Schoeps/Dorsey circuit theoretical questions

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

l0calh05t

Member
Joined
Nov 12, 2008
Messages
17
Location
Germany
(Theoretical because I'm not planning on doing this to my brand new Oktavas just yet :p)

I was recently looking at the dorsey mod schematic and at the schoeps schematic, and obviously the two are similar, but with a few differences. I was wondering how big these differences are.

The first one would be the DC-DC converter. How big of an advantage does the 60V polarisation voltage give over the 35V of the much simpler dorsey circuit? Does this improve sensitivity, linearity, noise?

The second is the choice of the input FET. The schoeps uses a BC264 which has an input capacitance of 4pF, and the dorsey uses a 2SK170 with an input capacitance of 30 pF. Considering the very high impedance at that point, isn't such a large input capacitance a bit of a problem?

At 16kHz the BC264 should have a input reactance of 2.5M (already quite small compared to the 1G precharge resistor), the 2SK170 on the other hand only 330k. Wouldn't it be better to use the BC264, despite the slightly higher noise figure?
 
l0calh05t said:
(Theoretical because I'm not planning on doing this to my brand new Oktavas just yet :p)

I was recently looking at the dorsey mod schematic and at the schoeps schematic, and obviously the two are similar, but with a few differences. I was wondering how big these differences are.

The first one would be the DC-DC converter. How big of an advantage does the 60V polarisation voltage give over the 35V of the much simpler dorsey circuit? Does this improve sensitivity, linearity, noise?

The advantage is almost 6dB of the sensitivity increase and as a consequence the same amount of S/N improvement.

The second is the choice of the input FET. The schoeps uses a BC264 which has an input capacitance of 4pF, and the dorsey uses a 2SK170 with an input capacitance of 30 pF. Considering the very high impedance at that point, isn't such a large input capacitance a bit of a problem?

At 16kHz the BC264 should have a input reactance of 2.5M (already quite small compared to the 1G precharge resistor), the 2SK170 on the other hand only 330k. Wouldn't it be better to use the BC264, despite the slightly higher noise figure?

The choice of 2SK170 is completely wrong, as used in Dorsey mod (as well as in many commercial mics). Having said that, it might be a very good choice, once all the voltages and operating points set right. Just look carefully at its data sheet to find the answers, otherwise, use what Schoeps does (and leave in the biasing pot). Look for Zapnspark messages concerning the right procedure of biasing.

Best, M
 
Marik said:
The advantage is almost 6dB of the sensitivity increase and as a consequence the same amount of S/N improvement.

If the polarization voltage isn't too high that is, right? Couldn't the higher voltage slightly change the tuning of the capsule, and thereby it's sound (or even bottom out the membrane)? Does anyone happen to have the original Oktava MK-012 schematic? (not the chinese version, should it be any different)

The choice of 2SK170 is completely wrong, as used in Dorsey mod (as well as in many commercial mics). Having said that, it might be a very good choice, once all the voltages and operating points set right. Just look carefully at its data sheet to find the answers, otherwise, use what Schoeps does (and leave in the biasing pot). Look for Zapnspark messages concerning the right procedure of biasing.

Best, M

Not quite sure which "answers" you mean. The 2SK170 is being operated at a much higher voltage than BC264 in the schoeps circuit (12V vs.  6.1V before the 2.2k). From the Ciss-Vds diagram in the 2SK170 I can see that the capacitance drops at higher voltages, but even at 30V it's still more than 25pF. Or am I missing anything?

Anyways, for everyones enjoyment, here are a couple relevant posts I found:

abbey road d enfer said:
Unfortunately "polarization voltage" may be related to many different subjects (chemistry, optoelectronics...), and many call that "bias voltage". But with "microphone", "capsule", "condenser", you get more targetted hits. Increasing the bias voltage will increase the sensitivity, not only for the obvious linear relationship, but also because the increased attraction force will reduce the space between the diaphragm and the fixed plate, thus inducing an increase in the actual capacitance of the capsule. That puts a definite maximum value to the vias value, if exceeded the diaphragm will touch the fixed plate, which is ashow-stopper. If the inter-electrode distance is too reduced, the actual signal will not be linear anymore. The increasd tension on the diaphragm will increase the resonance frequency, which may or may not be of consequence. In the other direction, decreasing the bias brings a loss of sensitivity, thus a loss of dynamic range (because the noise is essentially unchanged), which may be a good thing if you intend to use your mic with loud acoustic sources. I noticed just the other day that one manufacturer uses this technique for its 10dB pad, instead of the more common technique of shunting the capsule wit a fixed cap of ca. 3x the capsule's capacitance. Each method has its pros and cons, been the subject of an article in the AES journal a few years back.

zapnspark said:
What abbey road d enfer said.
Here is a diagram to help you visualize a couple of ways that capsules are typically "polarized"

TypicalCondPolCircs-1.png


The MXL 603 uses the method shown in fig. b

Cheers.

