Dumb idea for a mic circuit?

[OneRoomStudio]

I was thinking about the limitations of phantom power the other day, and I wondered what kinds of solid state mics would be designed if power weren't a consideration. One idea that occurred to me was to build a cascode circuit using bipolar 60V. The nice part about using +/- 60V is that it would make polarizing the capsule a piece of cake. The backplate could be grounded, +60 could be applied to the front capsule, and the rear capsule could be switched between +60V, 0V, and -60V for omni, cardiod, and figure-8 (respectively). You could even add some voltage dividers for additional patterns. Anyway...here's the circuit I was thinking of. I simulated it, and it seems to be solid. Thoughts?

 

[tk@halmi]

Seems like a lot of voltage across those two transistors. What current do you get through Q1?

 

[radardoug]

I was thinking about the limitations of phantom power the other day, and I wondered what kinds of solid state mics would be designed if power weren't a consideration. One idea that occurred to me was to build a cascode circuit using bipolar 60V. The nice part about using +/- 60V is that it would make polarizing the capsule a piece of cake. The backplate could be grounded, +60 could be applied to the front capsule, and the rear capsule could be switched between +60V, 0V, and -60V for omni, cardiod, and figure-8 (respectively). You could even add some voltage dividers for additional patterns. Anyway...here's the circuit I was thinking of. I simulated it, and it seems to be solid. Thoughts?

View attachment 140825

Power is not a consideration. The phantom supply gives plenty of current and voltage to build mikes. You are not powering a power hungry device.
And if you dont like phantom, there are plenty of ways once you build a dedicated supply.

 

[OneRoomStudio]

Power is not a consideration. The phantom supply gives plenty of current and voltage to build mikes. You are not powering a power hungry device.
And if you dont like phantom, there are plenty of ways once you build a dedicated supply.

Right, the point is that with a dedicated supply, +/- 60V would be convenient for capsule polarization…so the design was based on that.

Seems like a lot of voltage across those two transistors. What current do you get through Q1?

It pulls around 12.5mA, which is on the high side, but within the spec of the LSK170A (20mA max). There ends up being around 32.5V at the BJT collector, which again, it high-ish, but within spec.

 

[tk@halmi]

I see so the 4.7K resistor drops almost all of the negative rail voltage.

Make sure to use at least 1W resistors.

 

[k brown]

With a slightly higher trafo ratio, could C1 be eliminated?

 

[radardoug]

There are many ways to skin the cat, but putting +/- 60 volts on those two transistors makes no sense. Also, your cap values dont make sense. If you are doing a dedicated supply, +/- 15 seems more appropriate for the amp.

 

[radardoug]

With a slightly higher trafo ratio, could C1 be eliminated?

Those two are in no way connected.

 

[k brown]

Those two are in no way connected.

Doesn't bypassing the source resistor increase gain?

 

[OneRoomStudio]

There are many ways to skin the cat, but putting +/- 60 volts on those two transistors makes no sense. Also, your cap values dont make sense. If you are doing a dedicated supply, +/- 15 seems more appropriate for the amp.

The cap values were set large enough that they wouldn’t affect the frequency response for simulation purposes. C2 could go as low as 1uF maybe, without losing too much low end, and C1 could go down to maybe 4.7uF or could be removed, as @k brown mentioned. Removing it would improve linearity but reduce gain.

Since the output transformer is a step-down, the ratio would have to go down with C1 removed to retain a similar output level. Luckily, the output impedance of this circuit is pretty low and you could probably get away with a 3:1 or maybe even 2:1 OT.

 

[abbey road d enfer]

Thoughts?

It's necessary to add a capacitor at the input for proper simulation.
Without it, the circuit is improperly biased.
With the capacitor, the simulation show that Q1's Vds is only about 0.5V. I don't think it's right.
It shows in the fact that the circuit distorts heavily at about 300mV, which is paradoxical for a circuit powerd with +/-60V rails.
I agree that it may be good enough for typical capsule levels, but for me it indicates inadequacy.
Proper cascode operation requires a different biasing of the FET.

 

[abbey road d enfer]

With a slightly higher trafo ratio, could C1 be eliminated?

Q2's collector sits at about 35V. No xfmr, whatever the ratio should operate with that much DC.

 

[k brown]

Q2's collector sits at about 35V. No xfmr, whatever the ratio should operate with that much DC.

I think you're tallking about C2.

C1 is the source bypass cap I referred to.

 

[abbey road d enfer]

I think you're tallking about C2.

C1 is the source bypass cap I referred to.

