5-pattern mic circuit

[dhammer]

Attached is a schematic I traced from Chinese-built 5-pattern FET mic.  I'm very impressed by the mic & decided to trace the whole thing as an exercise.  My understanding of mic circuits begins and ends with that ubiquitous Schoeps-style circuit & this is something else, with the pair of JFETs at the front-end.  I see something similar but not identical in other 5-pattern mics (414s, U89s & TLM-170s).

Does anyone have time/interest in talking through the circuit?  And also giving any feedback on the way I've drawn the schematic?  I drove myself a little nuts trying to get the pinouts for the transistors correct but I think I have them.  The switches are not so carefully drawn but not so important to me either.  Is there anything about the way I organized the drawing in terms of circuit sections that could be done more logically? 

I took some voltage readings at the transistors and a couple other places and put them on the scheme.  I am very curious about the 31.6v just off XLR Pin 2 at R37 –– shouldn't this be 48v?

Thanks very much for any comments!

EDIT:  I removed the attachment – final rev'd file is here:
https://groupdiy.com/index.php?topic=66809.msg848075#msg848075

 

[RuudNL]

Phantom voltage is 48 V when not loaded.
But there are 6.8 K.ohm resistors in series with the phantom power voltage.
If you measure 31.6V on pin 2 of the XLR, this means you have a voltage drop of 48 - 31.6 = 16.4 V over one 6.8 K.ohm resistor.
This corresponds with a current of 2.4 mA.

 

[abbey road d enfer]

Does anyone have time/interest in talking through the circuit?  And also giving any feedback on the way I've drawn the schematic? 

Your schemo would be easier to grab if you rotated Q4 & Q5 180°, Q2, 6, 7, 8 9 & 11 90° CW, Q1 & Q10 90° CCW. Traditionally ground is at the bottom of the page.
R1 and C1 should be connected to the collector of Q1.
I have a doubt about the HPF; it seems it works by selective NFB, reinjecting a low-passed version of the output signal out-of-phase, but there is no signal at the collector of Q9, for lack of a resistor..

 

[ln76d]

Your schemo would be easier to grab if you rotated Q4 & Q5 180°, Q2, 6, 7, 8 9 & 11 90° CW, Q1 & Q10 90° CCW. Traditionally ground is at the bottom of the page.
R1 and C1 should be connected to the collector of Q1.
I have a doubt about the HPF; it seems it works by selective NFB, reinjecting a low-passed version of the output signal out-of-phase, but there is no signal at the collector of Q9, for lack of a resistor..

Thanks abbey!

yeap, definately there's hard to read schematic, i truly wanted to add my "few cents" to the topic but have no time to puzzle schematic.

 

[dhammer]

Thanks for the replies and suggestions!  I'll re-draw based on these notes & report back.

 

[dhammer]

Attached is a new pass at this schematic following Abbey's advice.  Any comments appreciated – again either about the drawing or about the circuit.    I have questions about the circuit but best to wait to see if the drawings is working and clear.

Thanks all!

EDIT:  I removed the attachment – final rev'd file is here:
https://groupdiy.com/index.php?topic=66809.msg848075#msg848075

 

[abbey road d enfer]

Attached is a new pass at this schematic following Abbey's advice.  Any comments appreciated – again either about the drawing or about the circuit.    I have questions about the circuit but best to wait to see if the drawings is working and clear.

Thanks all!

Rotate Q1, Q3, Q7 180°, Q11 90° CW, Q10 90° CCW, Check Q12 connections.

 

[ln76d]

Much better, still hurts my eyes - but now much more because it's a pretty expanded circuit as for microphone
You didn't wrote what it is.
Topology remind me 797 mikes.

I have doubts about R9 connection  - did you redrew it properly?

First of all - 22nF at the output of tthe mike - i would remove it or reduce value at least to 2.2nF (or even lower value).
If you will check R9 connection then we can talk about input topology

Is it K67 capsule?

 

[dhammer]

Attached is rev with corrections to wiring of R9 and the resistor on the emitter of Q12 – thanks guys

I rotated Q1 but haven't done the rest Abbey called out.  I'll do that in the next round.

In76 – I think it is 797 factory done for an American company.  The capsule is k67-style.  I was wondering if there is EQ built into the circuit to smooth that capsule – C13, the 1nF film cap on Q3?

EDIT:  I removed the attachment – final rev'd file is here:
https://groupdiy.com/index.php?topic=66809.msg848075#msg848075

 

[ln76d]

Is it microphone manufacturer and model a secret ?

