Old vs New Stock JFet in a D87 (U87 Clone)

[ktziavos]

Hello friends,
wondering the difference of sound in a D-87 using two different JFets . One new stock from mouser and one old stock(texas instmnts). The new ones gives me 2.4 mV at drain with 11 to 11,5 mV at Source and the old ones gives me 3 mV at drain with 11 to 11,5 at Source. changed some Jfets for experiment and the values are exact the same. How this voltage change effects the microphone sound?

 

[McIrish]

If the jfet is biased correctly, it should not impact much of a sound of its own. But that's the gotcha. Changing fets requires you to set the bias. I remove the capsule and inject a test tone and then set the bias for symmetrical clipping. I know a lot of new jfets, that claim to be the same as the original, have very different bias voltages.

 

[micolas]

In the same u87ai clone (not U87i), I tested 2N3819 from Motorola, Fairchild and Central Semiconductor, all with Idss approximately equal to 8...8.1 mA.
At Ud=11...11.3v I measured as follows:
Usource=1.8...1.9v for Motorola, Fairchild
Usource=2.4v for CEN
I liked the sound of the old ones more, it seemed more natural, more musical.
The new ones from CEN sounded more artificial, with a different tonal balance and went into distortion differently.
Personally, I like Motorola the most.
A stupid question:
Are the new ones also manufactured in Process 50, with exactly the same materials?

 

[McIrish]

I found the central jfets to be nothing like the originals. I ended up going with a better jfet for my custom build.

 

[micolas]

McIrish

What better jFET are you using in the U87 topology?

 

[kingkorg]

Hello friends,
wondering the difference of sound in a D-87 using two different JFets . One new stock from mouser and one old stock(texas instmnts). The new ones gives me 2.4 mV at drain with 11 to 11,5 mV at Source and the old ones gives me 3 mV at drain with 11 to 11,5 at Source. changed some Jfets for experiment and the values are exact the same. How this voltage change effects the microphone sound?

Whatever fet you test in u87 circuit, measure the frequency response of the whole circuit. The difference you hear could be due to the diference in the de-emphasis shape. It would be waste of time, money, energy to chase a certain fet when the only difference is built in EQ curve.

 

[Voyager10]

Exactly. The relevant factor here is the (open loop) gain of the FET amp, which determines how much effect the feedback components (from the FET drain back to the capsule bias voltage) will have on the frequency response.

In turn, the gain is determined by the transconductance (change in drain current for a given change in gate voltage) of the FET at its bias point.

 

[micolas]

...were also used
2SK105, BF256

 

[RuudNL]

I noticed that the gain (in circuit) may vary 1 to 2 dB between different 2N3819 batches.
It is important to adjust the bias for the specific FET under test, because an incorrect bias can also influence the gain.

 

[McIrish]

McIrish

What better jFET are you using in the U87 topology?

I used a LS846

 

[micolas]

McIrish thanks man!

The LS846 TO-92 seems adequate, but it is substantially more expensive than the 2n3819, especially since some of us prefer to buy a large quantity to select the best ones.
I will try it myself.
I did tests with BF245, BF256,2SK184,2SK118,
J201,J305,
2N5457,2SK30A,2SK105 etc, however I liked the 2N3819 NOS the best.
I will try from now on to make measurements, let's be objective, realistic, let's not waste resources, energy and time, with subjective things, KingKorg is right, that's the only way we can really be useful to the diygroup community.