R/E/P Community

R/E/P => R/E/P Archives => Klaus Heyne's Mic Lab => Topic started by: maxx on February 15, 2006, 05:18:30 pm

Title: U87 transistor inquiry
Post by: maxx on February 15, 2006, 05:18:30 pm
Can any one give a spec and description of component T1 in the schematic below. I have a U87 microphone with component S2436/TIP645. Searching for this transistor using typical cross reference charts leads to nothing. Is the S2436 transistor a proprietary one ?

Title: Re: U87 transistor inquiry
Post by: Klaus Heyne on February 15, 2006, 05:36:49 pm
The 2436 is similar to the 2N3819 for which you will find specs by Googling it.

Both transistors were used by Neumann interchangeably for at least 10 years, from the late 1960s to the late 1970s, and occasionally thereafter as well.

The two transistors mentioned above are also compatible with a couple of other ones Neumann used after the Texas Instrument 2N3819 became obsolete: K105 and SK107.

The specs are pretty close on all four of them, and you may liberally substitute them in any Neumann FET mic without negative consequences for the circuit or the sound of these mics, provided that you custom bias each FET, just as Neumann used to do, regardless of the brand of FET it used.
Title: U87 transistor inquiry
Post by: maxx on February 15, 2006, 05:47:26 pm
Thanks Klaus for the quick reply...

I believe the TIP645 or TIP64 is extinct. I cannot find any info on it anywhere.

I was'nt sure if the S 2436 was a 2Sk2436 or so.

I will certainly study the different data sheets on these.

Interestingly, The 2N3819 is also used in my KM84.

Is there a good primer on biasing fets on the web, I have seen a few but not sure how well they are discussed/written. Can one be recommended ?


Title: Re: U87 transistor inquiry
Post by: Klaus Heyne on February 15, 2006, 09:13:32 pm
Don't know hat you mean by "TIP645"...

The 2436 is not an SK-type. Neither is the SK 107, despite its letters...

Biasing of single FETs in mic applications is done to minimize distortion and maximize output. As FETs vary, it needs to be done for each individual FET in each individual mic.

I set these FETs up the following way:
I feed a 1K sine wave signal to the gate and slowly increase the source resistance from 0-10k Ohm with a temporary trim pot (10k) until I reach the lowest distortion of the FET.

This will be a pretty precisely defined point right after the highest output of the mic, right after the first heavy distortion audible when the FET turns on, and right before the more gradual onset of the distortion as the FET's output decreases again.

The trick is to feed enough 1k into the gate that at this ideal bias point the distortion is indeed barely audible or gone altogether, and clearly audible to each side of the ideal bias point.

All of this is best done at very low listening levels over headphones in a quiet environment. If you don't want to do it aurally, you can use an oscilloscope.

I then copy the ideal value I found  with the trim pot and choose a fixed source resistor of the same value for final installation.

If others have found another or better way to bias source-biased FETs in condenser mics, please post them!
But please be conscious of the forum's rules: keep the explanations simple!
Title: Re: U87 transistor inquiry
Post by: John Monforte on February 16, 2006, 12:45:20 am
I have used Klaus' method and it works well. But I would add that you might also want to test for noise.

Buy a bunch of these (they are cheap) and run through them to listen for 1/f noise. Sometimes this is called flicker noise or popcorn noise. It has a low level popping sound. It comes from contamination of the die when it is packaged and a device will either have it or not for its whole life. You will find some duds because noise is not a guaranteed spec on those parts.

You can also measure the "good" noise and find differences of a dB or two. To do this, you first need to bias it up and then replace the capsule with a 50pF capacitor (so you get rid of the acoustic stuff) and carefully shield your test setup to eliminate hum. You can measure a couple dB difference among a normal batch of parts.

If this last test sounds tedious to you, well it is! I built a little test fixture with a special test circuit which speeds up the process greatly. But now I am getting too techno...
Title: U87 transistor inquiry
Post by: maxx on February 16, 2006, 10:06:42 am

the transistor has two numbers on it S 2436 and below it TIP645. Seems TIP645 is irelevant given the right part # is above it on its plastic package. Was'nt sure ...

Would it be conceivable to find other compatible FET's that lend different tones or sound coloring to the U87 ?