Rather not C13, but topology is pretty busy, so first would be good to make simulation - at least i have problems to find some eq points in a short time

Anyway - do you want to mod it a little?

 

[dhammer]

I'm interested in circuit mods for sure but hoping to understand what I'm seeing and doing & not just following a cookbook recipe.... 

Thanks a ton for getting interested in it!

 

[ln76d]

If you want to understand then try simulation because staring at the "bushes" is tiring

Ok then - mentioned above 22nF caps at the output are for RF filtration, but as usual, in chinese microphones are oversized. Affects response and phase badly - worth to remove (if your work environement let you) or change for much lower values.

First mod which i would make is change of input topology. Original IMO is the worst possible.
For better results you could use M49 type of multipattern connection or C414  - this one is really inpressive in case of bright capsules.  Not much to explain. Difference is that you can skip capacitor between capsule and gate - and this is usual huge improvement.  Of course that both other topologies use caps to make connection between  both sides but don't affect so badly response as the cap in series.

From the other stuff - i really like replace fets in 797 topologies for 2n3819, but you will have to re-bias it.

 

[abbey road d enfer]

I was wondering if there is EQ built into the circuit to smooth that capsule – C13, the 1nF film cap on Q3?

No; this cap is for stabilization of the circuit. It's actually a discrete opamp, and as such needs this capacitor to establish a dominant pole that is necessary for closed-loop stable operation.

 

[abbey road d enfer]

Ok then - mentioned above 22nF caps at the output are for RF filtration, but as usual, in chinese microphones are oversized. Affects response and phase badly - worth to remove (if your work environement let you) or change for much lower values.

Really? -0.3dB and 15° phase-shift at 20 kHz is excessive? -3dB at 75 kHz wrong?

 

[abbey road d enfer]

First mod which i would make is change of input topology. Original IMO is the worst possible.

Care to elaborate?

From the other stuff - i really like replace fets in 797 topologies for 2n3819, but you will have to re-bias it.

In typical single-FET topologies, 2SK170 has too high capacitance, which unduly loads the capsule, introducing loss of signal and distortion, but here with the opamp topology, whre NFB bootstraps the g-s capacitance, this is much less of an issue. 2SK170 has a noise advantage over 2N3819.

 

[ln76d]

Really? -0.3dB and 15° phase-shift at 20 kHz is excessive? -3dB at 75 kHz wrong?

I have little bit different numbers, but i'm rather "simulator" not "calculator" type of geek
Someday maybe i will make simulation of it - i'm thinking to buy one of these mikes in some time - for fun
Anyway - everything what's even close to audible range affects badly mike response when is applied at the output, especially with bright capsules.
Rather don't want make theoretical  discussion about it, because it don't have any sense - it's a matter of try for the owner. These are only two caps - so i think there's no problem to make tests. I didn't found any chinese topology where this woudn't be significant improvement.

Care to elaborate?

Tried before - but i'm not the best in theory. Anyway i will try, if you want
1. Capacitor in series is responsible for overall degradation of sound - here matter its quality and value - 1nF is not always is good (depend on capsule, polarisation voltage and FET)
2. Changing topology, you can get higher impedance at the input - 1G against 500M - in most of FET mikes it make a difference (quite opposite to tube mikes)
3. Grounded capsule is much more sensitive to environmental conditions - wanna bet? ( ) - make test for different topologies without headbasket installed.

Overall response is better. I did many one to one (A/B) tests with different topologies.
Still the best is so-called "pure cardioid" (like U47 for example) but not always possible to adapt in multipattern mikes.

In typical single-FET topologies, 2SK170 has too high capacitance, which unduly loads the capsule, introducing loss of signal and distortion, but here with the opamp topology, whre NFB bootstraps the g-s capacitance, this is much less of an issue. 2SK170 has a noise advantage over 2N3819.

Again - matter of try, especially in microphone circuits theory is not enough usual, there's many things on the way, especially when we have high impedances. With both types used in 797 designs 2sk170 and 2n5457 difference is really significant when change it to 2n3819 (also other FETs - but this is one which i prefer most).
Here i did a lot of different tests also - and yes, nothing will substitute tests -  the effects can be really surprising as also not always will fit theory approach
Again - matter of try  - i think is better than staring at the "bushes" and "numbers"
Not really have option to compare noise figures - for 2n3819 there's not much in datasheet, and i'm not able to figure it out from the rest of the parameters. 
For capacitances - input capacitance is ten times higher in 2sk170, not really sure how it's changed in this topology (it isn't in parallel), but i imagine that even if it is lower still we have 10 times lower in 2n3819 - correct me if i'm wrong.
Still - these are only numbers, and there's a lot of other parameters and we need to add to this production divergence and many other factors - to make life easier - two FETs, one trimpot, soldering station and tin  

 

[Khron]

That's an interesting statement, considering the original Schoeps schematic also has those caps at 22nF. Silly germans didn't know what they were doing, hmm? ;D

Ok then - mentioned above 22nF caps at the output are for RF filtration, but as usual, in chinese microphones are oversized. Affects response and phase badly - worth to remove (if your work environement let you) or change for much lower values.