Aah, ok.
Without C1, the circuit's gain decreases from about 20dB to minus 20dB.

 

[abbey road d enfer]

Luckily, the output impedance of this circuit is pretty low and you could probably get away with a 3:1 or maybe even 2:1 OT.

The circuit's output impedance is almost entirely defined by Q2's collector resistor. It's the standard constant-current behaviour of a cascode. The output signal is largely affected by the load that's applied.

 

[jp8]

Adding an external supply only adds costs and complexity to the system, while there are plenty of good polarization voltage generator designs available. I don't see any advantage using an external supply to that end.

As to your design: consider adding a feedback capacitor from Collector to Gate, creating a KM84 style charge amplifier circuit. The gain will depend less on the gm of the JFET, you'll get lower output impedance, lower THD from JFET and transformer and extended LF range.

Jan

 

[thor.zmt]

I was thinking about the limitations of phantom power the other day, and I wondered what kinds of solid state mics would be designed if power weren't a consideration. One idea that occurred to me was to build a cascode circuit using bipolar 60V.

....

Anyway...here's the circuit I was thinking of. I simulated it, and it seems to be solid. Thoughts?

View attachment 140825

Other commented, I agree utter waste of the 120V input voltage, for ~ 30V on the load resistor.

The 2SK170 is not a good Fet for microphones. The cascode is "phunky" and the biasing suspect.

You could try one of the "usual suspects" with a Rush Cascode using a PNP for biasing and a second PNP for a Sziklai circuit, then a pair of E-102 CRD to give 2mA load instead of the resistor and then your transformer .

Honestly, I think the exact opposite direction is a better choice.

The classic "Schoeps" circuit is an excellent starting point to 'amp it up'. As P12/P24 are completely history, we can aggressively design for P48 (or even more Volz).

The output Bipolars can be changed to P-MOS, removing many issues in the circuit. For a PMOS we can set the voltage across the FET with MOhm resistors, so very low value and high quality coupling capacitors will work great.

A P-MOS & NMOS Sziklai circuit can make VERY low distortion cable drivers in this position, but currents in the PMOS start getting low. PNP at 100uA and N-Channel MOS might be better for this. The PNP+NMOS Sziklai will also increase the transistors input impedance by the gm of the mosfet (100's of mA/V) so again we see many MOhm and can again use low value film coupling capacitors

The circuit could be made to run on "P68" (68V via 6k8 per polarity) as option over P48.

At which point we get (say) 60V on the XLR and can use that to polarise the capsule and 2.35mA to run our Mic-Frontend.

That's enough for a J-Fet + P-MOS Szikai which can run at (say) ~ 24V giving around 4V maximum out at low distortion.

Running the same Mic on P48 (needs a P48/P68 switch) drops the polarisation to 40V which looses around 3dB SNR (probably Brownian noise of air molecules on the Diaphragm will limit SNR anyway) and sensitivity, possibly a good trade.

Ok, multi pattern polarisation is still hard, but why not build a few more mic's anyway?

Otherwise, why not directly integrate an A2D and Wifi audio module?

Thor

 

[jp8]

Right, the point is that with a dedicated supply, +/- 60V would be convenient for capsule polarization…so the design was based on that.


It pulls around 12.5mA, which is on the high side, but within the spec of the LSK170A (20mA max). There ends up being around 32.5V at the BJT collector, which again, it high-ish, but within spec.

According to the LSK170 datasheet from Linear Systems, the Idss range of the LSK170A is specified as 2.6 - 6.5mA. If any LSK170A that you buy should work in this circuit, design for Id < 2.6mA. And keep in mind that your simulator model is probably a typical JFET, not a worst- or best-case part. I wish manufacturers would supply worst-, typical and best-case models, but they seldomly do...

Model the transformer as closely as possible to the actual type that you intend to use and add a load resistor that equals your mic pre input impedance. You may find a resonant peak from C2 + transformer inductance that can become audible. In that case, increase C2, but then you'll pass subsonic signals, which in my opinion you should not. Or add a damping rssistor in series with C2. Or take a large C2 (e.g. 22 or47uF), followed by a parallel RC circuit. Select the capacitor value such that you have 20Hz low-cut frequency, then vary the parallel resistor until the resonant bump dissapears.

Jan

 

[tk@halmi]

There are several good circuits out there to create dual 60V polarization from phantom power. There is a bit of investment to create the polarization PCB that fits your mic body, but it is still cheaper than an extra enclosure that uses the mains power. From a cost perspective this solution ends up to be a far more expensive than the onboard voltage multipliers.