Oktava MK-319 Modifications Pt. 2: Biasing a Toshiba 2SK170BL JFET

Before I started the mods on the Oktava MK-219, I had read a lot on the interwebs about modifying the MK-219 as well as the MK-319.

Most of the mods center around damping body or headshell resonance, but there is plenty to be done with the circuit as well. One of the changes in the original Scott Dorsey article is changing the stock Russian JFET to a Toshiba 2SK170BL - a popular, and extremely quiet transistor.

There is also considerable debate about whether or not this change is worth it - or if another JFET, namely a J305 - is a better choice.

Since I had procured a handful of each type of JFET, I decided to find out for myself.

The JFET has to be properly biased in the circuit. I'm familiar with biasing a tube in an amplifier, but not so much in a transistor device. But I learned.

It's actually fairly easy to do. And if I can do it, you surely can too.

What we do is feed a 1kHz signal into the gate of the JFET via a capacitor, and then look at the resulting output. In the Oktava MK-219 and -319, the bias is set by two resistors, R7 and R8.  (I'll have a picture of the board with everything labelled in the next post).

I just connected the signal generator's input to C2 (I think that's it), the input cap that then connects to the gate of the JFET. Put the scope on the drain of the JFET (the output).

Then turn up the output of the generator until one wave begins to clip. Then adjust R7 and R8 until the clipping is equal at the top and the bottom of the wave.

Now you may say, how do I adjust R7 and R8 - since they are fixed resistors? Ah ha! We put two 5K trim pots in their place. That way we can adjust the pots until we see the correct waveform on the scope.

Now let's do it.

I first wanted to take a look at the bias as it was with the stock Oktava transistor and bias resistors in place, to confirm that this method of biasing would work - in other words, to 'see' the bias on the stock JFET as it came from the factory.

I used my HP 200CD generator for the 1kHz signal.  And I connected my trusty Tektronix 453 up to the microphone with the probe at the gate of the transistor, so I could ensure I was seeing a nice sine wave. Which I could, as you see in the picture on the right.

Then I moved the probe to the drain of the transistor.

This is the clipping I saw - both sides are clipping. They don't have the exact same amount of clipping, but there you have it.

From what I read, the Oktava factory set the bias with a pot, measured that, then put the closest value resistors into the mic.

I wonder how consistent their transistors were - did they do this for each mic or did they just sample a batch and use the average value resistors for all of the mics?

My scope is ex-IBM - you can see the label scratched off on the front.

It also has an IBM property number carved on the top.

I don't think they'll be asking for it back any time soon.

After I confirmed that my method would work, I trimmed one leg off two 5K trimpots. These will go onto the board in place of R7 and R8.

This was a test fit for one pot.

You can also see where I removed the stock Russian FET and labelled the board as to what is what.

Very long story short, I tried a J305, but I just couldn't get it to bias - or even amplify much at all!  Maybe I need higher values than 5K in this circuit. I need to learn more about transistors I think - I'm fluent in tubes but not transistors!

I wound up putting a Toshiba 2SK170BL in, and it worked perfectly, showing a lot of gain.

After tweaking the pots a bit, I got good symmetrical clipping on both the top and bottom of the wave.

Here's just one cycle. I was able to tweak the pots so the clipping is virtually identical.

Aren't scopes fun? I don't use mine a lot but I really like using it.

Rather than measure the pots' value, I just left them in so the bias will be spot on.

I did put a drop of lacquer on each pot so their adjustment won't move.

Now to finish the rest of the circuit mods.