Great info on biasing. I'm going to set up a test jig and try to do some biasing exercises to see how things work. If I find a jig set up that works I'll post its schematic so others can understand the working theory of this ( Tech Wanking ).

John :

Thanks for your info on biasing-very helpful, although a bit more complex but I see the method to your testing as more lab like. If you have a generic or basic test jig layout that you have used would it be possible to post it ?

If displaying a schematic of a test fixture is too Techno Wank then we won't go there. But, its just too interesting to pass up.
Title: Re: U87 transistor inquiry
Post by: Dale Ulan on February 16, 2006, 10:17:22 am
If you are in a pinch or in a hurry, and don't mind a compromise, you can use the same trimpot method except measure the voltage at the drain of the mic and get it to about 10 to 11 volts - that'll be in the general ballpark. Then, speak very loudly, very close, but without popping air at the capsule, into the mic while measuring the drain voltage on a DC voltmeter. If the voltage drifts around a lot, then adjust the pot a bit one way or the other to try to keep it from doing that. This is a very crude way of doing what Klaus' method is doing. Obviously his is better and more accurate, but I've done this in a pinch.

If you stop at the first (rough) part of this test (just setting the voltage), you might find that the mic will cut out occasionally on plosives or even loud passages.

If you are using this voltmeter method, make sure you measure the voltage using either a digital or a vacuum-tube voltmeter or an oscilloscope. You can't use a normal 'analog' voltmeter as it will upset the circuit too much.
Title: Re: U87 transistor inquiry
Post by: maxx on February 17, 2006, 03:26:24 am
Dale,  Hope all is well.

A nice description and also helpful info, thanks.

I will probably start with injecting a signal first, when my ears go out from the constant listening to a oscillator , then I take a break. That is if my ears don't fall off from fatigue.

Maybe I can bias T1 with some soothing music.  Rolling Eyes
Title: Re: U87 transistor inquiry
Post by: Klaus Heyne on February 17, 2006, 03:07:13 pm
maxx wrote on Fri, 17 February 2006 00:26

I will probably start with injecting a signal first, when my ears go out from the constant listening to a oscillator, then I take a break. That is if my ears don't fall off from fatigue.

This will not be an issue: The most reliable, discriminating use of your ears in this situation is at the threshold of hearing.
I found that both the onset of distortions are best audible, and listener fatigue minimized at the very low end of volume I recommend for this procedure.
Title: Re: U87 transistor inquiry
Post by: permeke on July 19, 2009, 07:42:50 pm
Dear Klaus.

Can you explain the bridges on R20 and R21 for U87i and U87P models as written on the schematic ?

I had the chance to compare my seventies U87 with an identical one from the same age , and mine has the bridges on and the other not.
other than that , I don't see any differences in capsules and electronics ( or can I have overseen something
Both have the microswitch on the connector removed.

kind regards

Title: Re: U87 transistor inquiry
Post by: David Satz on July 19, 2009, 08:52:54 pm
permeke, the "p" version of the output circuit was designed primarily to solve certain problems of the U.S. market in the 1950s and 60s, when many broadcast and recording consoles were still in use which had been designed for dynamic microphones, including ribbons. The output levels of professional condenser microphones were as much as 10 to 20 dB higher than the preamps in those consoles had been designed to handle.

In addition, the input transformers of some of those preamps were used in a circuit configuration which would cause audible errors in frequency response, and sometimes even parasitic oscillation, if the microphone's source (driving) impedance wasn't in the range of 150-200 Ohms. Thus the 50-Ohm setting of the microphone's output transformer might help with the input overload problem, while causing a frequency response problem at the same time.

In the "p" version of a Neumann microphone amplifier, the transformer is strapped for 50 Ohms and the bridges which you mention are opened so that the 47 Ohm resistors are in series with the transformer's secondary windings. This brings the microphone's output impedance up to around 150 Ohms and provides a further reduction in sensitivity, thus helping to avoid input overload while maintaining correct frequency response even with the problematic kinds of console preamps.

With modern professional preamps there is no need for the "p" variant of the amplifer any more, but during the reign of Gotham Audio Corp. as the North American distributors of Neumann microphones, the "p" configuration was the default--if you wanted 50- or 200-Ohm Neumann microphones you had to special-order them from Gotham (which I always did, and was always quizzed to make sure that I knew what I was doing).

