A very versatile input stage for SDC mics

[kingkorg]

which likely more than the capsule would ever produce.

Capsules produce way more than people give them credit for. This is a plosive into a SDC vocal mic with no protection, and the diaphragm didn't even hit the backplate. 3.2Vpp will be enough for most "normal" conditions, but i wouldn't use these on close miking loud sources, or live vocal recordings. Yes, that's 22.2Vpp smal diaphragm mic polarized at 40V.

 

[Cqwet Dbdfte]

Very impressive!
This follower circuit does not look to able to handle 22Vpp in/ out.

 

[abbey road d enfer]

The transformer inductance adds some impedance isolating the power source, if this is a center tapped transformer not much AC signal survives the R7/C7,

Think it over; if the 48V source has a resistance of 3.4kohm, the signal develops from J1 though R8, L1, L2 and R7.
The problem with your simulation is that L2 is shorted through V2, R7 and C7.
You should really apply phantom via two 6.8k resistors.

The circuit is single ended, boot strapped voltage follower/buffer for the input capsule. No gain. It starts clipping at 3.2Vpp (with a 1k load) , which likely more than the capsule would ever produce.
I have no number handy for the transformer inductance. I used 8H, may not matter if large enough.

That's a pretty good figure, close enough to most mic pre input transformers.

Not having 180° phase difference between the mic conductors makes that configuration lose some of its noise immunity?

No. Balance is a matter of impedance, not level. A signal can be perfectly balanced with zero signal on one of its legs.

Could some small transformer fit inside the mic?

You don't need to cascade two transformers.

 

[abbey road d enfer]

Very impressive!
This follower circuit does not look to able to handle 22Vpp in/ out.

Probably not, although it depends on the actual load impedance. Considering the 27k resistors and a typical mic pre impedance of about 2k, I would guess the actual clipping happens at about 2Vpp. Which is not a bad performance for common usage, but sure not enough for close micing of loud sources.

 

[MicUlli]

Probably not, although it depends on the actual load impedance. Considering the 27k resistors and a typical mic pre impedance of about 2k, I would guess the actual clipping happens at about 2Vpp. Which is not a bad performance for common usage, but sure not enough for close micing of loud sources.

Not correct. The output stage has a clipping limit of 3,8mA pp. This leads to a clipping voltage of 7,6V pp with 2k load. As always the field of usage has to be considered. If you want to mike a starting jet or what kind of explosion ever simply reduce polarisation voltage..

 

[abbey road d enfer]

Not correct. The output stage has a clipping limit of 3,8mA pp. This leads to a clipping voltage of 7,6V pp with 2k load. As always the field of usage has to be considered. If you want to mike a starting jet or what kind of explosion ever simply reduce polarisation voltage..

As I said, it was a guess. I did not want to run a sim. Thank you for providing the real numbers.

 

[Cqwet Dbdfte]

I did not have the correct JFET model for Ulli's circuit. I modified his circuit and got a higher clipping level, like 22Vpp before getting nasty. R1 is lowered to 4.7k and a low noise, low Idss JFET is used. Current would idle at 3.8mA with R1's value.
An available P-MOSFET in default library is used in lieu of the PNP. There are likely better types, maybe the DMP10H4D2S-7?

 

[abbey road d enfer]

R1 and R6 are the main contributors to the output drive capability of this circuit. R1 and R6 should be similar (not equal) in order to balance DC loads.

 

[Cqwet Dbdfte]

R6 does not do much to the signal, may just provide some idle current to M1, to keep it from starving.
The value for R6 provides about 1mA, total draw about 3.8mA.
True differential signalling it is not, at least at audio freq.
I added a 1k grid stopper on M1. The JFET is not an UHF part and may still benefit from a grid stopper.
The sim looks really clean up to 22Vpp input.

 

[abbey road d enfer]

According to my simulation, in the original circuit, R4 and R5 (your R1 and R6) are equal contributors to the positive swing.
The negative swing is governed by the saturation voltage of the BJT (or Rdson of MOSFET), but the positive swing is governed by the sum of currents supplied via R4 and R5.

 

[Cqwet Dbdfte]

According to my simulation, in the original circuit, R4 and R5 (your R1 and R6) are equal contributors to the positive swing.
The negative swing is governed by the saturation voltage of the BJT (or Rdson of MOSFET), but the positive swing is governed by the sum of currents supplied via R4 and R5.

Which model did you use for the JFET?

 

[abbey road d enfer]

Which model did you use for the JFET?

I used a 2N3819, as it seems to be close enough to the 2SK117 proposed by the OP.

 

[Cqwet Dbdfte]

The 2N3819 can work withe the BJT up to 22Vpp as well, IF the current is increased. Harmonics are a little higher. A PNP base current is higher than the gate current (uA vs nA) making it easier to RC filter a bias supply. Does the PNP have a substantial noise advantage over the P-MOS in this config?
R1 is the main determinant of max voltage swing. The LSK170A does have nice noise specs for audio, not designed for VHF/UHF, with higher input capacitance, thus maybe less issues with RF interference. 2N3819 has a long proven history of mic use, certainly good enough, with a 25V rating it may be a slim margin for a 22V output in a 48V circuit.
I tried to see if this LSK170A and PMOS config can handle 22Vpp gracefully and it looks like it can. I'll order up a PCB and some parts to test it out.

