Takstar CM 60 suggested mods/improvements (?)

[Voyager10]

The CM-63 electronics is better in most aspects. The JFET stage runs on 13.4V. Much better. But the bias adjustment was totally off...

The CM-60 can be fixed to use a higher voltage (in this schematic, raising the value of R10 will do it). But I do wonder whether it's really worth saving $10 over the CM-63, given that you need to fit a bias pot as well (plus the RFI stuff is harder to fix, and you won't have a pad switch).

 

[MicMaven]

Are these actually recorded at the same time? To me the mic position sounds different.

They were recorded one after another at the same position and distance from the speaker/player with exactly the same flat software settings. Crazy eh? James

 

[pmfalcman]

1000pf is commonly seen between the capsule and the FET.

But, on the photo I see 100nJ marking, which, to my knowledge, is 100 nF 5 %.
Hence my question.

 

[jp8]

The CM-60 can be fixed to use a higher voltage (in this schematic, raising the value of R10 will do it). But I do wonder whether it's really worth saving $10 over the CM-63, given that you need to fit a bias pot as well (plus the RFI stuff is harder to fix, and you won't have a pad switch).

On the CM-63 schematic, R6 is the same resistor as R10 on the CM-60 schematic and is 3k3 instead of 150 Ohms. That should fix the low voltage on the JFET circuit. And, for somewhat lower noise, the JFET can be replaced by a 2SK209-Y (mind the Y). I did not try that, though. But it should work. I bought the CM-60 and CM-63 just to serve as donor bodies and as a reference to compare my own PCBAs with. I did not intend to modify the stock PCBAs. Replacing the two resistors on the CM-63 was all I did.

Jan

 

[HerbertR]

On the CM-63 schematic, R6 is the same resistor as R10 on the CM-60 schematic and is 3k3 instead of 150 Ohms. That should fix the low voltage on the JFET circuit. And, for somewhat lower noise, the JFET can be replaced by a 2SK209-Y (mind the Y). I did not try that, though. But it should work. I bought the CM-60 and CM-63 just to serve as donor bodies and as a reference to compare my own PCBAs with. I did not intend to modify the stock PCBAs. Replacing the two resistors on the CM-63 was all I did.

Jan

It is not entirely clear form me why changing the R10 150Ohm from the DC-DC input converter should bring a higher voltage to the JFET source. The voltage measured at about 5-6V on the JFET is relatively low and is, as I understand, provided directly and unregulated from the phantom power via R11 6k8 and on to R6 2K. The DC converter in my case also provides about 46V before the R12 1G Ohm connection and the "big" filter capacitor 100nF to provide the capsule voltage. I think to increase the JFET voltage the only way would be to lower this "main resistor" R11 6K8 (the voltage drop is about 27V across the resistor, drawing about 4mA for the whole circuit), but this would also affect the DC converter input voltage at the same time. With the R9 2K, there is still a drop of about 2 - 2.5 V/1.2 mA across the JFET voltage rail. Theoretically, the R11 should be reduced to 5K, which would only drop about 20V and give an "optimal" 12-13V at the FET after the R6 2K ohm. In parallel, the R10 150 Ohm would also have to be increased to about 3K (2.7mA) to hold the 7V in order to avoid a burst of about 14.5V at the DC converter after the R10 150 Ohm. Or am I wrong?

But the good thing is that the microphone seems to work very good even with just 5 V on the FET and some mods.
Yes, the C7 capsule capacitor to the gate is 1000pF on the CM60. I cannot say whether this is a C0G SMD. The switch to Styroflex is noticeable fuller a note imo, the 2SK117 also feels noticeably more transparent and nicer in the low end. The 100nJ Cap C2 is a Capsule Voltage Filter Cap. Nothing super special. Regards

 

[kingkorg]

You are ignoring the zener guys. I remember changing ZD1 to higher value, and adjusting R11. Can't remember the values off the top of my head. I get better noise performance as well with ca.60V on the capsule. However original Schoeps has voltages in this range at the drain 5-6v or something.

 

[Voyager10]

It is not entirely clear form me why changing the R10 150Ohm from the DC-DC input converter should bring a higher voltage to the JFET source. The voltage measured at about 5-6V on the JFET is relatively low and is, as I understand, provided directly and unregulated from the phantom power via R11 6k8 and on to R6 2K.

Think of R11 and R10 as a voltage divider, where one end is at +32V or so, and the other end is fixed at 6.8V by the zener ZD1. Increasing R10 reduces the voltage divider's ratio so the R10/R11 junction will then be at a higher voltage.

(The zener passes as much current as is necessary to keep the ZD1/R10 junction at +6.8V. Think of it as 'absorbing' all the surplus current which comes through R10 but isn't needed to power the DC converter).

 

[HerbertR]

Think of R11 and R10 as a voltage divider, where one end is at +32V or so, and the other end is fixed at 6.8V by the zener ZD1. Increasing R10 reduces the voltage divider's ratio so the R10/R11 junction will then be at a higher voltage.

(The zener passes as much current as is necessary to keep the ZD1/R10 junction at +6.8V. Think of it as 'absorbing' all the surplus current which comes through R10 but isn't needed to power the DC converter).

