More JFET Biasing Questions

[javacody]

From Peter Snowberg:

For the 2N5457, if you are still taking the signal from the source, try eliminating the drain resistor (short it) and use a 4.7 in series with a 10K pot (set at 50%) for the source. If you want more gain, add a cap across the source resistance. Try .68uF for a Marshally tone, and 22uF for a full spectrum to be more Fenderish. You can always toss a 10uF in there and see how you like it. Really low values like 0.1uF will boost higher treble.

First, should that resistor in series with the 10k trimmer be 4.7 or 4.7k?

Second, I'm not sure if I'm picturing this in my head correctly, so here is a schem to verify:



Does this look correct?

Also, is there a name for taking the signal from the source vs. from the drain? Is it simply a matter of phase reversal?

[Peter Snowberg]

:oops: Sorry, I was being dimwitted when I wrote that. I was thinking of two different configurations hybridized into one non-functional circuit. My apologies.

When you don't have a drain resistance, you can't use a cap across the source. Yikes.... how did I write that? I don't even drink. :?

To use a voicing cap, you must wire the stage as a drain follower and you must have a drain resistance.

In any case, the resistor on the input needs to be a little different. That configuration will still let sound through, but it will cut the level by a huge amount and it will eat all the highs. It's a good trick for removing gain and you'll see it between stage 2 and 3 of Fender amps, but in this case I would bring the signal into the gate directly, or through a MUCH smaller resistor and then attack the 3.3M to ground.

That's a good place to start with the resistors. JFETs vary over a huge range, so you may end up needing to raise or lower the 4.7K in order to get the best sound. It depends on who made the JFET and when it was made too. Semiconductor fab processes are constantly on the shift.

As far as names for the configurations... You're looking at either a source follower, or a drain follower. The name really applies to the next stage, but its named by where it's fed from. In the BJT world, you're talking about emitter followers and collector followers, and in tube land the names are cathode follower or anode follower. The inverting stage is the de-facto standard so most of the time you simply don't hear names like anode follower. Typically, source followers have a lower impedance, but that’s not a rule at all.

Take care,
-Peter.

[gez]

First, should that resistor in series with the 10k trimmer be 4.7 or 4.7k?

Also, is there a name for taking the signal from the source vs. from the drain? Is it simply a matter of phase reversal?

If you take the signal from the source then it's called a Common Drain circuit or a Source Follower (usually the later).  It gives you unity (slightly less in practice) gain and reasonably low output impedance - it's a buffer.

The last thing you want to do is bypass the source resistance (I think Peter's comments about this were in relation to a common source circuit), all you're doing is shunting the signal to ground with that 22u cap - no signal at the output!  

The 3M3 should connect to ground and the input signal should be coupled directly to the gate (input cap optional - but it's good housekeeping to include one IMO).

I'm pretty sure the resistor value is 4k7 in series with the pot, though you only need this if you're designing a common source amp to get the drain to bias at half the supply OR if you want an specific quiescent current - a lot of the time you can just sling in a fixed value and it gets you in the right area.  

[javacody]

Doh! I was in too much of a hurry and it shows. R1 should connect to ground.

But I am getting very confused!

How do I control the amount of gain I see?

Here is a proposed schem, how should I tweak it?





Here is a link with a similar schem:

http://ourworld.compuserve.com/homepages/g_knott/elect247.htm

I see he uses a 100k resistor on the drain? I'm guessing that this is in relation to a higher v+?

[javacody]

I see on this page http://www.geocities.com/tpe123/folkurban/fuzz/snippets.html that on the Jinx, Tim used a 50k trimmer in series with a 1k resistor on the drain and a 10k trimmer on the source with a 10uF cap on the wiper of the trimmer. What does that 10uF cap on the wiper of the source's trim pot do? Act like a tone control? Adjust gain?

[Peter Snowberg]

The gain is controlled by the resistance change of the JFET versus the resistors on the source and drain. Now you're in voodoo land because you have think in terms of current flow through the JFET, and what that does to the voltage. Time for Ohms law.

In your diagram, R3 + R4 establish the total voltage across the JFET or better yet the amount of current traveling through the JFET, while R3 adjusts to satisfy Vgs(off). This is all about balance. The 100K in his diagram is what sets the amount of current running through the JFET. It's actually a current source. As this current decreases, the JFET has an easier time influencing the amount of power flowing through it and the gain goes up.

To put it bluntly.... a small resistor on the drain will give you small amplification while a large resistor will increase gain. As the gain goes up, details that didn't matter before start to become big factors. Thermal noise in the resistors will become a big deal.

What does that 10uF cap on the wiper of the source's trim pot do? Act like a tone control? Adjust gain?

The 10uF provides an AC path to ground while the resistor it's across provides a DC voltage to allow the source to be at a higher voltage than the gate. As you turn the pot up, the cap gets the AC path more directly connected to the source which lowers the impedance of the source supply while the resistance stays the same. I told you that you were in voodoo land now. Since bass take more energy to produce, it effectively works as a bass boost as the cap gets closer to the source.

Here are a few buffer designs:


Take care,
-Peter

[Peter Snowberg]

One (OK, maybe two) more thing(s)... I drew resistors on all the gates above, but i'm assuming that the input has NO DC on it so a coupling cap should also be used if you can't be 100% sure that in the absence on signal the gate will sit at ground potential. If you have any DC there, the transistor bias will get whacked.

The series resistor is totally optional, but it will work like a resistive divider with the gate load resistor to decrease the drive level, or to shape the tone depending on the value. Try using a 1 meg pot in this position and see how it changes things. There's nothing like empirically gained knowledge.

Take care,
-Peter

[javacody]

Here is what my experimentations have landed on:



I initially put a 100k pot in on the drain and a 10k trimmer on the source. I also went with a 22uF cap on the source. With a 100KOhms of resistance on the drain, it produced tons of gain and distortion. Very crunchy. I don't have a multi meter so I decided to start experimenting with different resistors. I decided I would keep halving the value of the resistors until I found one I liked. 39k, 22k, 10k, and then I landed on 4.7k.  10k was probably usable but left me with no clean at the lowest gain setting on the lm386.

I think I could understand this stuff better if I had a multi meter. Then when I tweaked stuff I could check the restance, voltage, etc. etc. I should be getting one soon, and then I will repeat this experiment. I like what this buffer sounds like right now, but I still don't understand much of what is going on with the JFET. I'm still not 100% certain what raising/lowering resistance on the source does.

[Peter Snowberg]

I think I could understand this stuff better if I had a multi meter.

This is a place where the classic "a picture is worth 1000 words" comes into play.

Given the circuits that you're messing with, I would advise getting a meter (a $30 unit is just fine), and then scanning eBay for a used scope.

The resistors change the position of the doorway, but if the room is pitch black the relativity of where you're standing goes out the window.

Take care,
-Peter