 

[abbey road d enfer]

The 2N3819 can work withe the BJT up to 22Vpp as well, IF the current is increased.

I used the 2N3819 as a quick substitute for the 2SK117, in order to determine the possible drive capability. No doubt optimization would lead to using a different FET.
I wouldn't worry about the former's 25V rating, since the d-s voltage never exceeds 15V, even at turn-on.

I tried to see if this LSK170A and PMOS config can handle 22Vpp gracefully

Only because the load is unusually high for a mic input. Using an actual 2k load shows different. Only 10Vpp.
In this respect I don't see any significant difference between 2N3819 and LSK170A.
Neither do I see a significant difference between BJT and MOS.
With R1 and R3 so different, the DC current draw is quite unbalanced. It is not a problem if DC-blocking caps are used, but most xfmr-input preamps don't have them. Your arrangement would result in significant magnetizing of the xfmr core. And making them similar increases the output capability by about 20%.

 

[Cqwet Dbdfte]

I used the 2N3819 as a quick substitute for the 2SK117, in order to determine the possible drive capability. No doubt optimization would lead to using a different FET.
I wouldn't worry about the former's 25V rating, since the d-s voltage never exceeds 15V, even at turn-on.

Only because the load is unusually high for a mic input. Using an actual 2k load shows different. Only 10Vpp.
In this respect I don't see any significant difference between 2N3819 and LSK170A.
Neither do I see a significant difference between BJT and MOS.
With R1 and R3 so different, the DC current draw is quite unbalanced. It is not a problem if DC-blocking caps are used, but most xfmr-input preamps don't have them. Your arrangement would result in significant magnetizing of the xfmr core. And making them similar increases the output capability by about 20%.

A 2k load on the 3.4K source is perhaps not ideal if 22Vpp is an objective. I would have thought most transformer inputs would be isolated from phantom power avoiding DC magnetization. A 22Vpp signal would likely be too much meat for input stages, and require attenuation, or a step-down transformer.
A JT-13K7-A may be a candidate for step down with a 5:1 ratio and +7dBu tolerance.
A center tapped transformer feeding "phantom" power, without the 3.4K limit, would of course allow lower impedance loads. (an old idea it seems https://groups.google.com/g/rec.audio.pro/c/1AIYvpjTpFY?pli=1)
Most mic transformers are step down so this 22Vpp case is a bit unusual.

 

[kingkorg]

One thing i keep forgetting is that quiet, high headroom impedance converters can be very useful, and actually necessary if undamped capsule is used. These have very high output in the midrange. Something like this

https://groupdiy.com/threads/diy-push-pull-symetrical-undamped-f8-capsule.84868/

Or this:

https://groupdiy.com/threads/sennheiser-mkh800-mkh80-capsule-ks80.83049/

 

[abbey road d enfer]

A 2k load on the 3.4K source is perhaps not ideal if 22Vpp is an objective.

22Vpp cannot be an objective for a recording mic. Most preampa are designed to handle signals that rarely exceed -10dBu. 22Vpp is about 30dB abpove that.

I would have thought most transformer inputs would be isolated from phantom power avoiding DC magnetization.

No. Sometimes, capacitors are included as a HPF, but most have a galvanic connection to the xfmr.

A 22Vpp signal would likely be too much meat for input stages, and require attenuation, or a step-down transformer.
A JT-13K7-A may be a candidate for step down with a 5:1 ratio and +7dBu tolerance.
A center tapped transformer feeding "phantom" power, without the 3.4K limit, would of course allow lower impedance loads. (an old idea it seems https://groups.google.com/g/rec.audio.pro/c/1AIYvpjTpFY?pli=1)
Most mic transformers are step down so this 22Vpp case is a bit unusual.

I do not know of any commercial mic preamp that uses a step-down xfmr.
What's the point of carrying very hot levels for having to attenuate them at the receiving point?

 

[abbey road d enfer]

One thing i keep forgetting is that quiet, high headroom impedance converters can be very useful, and actually necessary if undamped capsule is used. These have very high output in the midrange.

The main advantage of underdamped capsules is that they have a very low acoustic impedance noise, which makes them desirable for capturing low SPL's.
Why would one use such a capule for capturing loud signals?

 

[kingkorg]

The main advantage of underdamped capsules is that they have a very low acoustic impedance noise, which makes them desirable for capturing low SPL's.
Why would one use such a capule for capturing loud signals?

There aren't many undampened single diaphragm F8 capsules. My experience with the ones i worked with is they have way superior off axis and on axis consistency, and no phase issues compared to dual diaphragm ones. According to Sennheiser research they have no distortion artefacts at higher frequencies caused by damping in damped ones.

 

[Cqwet Dbdfte]

The main advantage of underdamped capsules is that they have a very low acoustic impedance noise, which makes them desirable for capturing low SPL's.
Why would one use such a capsule for capturing loud signals?

Kingkorg's example from a plosive into an SDC produced over 22Vpp. I would think close miking a drum, trumpet, or vocal without a suitable screen would necessitate some means of limiting the signal or deal with a 30 to 140+ dBspl dynamic range.
In that scenario a step down conversion may be useful.
My schematic had two caps in series with the signal where only one is needed to avoid DC magnetization of the core.