You are a "fox" ... I completely overlooked that, thanks!!!

 

[HerbertR]

Today I played around with the CM63 mics and compared them to the CM60 with the 3U capsules. The FET voltage of the CM63 is set to about 9.2V from the factory on both CM63s. Both CM63s with 3U SDC cards sounded much too "colored" for my taste. Bias calibration is clearly needed. The factory setting was way off with lots of harmonics. I ended up with an optimal 10.2V on my two Mics where all harmonics are lowest/disappeared in balance with as low a second harmonic as possible. Also the capsule voltage before the capsule resistor R12 1G is about 54V compared to about 45V on the CM60, so almost 10V higher. The output level of the CM63 mic is therefore about 2-3dB higher and the noise level a few dB lower. Exactly what you reported in the thread, thank you guys!!!! The CM63 (with the need for bias adjustment) is clearly the better choice for an excellent Mic. Great pieces for the price, wow.
And yes, the factory Omni capsules on the CM63's, well, they are causing not ear cancer imo....but... hmm yes...

 

[Voyager10]

After a bit of tinkering, I've come up with the following mods to the CM60 circuit board.




It will help to have the CM60 schematic and board overlay from this thread to hand.

Mod A​

Replace R1 and R2 (1G, probably) with 2G resistors - this lowers the low-frequency noise by 2-3dB. I used these from Mouser - they are tiny and a real pain to solder.

Mod B​

Replace 2SK208Y FET (Q1) with a 2SK209GR. This increases gain, and lowers noise, in conjunction with the bias mods (C and E) below.

Mod C​

Correctly bias the FET, for best headroom. The board has space for a trimmer, but as I didn't have one that was small enough, I used trial and error with different resistors. You can just push one through the holes (no need to solder, yet) and measure the voltage across R5 (or equivalent pins on the switch, or C10).

Putting a resistor as shown in the photo puts an additional resistor in parallel with R3, which raises the bias current. Ideally (in conjunction with mods B and E) you need 1-1.3mA bias current, i.e. 2-2.6V across R5. (This post has more detailed data on JFET biasing for the Schoeps circuit).

For my 2SK209, I ended up with an 82K resistor which gave 1.3mA. It took a long time for the current to settle on its final value, so patience is required.

Mod D​

Raise capsule polarisation voltage, for increased sensitivity and reduced noise. The PCB allows you to fit a trimmer in place of R9, but again I just experimented with a few fixed-value resistors (in parallel with R9). A 2K7 resistor gave about 58V measured on R7, up from 45V or so on the original board.

Mod E​

Raise FET operating voltage, for increased headroom/lower distortion. This requires an increase in the value of R10, as discussed above, so I replaced the original 150R R10 with a 2K7 resistor. Working voltage is just over 12V working voltage, up from 7.2V.
R10 doesn't need to be surface-mount - in the photos it's wired between the +ve end of C12 and one of the R10 pads.


The CM63 already has the equivalent of mods D and E, and mod C is just adjusting the trimmer. The board will need a good cleanup afterwards, especially if your soldering is as neat as mine ;-).

 

[k brown]

Looking at the CM-60 schem; wondering if worth adding ~ 470pf caps in parallel with the Qs' 150k resistors, as in the Schoeps circuit.

 

[jp8]

Perhaps needless to say, but the CM-63 will also benefit from mods A and B.

 

[jp8]

This is in the '60 or the '63?

You were quick in replying to a message that I accidentally posted and now revised. The removal of the track refers to the 60, though the 63 has this audio track running under the oscillator too. But when I removed the other output cap, I measured the same 2.5MHz level on that output. There is probably another coupling path the signal enters the audio tracks. Maybe common impedance in the ground, but as that is a solid plane, I'm not sure of that. So I was a little quick concluding the track underneath the oscillator is a real issue. But if I were to route such a board, I would keep the audio tracks as far away as possible from the oscillator. And give the oscillator its own ground plane that ties to the auduo circuit at the first decoupling elcap. That's how I build up my SDC PCBs with CMOS oscillator and there's zero RF noise from the oscillator on the outputs.

Jan

 

[Voyager10]

Perhaps needless to say, but the CM-63 will also benefit from mods A and B.

Yes, definitely. I'll post some measurements when I have them.

 

[jp8]

Looking at the CM-60 schem; wondering if worth adding ~ 470pf caps in parallel with the Qs' 150k resistors, as in the Schoeps circuit.

I doubt the usefulness of these capacitors. If they were intended as an RF filter to avoid AM demodulation in the output BJTs, I would have expected them between emitter and base. The AC path through C9 to chassis ground has too much parasitic inductance and reduces the effectiveness of the caps in the MHz range. Maybe even creating a parasitic LC resonant circuit.

It is also not suitable as a filter for the 2.5MHz oscillator frequency because the simulated cut-off frequency of over 3MHz is too high. Note: I simulated cut-off frequency without the output capacitors.

Simulations suggest there is gain peaking of the JFET stage between 10 and 100MHz, which is indeed effectively suppressed by the 470pF caps. But does this gain peaking actually occur in real-life circuits? I'm not so sure about that. Did anyone measure FR of the Schoeps circuit at frequencies beyond 10MHz and without any filter caps?