ZAP

Rossi said:
Gus' advice is good. What has become known as the Dorsey circuit is a very stripped down version of the Schoeps circuit. Having written similar articles, I understand that doing away with everything that's not essential was necessary in order to turn this circuit into a good beginners project. But if you're beyond the absolute beginner stage, you should have another look at the details left out.

Some random remarks:

The original version has a 60V dc/dc-converter for proper polarization voltage. Without it, the Dorsey version leaves only about 35V for capsule bias; you lose about 5 dB S/N compared to 60V.

The original version also has a bias adjustment pot which allows for lower THD and higher headroom. You only need an additional 1 Meg (precision) trimmer for that.

About your layout: commercial mic pcb usually use teflon stand-offs in the hi-Z area. Soldering the capsule wires and Gig-resistios to the pcb may invite leakage currents.

Also, 1u caps between the FET and BJTs may be a little large. There's practically no LF reduction going on other than those caps, and you may run into problems with sub-frequencies if you choose them this large. 470n is the largest value I would recommend. The original Schoeps circuit uses only 100n.

zapnspark said:
IMO, Rossi covered the bases very nicely.
There is another important part omission from the Dorsey circuit and - that omission has been repeated numerous times in the various Schoeps-like schematics found on the internet.
There should be a 47 uF/50 V electrolytic from the joined collectors of the PNP emitter follower transistors to ground.
This lowers the mic. output impedance, increases drive and lowers noise sensitivity.
It also helps to reduce (mainly) 2nd harmonic distortion.
Lots of folks get by without that cap but - I believe it is important to have it there for the reasons given.
(You can see that cap in the original Schoeps CMC mic. schematic)

Cheers.

ZAP
 
Marik said:
The second is the choice of the input FET. The schoeps uses a BC264 which has an input capacitance of 4pF, and the dorsey uses a 2SK170 with an input capacitance of 30 pF. Considering the very high impedance at that point, isn't such a large input capacitance a bit of a problem?

At 16kHz the BC264 should have a input reactance of 2.5M (already quite small compared to the 1G precharge resistor), the 2SK170 on the other hand only 330k. Wouldn't it be better to use the BC264, despite the slightly higher noise figure?

The choice of 2SK170 is completely wrong, as used in Dorsey mod (as well as in many commercial mics). Having said that, it might be a very good choice, once all the voltages and operating points set right. Just look carefully at its data sheet to find the answers, otherwise, use what Schoeps does (and leave in the biasing pot). Look for Zapnspark messages concerning the right procedure of biasing.

Best, M
Can you elaborate on that? I have no idea what the Dorsey mod is, but I don't see why the 2SK170 would be such a bad choice.
 
l0calh05t said:
Considering the very high impedance at that point, isn't such a large input capacitance a bit of a problem?
In fact, the impedance here is governed by the capsule's capacitance, which is probably around 40pF. And the reactance of the impedance converter is deeply modified by the topology of the stage. If configured as a common-source, the FET's g-d capacitance is multiplied by the gain of the stage (Miller effect), making the g-s cap negligible. That's why a cascode stage is often preferrred there (bottom element of a cascode operates with no voltage gain, hence reducing Miller effect). If configured as a voltage-follower (common-drain), the input capacitance (Cgs) is bootstrapped by the gain i.e. divided by 1-G (roughly divided by 100) with the g-d capacitance becoming the dominant. I don't know the schematic.
Since the source impedance is capacitive, the input capacitance of the impedance converter acts as a linear attenuator, contrary to the input resistance, which acts as a high-pass filter.
[/quote]
 
abbey road d enfer said:
Marik said:
The second is the choice of the input FET. The schoeps uses a BC264 which has an input capacitance of 4pF, and the dorsey uses a 2SK170 with an input capacitance of 30 pF. Considering the very high impedance at that point, isn't such a large input capacitance a bit of a problem?

