltage for the oscillator?

Click to expand...

RuudNL said:

Yes, this is a very common circuit used in several Chinese microphones!

Click to expand...

Exactly and in fact it's slightly modified version of modified schoeps converter used in many chinese microphones.

midwayfair said:

I recognize that one from the CAD circuit. Figure 8 is set to 30V, cardiod is the full 60V, right?

Click to expand...

I don't remember what values were used in several chinese microphones, but i build that type of converter to use with higher voltages.
Anyway even if it would be used with 30V/60V, then cardioid should be 30V and figure of eight 60V.
Or just fig 8 need higher voltage.
Look at schematic, R16/R17 form voltage divider for backplate polarisation, then look at the switch SW3.
For cardioid SW3 setup, R15/R19 form divider for back diaphragm polarisation - now backplate and back diaphragm have the same potential, which is true for cardioid pattern with that type of input toplogy.
For fig 8 SW3 have no connection so on the back diaphragm polarisation voltage is doubled in reference to the backplate voltage - it's correct for 8 pattern.
I didn't checked omni option, looks strange like a short, but maybe it works
Am using that type of converter with "true" cardioid, omni without connection to the polarisation voltage or ground so the only polarisation voltage at the back diaphragm is for fig 8.

Anyway higher polarisation voltage can be done same way as in schoeps or chinese "schoeps" converter.
Adding pot like in original schoeps circuit is usefull but also you can rise zener diode voltage value.
Note that the main difference between these converters is that the cad have two additional diodes for multiplier

ln76d said:

I didn't checked omni option, looks strange like a short, but maybe it works

Click to expand...

It just grounds the rear's "supply voltage". There's a 1G between the capsule and ground and 51M between the capsule supply and ground. The 51M isn't going to draw any meaningful current from the capsule supply, and having the 1G there is just convenient for the switching. You could shuffle components around so that you switch the rear between + and ground and the center to 1/2 voltage in Figure 8 and it's one less cap and one less large value resistor. That's how I did it (running on just 48V), but you end up needing to wait for a filter cap (the equivalent of C16) to drain when switching modes.

EDIT: I was still a little unclear on something.

Khron said:

2) That series resistor, in conjunction with the capacitors before and after it, forms an RC low-pass filter (to clean up the bias voltage going to the capsule)

Click to expand...

Again, would there be any benefit or need to putting another VERY large resistor (100M, 1G, whatever) after the 1.5M, before the voltage reaches the capsule (this is what the u87Ai does), or would it be better (or just as good) to just increase the size of that 1.5M? I understand that the 1.5M/10nF is an RC filter for the bias voltage. This is a case where it's extremely hard for me to experiment -- assuming I could even measure the difference in noise or whatever. I don't want to solder and desolder parts repeatedly if there's a clearly best solution.

midwayfair said:

It just grounds the rear's "supply voltage". There's a 1G between the capsule and ground and 51M between the capsule supply and ground. The 51M isn't going to draw any meaningful current from the capsule supply, and having the 1G there is just convenient for the switching. You could shuffle components around so that you switch the rear between + and ground and the center to 1/2 voltage in Figure 8 and it's one less cap and one less large value resistor. That's how I did it (running on just 48V), but you end up needing to wait for a filter cap (the equivalent of C16) to drain when switching modes.

EDIT: I was still a little unclear on something.

Click to expand...

I did it different way, because i don't like grounded capsule topology.

midwayfair said:

Again, would there be any benefit or need to putting another VERY large resistor (100M, 1G, whatever) after the 1.5M, before the voltage reaches the capsule (this is what the u87Ai does), or would it be better (or just as good) to just increase the size of that 1.5M? I understand that the 1.5M/10nF is an RC filter for the bias voltage. This is a case where it's extremely hard for me to experiment -- assuming I could even measure the difference in noise or whatever. I don't want to solder and desolder parts repeatedly if there's a clearly best solution.

Click to expand...

Or 1M? At least my schematic shows 1M instead of 1.5M, but there were different schematics.
Anyway, here you have schematic where i removed all components which aren't used in U87Ai but populated on the board (same board as for U89 which use these extra parts for other patterns).
Now it should be better visible
As you can see 1M/10nF form RC filter and then 1G/100nF form another RC filter.
You could just use 1G/100nF and see is it enough.

Couldn't get the u87Ai oscillator to ... well, oscillate, but I did have good success with the trimpot. (Probably just misread something on the schematic or had a mistake on my perfboard layout.) I was able to take the regulated 33V from the emitter and used a 25K trimpot to set the capsule voltages to +-60V. I sort of mashed up the designs a little to get the extra filtering from the CAD and U87 schematics but kept most of the values from the Audio Improv.

I am a little nervous that it goes all the way up to 120V, so tweaking the pot could cause the diaphragm to suck in if I'm not paying attention and twist it the wrong way, but it should be fine for my own uses and I can just set the capsule voltage before soldering. If I do an actual PCB layout I will probably put a limiting resistor to keep it to 80V or lower.

Also realized, before soldering, that a 9V TO-92 regulator for the single-transistor oscillator was a bad idea coming from a 47V source voltage!

midwayfair said:

Couldn't get the u87Ai oscillator to ... well, oscillate, but I did have good success with the trimpot. (Probably just misread something on the schematic or had a mistake on my perfboard layout.) I was able to take the regulated 33V from the emitter and used a 25K trimpot to set the capsule voltages to +-60V. I sort of mashed up the designs a little to get the extra filtering from the CAD and U87 schematics but kept most of the values from the Audio Improv.

Click to expand...

What values of chokes did you used for U87Ai oscillator?

midwayfair said:

I am a little nervous that it goes all the way up to 120V, so tweaking the pot could cause the diaphragm to suck in if I'm not paying attention and twist it the wrong way, but it should be fine for my own uses and I can just set the capsule voltage before soldering. If I do an actual PCB layout I will probably put a limiting resistor to keep it to 80V or lower.

Click to expand...

I'm rather fan of simplifying the circuits than adding extra parts
Yes as you wrote, you can set polarisation voltage without capsule connection.
Anyway i would find more optimal trimmer value and do voice test to find sweet spot of the capsule
If you use 32mm capsules then these usual works pretty good with voltages over 100V.

I used two 100uH in series because I didn't have 220uH for the u87ai. I doubt it was them, the frequency should have been similar.

I'm using the group buy capsules, I'll have to see what their max is but I don't think they can take 120v

midwayfair said:

I used two 100uH in series because I didn't have 220uH for the u87ai. I doubt it was them, the frequency should have been similar.

Click to expand...

Probably, didn't tested different values but rther asked out of curiosity

midwayfair said:

I'm using the group buy capsules, I'll have to see what their max is but I don't think they can take 120v

Click to expand...

In standard connection with backplate polarisation rather not but polarising diaphragm like for multipattern i used both (k47 and k12) up to 150V for figure of eight.

Also i used k47 in schoeps topology for cardioid pattern with 75V - some works ok with backplate polarisation, some not, but all worked with reversed polarisation where signal was taken from backplate and polarisation voltage was applied to the diaphragm.