I'm somewhat confused about the capsule biasing arrangement, though. If the front diaphragm is permanently grounded, shouldn't the backplate be biased to some voltage?

For example, for the omni pattern, ok, both diaphragms need to be at the same potential, but different from the backplate. As the last schematic is drawn, when omni is selected, the bias voltage coming from the DC-DC circuit is shunted straight to ground.

There's something really funky in the pattern select section of the schematic, at least

 

[abbey road d enfer]

Anyway - everything what's even close to audible range affects badly mike response when is applied at the output, especially with bright capsules.

That may be the case when the output stage is pushed into slew limiting, due to a combination of excessive sibilants and strong capacitive loading. I must admit there is lack of data to properly assess this point.

Rather don't want make theoretical  discussion about it, because it don't have any sense - it's a matter of try for the owner. These are only two caps - so i think there's no problem to make tests. I didn't found any chinese topology where this woudn't be significant improvement.

Since most chinese capsules are oversibilant it is quite possible the output stages are put under stress.

1. Capacitor in series is responsible for overall degradation of sound - here matter its quality and value - 1nF is not always is good (depend on capsule, polarisation voltage and FET)

I would disagree somewhat there. Capacitors in series introduce distortion when their value is small enough for a significant voltage to develop across it. In this example, the cap is about 15 times larger than the capsule and loaded with a high impedance so no significant voltage develops there.

2. Changing topology, you can get higher impedance at the input - 1G against 500M - in most of FET mikes it make a difference (quite opposite to tube mikes)

Agreed, but my question was regarding the head amp topology, not the bias. You mentioned input topology, which IMO concerns the head amp, not the bias circuitry. But ok, matter of language.

3. Grounded capsule is much more sensitive to environmental conditions - wanna bet? ( ) - make test for different topologies without headbasket installed.

Agreed. But due to the multi-pattern type, one can't ground the diaphragms.

Again - matter of try, especially in microphone circuits theory is not enough usual, there's many things on the way, especially when we have high impedances. With both types used in 797 designs 2sk170 and 2n5457 difference is really significant when change it to 2n3819 (also other FETs - but this is one which i prefer most). 

Agreed. In a single-FET head amp, 2SK170 is not a good option. Less a problem in a source-follower configuration, though.

Not really have option to compare noise figures - for 2n3819 there's not much in datasheet, and i'm not able to figure it out from the rest of the parameters. 

2N3819: 6nV/sqrtHz, 2SK170: 0.95 nV/sqrtHz.

For capacitances - input capacitance is ten times higher in 2sk170, not really sure how it's changed in this topology (it isn't in parallel), but i imagine that even if it is lower still we have 10 times lower in 2n3819 - correct me if i'm wrong.

Bootstrap tends to make the difference less noticeable. Wet finger says capacitance is halved, which is still 15pF. But OTOH, in a common-source topology, Miller effect adds about 20pF to a 2N3819.

 

[dhammer]

Khron's right – I had the pattern switch wired wrong.  The switches are tricky for me because they are 2-pole/5-position rotaries being used to cover 5 positions.  I fixed it in the attached.

I also reoriented the transistors according to Abbey's advice further up the thread.  I have a feeling I didn't understand the real goal of these rotations so I don't know if the way I did them really helps the scheme make better sense.

The advice from Piotr & Abbey's analysis are very interesting & questions abound.  I'll come back tonight to get some of them out there...

EDIT:  I removed the attachment – final rev'd file is here:
https://groupdiy.com/index.php?topic=66809.msg848075#msg848075

 

[abbey road d enfer]

I also reoriented the transistors according to Abbey's advice further up the thread.  I have a feeling I didn't understand the real goal of these rotations so I don't know if the way I did them really helps the scheme make better sense.

It makes understanding the circuit easier, with positive voltages going up, current going down in all the transistors.
One last thing, you should mirror Q3 and Q11 so teh base is on the left; traditionally input is on the left, output on the right.