At 16kHz the BC264 should have a input reactance of 2.5M (already quite small compared to the 1G precharge resistor), the 2SK170 on the other hand only 330k. Wouldn't it be better to use the BC264, despite the slightly higher noise figure?

The choice of 2SK170 is completely wrong, as used in Dorsey mod (as well as in many commercial mics). Having said that, it might be a very good choice, once all the voltages and operating points set right. Just look carefully at its data sheet to find the answers, otherwise, use what Schoeps does (and leave in the biasing pot). Look for Zapnspark messages concerning the right procedure of biasing.

Best, M
Can you elaborate on that? I have no idea what the Dorsey mod is, but I don't see why the 2SK170 would be such a bad choice.

http://www.pfarrell.com/images/miccircuit.gif

Intended for LDC mic.
 
abbey road d enfer said:
Can you elaborate on that? I have no idea what the Dorsey mod is, but I don't see why the 2SK170 would be such a bad choice.

See here: http://www.recordingmag.com/resources/resourceDetail/322.html (better schematic to be found here: http://www.pfarrell.com/music/shanghaimic.html )
Also of interest, as it applies to the MC-012 (the chinese copy of the MK-012): http://www.recordingmag.com/resources/resourceDetail/315.html
 
Marik said:
Not sure I understand what you mean

By what? That it's also of interest? That's because I think it's interesting (although not really relevant to the comparison. just another circuit to compare to). Or about the MK-012/MC-012 difference? All real Oktava MK-012 microphones from Russia were always labeled as MK-012 and only the chinese copies/fakes ever had the MC-012 label.

After your edit:
The article's title is "Oktava MC-012 Upgrades", so the schematic is probably derived from the chinese copy. Which is why I asked for the original MK-012 circuit earlier.
 
l0calh05t said:
Marik said:
Not sure I understand what you mean

By what? That it's also of interest? That's because I think it's interesting (although not really relevant to the comparison. just another circuit to compare to). Or about the MK-012/MC-012 difference? All real Oktava MK-012 microphones from Russia were always labeled as MK-012 and only the chinese copies/fakes ever had the MC-012 label.

Sorry, I did not challenge the fact that it is interesting. FYI, the MK012 and MC012 are the same Russian microphone and the schematics you posted is of Russian version and has nothing to do with Chinese copycat.

Hope it helps.

Best, M
 
Marik said:
Sorry, I did not challenge the fact that it is interesting. FYI, the MK012 and MC012 are the same Russian microphone and the schematics you posted is of Russian version and has nothing to do with Chinese copycat.

Hope it helps.

Best, M

Hm, maybe I am mistaken. It appears that MC 012 was used for an older version of the MK 012 (but the chinese ones were labeled MC 012 as well), which would still hint at different circuits ( look at the top image here, text is in german: http://www.musiker-board.de/vb/studio-mikrofone-instrumente/355214-oktava-mk-012-modifikation-erfahrungsbericht.html )
 
l0calh05t said:
Marik said:
Sorry, I did not challenge the fact that it is interesting. FYI, the MK012 and MC012 are the same Russian microphone and the schematics you posted is of Russian version and has nothing to do with Chinese copycat.

Hope it helps.

Best, M

Hm, maybe I am mistaken. It appears that MC 012 was used for an older version of the MK 012 (but the chinese ones were labeled MC 012 as well), which would still hint at different circuits ( look at the top image here, text is in german: http://www.musiker-board.de/vb/studio-mikrofone-instrumente/355214-oktava-mk-012-modifikation-erfahrungsbericht.html )

The Russian version was labeled either MK or MC, which stand for either Микрофон Конденсаторный (Mikrofon Kondensatornij), or Microphone Condenser. The first is a Cyrrilic version, and the second being an export version. There is no difference between them.

Best, M
 
Marik said:
The Russian version was labeled either MK or MC, which stand for either Микрофон Конденсаторный (Mikrofon Kondensatornij), or Microphone Condenser. The first is a Cyrrilic version, and the second being an export version. There is no difference between them.

Best, M

I know a little russian, which is why the MC on Russian originals surprised me (on the other hand Oktava isn't written in Cyrillic either).
The current versions sold in europe are all labeled MK-012 and from the images, they would appear to be quite different (newer pcb, different components, but probably the same circuit as before)
 
I don't think there were any 012 Chinese copies when Dorsey's article appeared.