If a resistive pad is needed to protect a preamp from input overload, that pad belongs at the input of the preamp rather than in the microphone. A pad at the input of a preamp attenuates any noise pickup as much as it attenuates useful signals. A pad in (or at the output of) a microphone reduces its useful signals while leaving any interfering noise unaffected. When you turn up the preamp gain to compensate for the pad, you raise the noise level even further. So the "p" circuit arrangement is really to be avoided nowadays.

--best regards
Title: Re: U87 transistor inquiry
Post by: permeke on July 20, 2009, 03:58:15 am
Thank you very much for the great explanation.

Title: Re: U87 transistor inquiry
Post by: David Bock on July 21, 2009, 04:49:33 pm
There are actually TWO ideal source resistor values for the 87 fet.

One is the ideal moderate signal value using the method Klaus suggests, which may also be done with a distortion analyzer.

The other is a best value for max spl resistor, where you MUST use a distortion analyzer to find.

They will not be the same resistor value, so you do have a choice as to which one you want to optimize for. If your 87 is stock you can use the test input for your osc, that way you know you won't interfere with the delicate biasing of the fet by loading the input to a place it never lives in in real life.
Title: Re: U87 transistor inquiry
Post by: permeke on October 10, 2010, 05:26:59 pm
Ok, It's been a while now,  Very Happy

but in the meantime I got me a second U87.
Here's what I discovered : both sound pretty close to each other. One has a fairly higher output and has a 9K3  Fet resistor. The other one has a 4K7 resistor. They both have the2N3819 Fet.

I believe both of the microphones have been "serviced" somehow during the years.

Now my question is, what's the closest value for the resistor to a stock U87 , I can't believe that the tuning value can't differ that much.

kind regards

Title: Re: U87 transistor inquiry
Post by: Klaus Heyne on October 10, 2010, 06:06:30 pm
Yes,"tuning" resistors- in this case, RS determines the ideal operating point of the FET- can vary widely- as much as 7k to 8k Ohms with the same type of FET used.

Do not assume the resistor you found in there will be right after a FET replacement.
You MUST select the correct source resistor when you replace the FET in this type of circuit. If you don't know how, take the mic to a condenser specialist who does. (It requires experience, a signal generator, and either a good ear or a scope.)
Title: Re: U87 transistor inquiry
Post by: Jim Williams on October 11, 2010, 10:35:58 am
Readers should know a scope will not show THD levels below about 3%, the limits of the trace errors. 3% is very high and should be trimmed down to a lower level.

As David says, a distortion meter is best for this. An analyzer like Audio Precision will show THD vs amplitude and will allow you to fine tune the THD. You will see the distortion change with levels so you can optimize it for high or low SPL's.

Then run a THD vs frequency sweep to see the effects at 20 and 20k hz. Small tweaks may allow you to minimize THD at one end or the other.

One can also use that rig to test jfets for noise. Then you can place various loads on the mic and test for loading effects.

Test gear like that eliminates any guessing and the inherent limitations of the ear to detect small distortion effects. It is necessary if you want to "nail it".
Title: Re: U87 transistor inquiry
Post by: MDM, on October 12, 2010, 10:00:10 am
I once used a spectrum analyzer on a DAW input, feeding a sine-wave.

As the transistor went in and out of the linear region harmonics would spring-up in the high frequencies.

when the harmonics are at their lowest then the transistor is probably biased reasonably well, I imagine.
Title: Re: U87 transistor inquiry
Post by: Klaus Heyne on October 12, 2010, 11:48:14 am
Indeed, the "trough" between the two types of distortions is what I have been aiming for when I bias these fets.

Unlike some others who chimed in, this old ludite still uses his ears to find that spot, but that requires a very quiet environment, as the spot is easiest to find at very low listening levels of the sine wave.
Title: Re: U87 transistor inquiry
Post by: Galil on October 13, 2010, 11:11:24 am
Would it be proper to assume that the less expensive condenser microphones on the market have not had their FETs tuned in an optimal fashion, as described in this thread?

Stephen Galiley
Title: Re: U87 transistor inquiry
Post by: permeke on October 14, 2010, 12:33:55 pm
Thanks again for the excellent explanation.

Back to the resistors: Is t just a matter of bridging the resistors or should something be rewired at the transformer as well ?

Kind regards-