Jan

 

[jp8]

Well, I fired up my scope and function generator and up to 50 MHz I did not observe gain peaking on the Drain and Source terminals. What's wrong with my LTspice sim? (Rethoric question, but any suggestions are welcome.)

Btw, the same question could be raised regarding other parts in the Schoeps circuit that are not found in the CM-60, such as the RFI ferrite beads Dr3, Dr4 and Dr5. Takstar improved on that in the CM-63 with a CM choke and three ferrite beads, which I consider a plus. Trying to get a truly effective filter on the stock CM-60 PCB will be challenging, though. Ideally, for a RFI filter to be effective, you would also like to have an XLR insert like the one on the CM-63, which has a short conductive path between pin 1 and the tube. Just inserting some flying inductors won't cut it, I'm afraid. I have recently made a KM84 circuit RFI-proof this way, but I had to design the PCB from the ground up as if it were an RF design and with the CM filter and ferrites as close as possible to the XLR. All other Frankensteined circuits failed miserably.

Jan

 

[kingkorg]

After a bit of tinkering, I've come up with the following mods to the CM60 circuit board.

View attachment 132642
View attachment 132643

It will help to have the CM60 schematic and board overlay from this thread to hand.

Mod A​

Replace R1 and R2 (1G, probably) with 2G resistors - this lowers the low-frequency noise by 2-3dB. I used these from Mouser - they are tiny and a real pain to solder.

Mod B​

Replace 2SK208Y FET (Q1) with a 2SK209GR. This increases gain, and lowers noise, in conjunction with the bias mods (C and E) below.

Mod C​

Correctly bias the FET, for best headroom. The board has space for a trimmer, but as I didn't have one that was small enough, I used trial and error with different resistors. You can just push one through the holes (no need to solder, yet) and measure the voltage across R5 (or equivalent pins on the switch, or C10).

Putting a resistor as shown in the photo puts an additional resistor in parallel with R3, which raises the bias current. Ideally (in conjunction with mods B and E) you need 1-1.3mA bias current, i.e. 2-2.6V across R5. (This post has more detailed data on JFET biasing for the Schoeps circuit).

For my 2SK209, I ended up with an 82K resistor which gave 1.3mA. It took a long time for the current to settle on its final value, so patience is required.

Mod D​

Raise capsule polarisation voltage, for increased sensitivity and reduced noise. The PCB allows you to fit a trimmer in place of R9, but again I just experimented with a few fixed-value resistors (in parallel with R9). A 2K7 resistor gave about 58V measured on R7, up from 45V or so on the original board.

Mod E​

Raise FET operating voltage, for increased headroom/lower distortion. This requires an increase in the value of R10, as discussed above, so I replaced the original 150R R10 with a 2K7 resistor. Working voltage is just over 12V working voltage, up from 7.2V.
R10 doesn't need to be surface-mount - in the photos it's wired between the +ve end of C12 and one of the R10 pads.


The CM63 already has the equivalent of mods D and E, and mod C is just adjusting the trimmer. The board will need a good cleanup afterwards, especially if your soldering is as neat as mine ;-).

CM60 has hi-z area covered in transparent mask, have you done anything to protect the area?

 

[HerbertR]

After a bit of tinkering, I've come up with the following mods to the CM60 circuit board.

View attachment 132642
View attachment 132643

It will help to have the CM60 schematic and board overlay from this thread to hand.

Mod A​

Replace R1 and R2 (1G, probably) with 2G resistors - this..........

Great, you listed a lot of mods together!

I implemented the U87is RFI filter with 47uH and 330pF in the CM60s. This required cutting out the connection traces on the board from XLRs 2/3. XLR 3 is behind the SMD side which is connected to pads through the board...very tricky but...why not. I also "decoupled" the board ground with a ferrite bead to XLR1, a lot of precision work to "dig" the XLR ground out of all the board connections. But oscillator and audio ground are not disconnected as mentioned as this would be more effective from JP8. CM63 is clearly better designed with a much better noisefloor.

 

[Voyager10]

CM60 has hi-z area covered in transparent mask, have you done anything to protect the area?

I've got some 'Flexible Silicone conformal coating': https://uk.rs-online.com/web/p/conformal-coatings/0535525?searchId= - which seems to work OK.

I'm obviously concerned that I don't seal in any dirt or moisture, so I've not done the Takstar boards yet. When I've finished fiddling with them I'll give them a very good clean then apply some coating.

 

[Voyager10]

Here are some measurements of the mods from post #30. Done with the same setup as this post, so the results should be roughly comparable.

Headroom / THD​

Mods	Gain	THD @ 100mV RMS in	Vin for 0.5% THD
None	1.36x	0.14%	0.35V
A, B, C, E	1.48x	0.012%	2.02V

(Note mod A, the 2G resistors, isn't relevant to these numbers)

Noise​

Trace	Applicable mods	Equivalent input noise dBV (A)
1	None	-118.6
2	A	-119.4
3	A, B, C, E	-120.8