IMO the 012 circuit nicely complements the capsule. The Dorsey variant of the Schoeps circuit was suggested for LD mics, which can be okay with a 35V bias, the capsule giving a higher output than a SD mic. The "Dorsey circuit" was meant to do away with the not so great transformers in most Chinese mics, and, being a beginners' project, it had to be very simple. So he left out a DC converter and all the RF protection. He also put in a hefty (and somewhat rare) 4.7u/50V tanalum cap to filter the capsule bias when he could have used a small film cap in a hi ohms filter network. Or even better two filter stages in series. But that would have complicated the schematic. So keep in mind that the "Dorsey circuit" is optimized for simplicity, not for best performance.

The "Dorsey circuit" is not a good circuit for a SD mic as S/N is likely not satisfactory. The 012 circuit consumes less current and thus leaves about 42V for capsule bias, which is sufficient, the capsule being on the large side of SD.

Most capsules are constructed for 60V. And in those designs, a 60V bias does not change the sound very much compared to lower voltages. A K67 for instance collapses at 120V. You obviously need a voltage well below that so the diaphragm can move with enough amplitude to cope with high SPLs. 60V was found to be optimal. If you choose a voltage high enough to seriously alter the diaphragm tension and thus the sound, you're probably too close to the collapsing voltage.





 
I have done a bit of noise analysis on these circuits.

As far as Fet input capacitance... I note that high input capacitance occurs with very low input voltage noise fets. As if there were many fets in parallel.

So as far as dynamic range, the dividing action of the high capacitance fights the lower input noise. I did an analysis of this and found that optimum occurs when fet input capacitance equals capsule capacitance, in the most general case.

But there is one term I don't have a number for. The real part of the capsule motional impedance. Gradient condenser microphones are largely acoustic resistance controlled, and this would transform into a capacitor with series resistance...the latter being a thermal noise source.

Does anyone have a rough number for this resistive component in typical capsules? Of course it would be frequency dependent.
 
Rossi said:
I don't think there were any 012 Chinese copies when Dorsey's article appeared.

IMO the 012 circuit nicely complements the capsule. The Dorsey variant of the Schoeps circuit was suggested for LD mics, which can be okay with a 35V bias, the capsule giving a higher output than a SD mic. The "Dorsey circuit" was meant to do away with the not so great transformers in most Chinese mics, and, being a beginners' project, it had to be very simple. So he left out a DC converter and all the RF protection. He also put in a hefty (and somewhat rare) 4.7u/50V tanalum cap to filter the capsule bias when he could have used a small film cap in a hi ohms filter network. Or even better two filter stages in series. But that would have complicated the schematic. So keep in mind that the "Dorsey circuit" is optimized for simplicity, not for best performance.

Yes, it's definitely a very simple circuit.

The "Dorsey circuit" is not a good circuit for a SD mic as S/N is likely not satisfactory. The 012 circuit consumes less current and thus leaves about 42V for capsule bias, which is sufficient, the capsule being on the large side of SD.
This is pretty much what I thought. But the original Schoeps circuit should function quite well in an SDC, considering that the CMC5 is an SDC circuit.

Most capsules are constructed for 60V. And in those designs, a 60V bias does not change the sound very much compared to lower voltages. A K67 for instance collapses at 120V. You obviously need a voltage well below that so the diaphragm can move with enough amplitude to cope with high SPLs. 60V was found to be optimal. If you choose a voltage high enough to seriously alter the diaphragm tension and thus the sound, you're probably too close to the collapsing voltage.

Shouldn't the capsules in microphones be constructed for the voltage actually used? I.e. shouldn't the MK-012 capsules be designed for 40V?
 
I was referring to the Dorsey variant, not the Schoeps Cirucit, which has a 60V DC converter and other goodies. The Schoeps Ciruit is a very good circuit for a SD mic. The Dorsey variant is a so-so circuit even for a LD mic. As I said, the Dorsey variant may be a good circuit to learn from and for amateurs to build but it is not an optimal circuit for any mic.

I'm not sure you can build a capsule that works better at 40V than at 60V. There are physical limits. That said, there are always compromises to be made. A 60V bias would require additional circuitry which would make the mic larger (and more expensive). And for some applications the additional headroom of the lower capsule output may be more desirable than lower self noise, anyway. Keep in mind that the 012 has no pad swich but requires a separate attenuation pad.
 

Latest posts

Back
Top