How critical are matched transistors in Schoeps type circuits? (curious)

Help Support GroupDIY:

Blue Jinn

Well-known member
Joined
May 28, 2009
Messages
394
Location
USA Northwest
Looking to build an "Alice" type mic and maybe a straight forward "GenChina" on some of RuudNL's boards from a few years back.

Although it looks like a bare bones tester is under $US20, and transistors are pretty cheap, just wondering how critical this is.
 

Bo Deadly

Well-known member
Joined
Dec 22, 2015
Messages
2,687
Location
New Jersey, USA
Matching components in electronics in general usually has to do with either temperature compensation, common mode rejection or distortion. There are no obvious problems with temperature compensation in a Schoeps circuit. Distortion is not as much of an issue in a Schoeps pre compared to say a guitar amp transformer output where each transistor is handling half or the signal. CMRR looks like it could be a factor if there is enough difference in the emitter resistance because the parts are biased slightly different. But the caps to ground are going to present the lowest impedance at high frequencies and CMRR gets better at low frequencies so in practice it probably doesn't matter a whole lot.

Bottom line, probably not much but it's complicated enough that I personally would not feel super confident about the answer without making empirical measurements of CMRR and THD under different temperature conditions.

But if you're just making one mic or a few, it's pretty easy to just do a basic single-pass matching with a cheap transistor tester. If you really want to be pedantic about it, create a simple circuit to test a transistor for noise and AC gain and pick two that are close.
 

RuudNL

Well-known member
Joined
Apr 26, 2009
Messages
2,455
Location
Haule / The Netherlands
The most obvious problem you will get with unmatched transistors, is that there will be a DC offset between the two output lines (pin 2 and 3 of the XLR). Especially transformer balanced inputs will not like that very much.
For electronically balanced input this isn't usually a big problem. With carefully selected transistors, the DC offset can be as low as 1 mV. This isn't a problem for a transformer coupled input.
Personally I select and match transistors for Hfe and Vbe.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,285
Location
Marcelland
Regarding CMRR, even if the transistors were perfectly matched for Hfe, one of the legs would have a different impedance than the other.
This is because the output impedance reflects the source impedance seen by the base.
Since one base sees the FET source and the other the drain, there is a structural difference. The output impedance of the FET is higher at the drain than at the source.
The actual calculation implies adding the dynamic emitter resistance (26/Iq) and the emitter spread resistance.
With the usual 2.2k resistors, the output Z at the drain is, of course, 2.2k and the output Z at the source is about 500r (depending on the FET's Gm).
With the typical transistors used the reflected impedance at the output is about 7r for the lower transistor and 15r for the upper.
Since protection resistors need to be added, it is not too difficult to make the south one higher by about 8 ohms than the north one. For example 47r for the south and 39 for the north.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,285
Location
Marcelland
With carefully selected transistors, the DC offset can be as low as 1 mV. This isn't a problem for a transformer coupled input.
It begs the question "how much DC current is tolerable in an input xfmr? I know id depends very much on the transformer and what is tolerable.
Has anyone done and published an evaluation? Maybe xfmr manufacturers have facts?
 

rogs

Well-known member
Joined
Jun 23, 2006
Messages
353
Location
UK
.............The output impedance of the FET is higher at the drain than at the source...........
I've seen this mentioned before, but I do have trouble understanding quite how it works, when used with JFETS like the 2N5457 and J113 in the 'Alice' / Schoeps style phase splitter configuration?
(t's an 'Alice' type project the OP mentions in his first post).

The data sheets for both those devices state that the drain and source terminations are 'interchangeable' ---- and in a typical 'Alice' /Schoeps circuit both are loaded with the same value resistors. (typically 2k2)
What happens to the relative drain and source impedances if the terminations are reversed?
Trying it out in practice doesn't seem to reveal any difference.....The drain and source do seem to be interchangeable?
Obviously I'm missing something here......
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,285
Location
Marcelland
I've seen this mentioned before, but I do have trouble understanding quite how it works, when used with JFETS like the 2N5457 and J113 in the 'Alice' / Schoeps style phase splitter configuration?
(t's an 'Alice' type project the OP mentions in his first post).

The data sheets for both those devices state that the drain and source terminations are 'interchangeable' ---- and in a typical 'Alice' /Schoeps circuit both are loaded with the same value resistors. (typically 2k2)
What happens to the relative drain and source impedances if the terminations are reversed?
Trying it out in practice doesn't seem to reveal any difference.....The drain and source do seem to be interchangeable?
Obviously I'm missing something here......
It's the arrangement that determines the output impedance. It is correct, source and drain have no structural difference, what makes the difference is the direction of current. Just like many characteristics of an active device depend on a number of factors (Iq, Vds, Vgs)
 

rogs

Well-known member
Joined
Jun 23, 2006
Messages
353
Location
UK
It's the arrangement that determines the output impedance. It is correct, source and drain have no structural difference, what makes the difference is the direction of current. Just like many characteristics of an active device depend on a number of factors (Iq, Vds, Vgs)
It seems as if many (non transformer) hobby mic projects look to the famous Schoeps CMC5 circuitry for inspiration -- me included!
It does look as if Schoeps has made his circuit symmetrical -- although, to be fair, the values of the output inductors DR3 and DR4 are not listed, so they might be marginally different values? --
Or - if not - then Schoeps had presumably not thought the differences significant enough to be of concern?
 

Khron

Well-known member
Joined
Apr 8, 2010
Messages
1,986
Location
Finland
It does look as if Schoeps has made his circuit symmetrical -- although, to be fair, the values of the output inductors DR3 and DR4 are not listed, so they might be marginally different values? --
Or - if not - then Schoeps had presumably not thought the differences significant enough to be of concern?

I'm no expert, but that's most likely, i'd say. And anyway, if there's enough RFI that can upset things "within" the cable, that's more than enough to "gag" the microphone itself, so a couple dB of CMRR one way or the other become a bit of a moot point.
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,285
Location
Marcelland
It does look as if Schoeps has made his circuit symmetrical -- although, to be fair, the values of the output inductors DR3 and DR4 are not listed, so they might be marginally different values? --
Or - if not - then Schoeps had presumably not thought the differences significant enough to be of concern?
Apparently Jorg Wuttke knew that his design introduced some assymetry in the output balance, but he thought it was not a major issue, particularly considering that transformerless preamps did not exist at the time. An unbalance of 10 ohms results in a theoretical CMRR of about 56dB.* In the typical radio-electric environment of the 1970's, it was deemed acceptable. In consideration of the actual much noisier RF environment, it seems the designers at Schoeps, like many others have favoured rejection filters (particularly those that "dump" RFI to earth) over a more elaborate CMRR arrangement.
I don't disagree with this choice.

*This is for a typical circuit with P48 supplied by two 6.8k resistors. It may be possible to achieve a better figure with a 3.4k resistor feeding a center-tapped primary; it may also be worse :).
 

Latest posts

Top