JFET Buffer Help

[Dylfish]

Hey Guys,

I've been reading about JFETS and how the common drain config has a High z input and a Low z output.

I also know that the "stream" between D&S is almost like a variable resistor that's controlled by the voltage of VGS, but i'm struggling to see (I'm easy terms) the common drain keeps the output at unity. Is it due to an internal resistance in the JFET source and the low value source resistor forming a voltage divider that the output signal comes from?

This is the circuit I've been looking at.



Thanks,

[kingswayguitar]

When I was re-learning this stuff for guitars, I developed a hockey analogy.

The source resistor is like the goalie.  He's solid... can keep you  in the game, keep you from falling behind on the scorebaord (unity), but can't help on offense.  The other 5 guys do all the skating around and passing to put you ahead on the scorebaord (put you ahead as in gain/amplification).

[Mike Burgundy]

There's a big bucket of math involved, but the eventual formula for source follower (=common drain) voltage gain is
gain=(g_mR_source) / (g_mR_source +1)
with g_m = transconductance of the device.
It's reasonable to assume transconductance times R_source to be large, so gain will inevitably approach one.

If you think of it in a less theoretical way, the FET works like a fosset, opening (very low resistance) and closing (very high resistance) controlled by the tap/gate - actually  the gate referenced to the source. This bit is important. The whole gain of 1 bit comes in because the opening/closing is controlled by the voltage between gate and source. If you look at *just* drain and source, you'd think with the FET fully open, Vs would be very close to 9V right? But here's the catch - this will move the source voltage up so Vgs will change. It's keeping itself in check. More source voltage means the tap is closing more.
Does that help?

[Dylfish]

Thanks Guys,

To be honest I don't think I'm an idiot by any stretch but it's just been really hard to drill these concepts into my head.

I don't want to throw parts of pedals together and hope they work but at the moment I'm struggling a bit =)

Should the transconductance be on the datasheet? I'm struggling to find it.

Cheers!

[Jabulani Jonny]

This may reveal my lack of understanding, but is it due to the difference in resistance between R1 & R2?  I would think R1 is to block DC, but may there's a correlation. I dunno.

[Dylfish]

My understanding is R1 sets the input impedance and makes sure the gate is 0v to allow it to be biased by Vs (Vgs = Vg - Vs).

I too could be incorrect

[Mike Burgundy]

Transconductance (gcan be on datasheets. sorry, I  should have explained what that is - it is the ratio of how much a voltage change on the input causes a CURRENT change on the output. In this case drain current devided by the change in Vgs. Small change in Vgs causes big change in drain current for large transconductance. g_m=ΔI_D/ΔVgs in which Δ stands for "change"
Hows the fosset analogy working for ya?

[Jabulani Jonny]

Math....bastards.    Tapout.   

[PRR]

> the "stream" between D&S is almost like a variable resistor that's controlled by the voltage of VGS, but i'm struggling to see (I'm easy terms) the common drain keeps the output at unity.

You can't see electrons. So it is hard to "visualize".

We can see water. If you drape a garden-hose nozzle hanging down, and turn on water, it rises. If you have a variable valve, at low flow the nozzle rises less. With a powerful flow and a sensitive valve, the nozzle-lift may be much larger than the valve lever force.

So let's lift an elephant with a fingertip.

Let's make the height of the elephant =equal= the height of our finger.

We need a steady current of water, the pump (or battery).

For the Source Follower, we allow the valve to move up-and-down, and stand the elephant on the valve. Assume the valve is on vertical sliders.



When valve is open, valve and elephant rise; when valve is closed valve and elephant fall.

Now hold you finger at a fixed height, say 2 feet. If you have the valve lever hooked up right, if valve and elephant are below 2 feet they will rise. When they rise above 2 feet the valve will close, valve and elephant fall.

So the height of the elephant will stay 2 feet.

If you move your finger to 3 feet, valve and elephant rise until they reach 3 feet, again the valve closes, they drop, and find a balance at 3 feet.

If you jiggle your finger height, the elephant's height jiggles.

> a High z input

It is possible to fingertip control large forces. Car power steering.

> a Low z output.

If you replace the elephant with a dog, the valve will still close/open at fingertip level. True, the valve may be more-closed with a dog-load, so the valve may have to rise a little bit to balance the dog instead of the elephant. But say 2.0 feet versus 2.1 feet, for load from 2,000 lbs to 20 lbs-- that's good height regulation which is all that "low Z" means.

[Jabulani Jonny]

That's probably one of the most helpful posts I've seen in 2 years. Seriously. Thanks for that.

[goldstache]

mind blown!  Thanks for the analogy

[mth5044]

PRR how high are you right now.

[Dylfish]

That a brilliant analogy PRR.

This might be a very stupid question but what internal resistance is generated by the JFET itself?

I've constructed this so there is nothing on the gate and in my head (at least) the 9v will be flowing through the JFET uninterrupted since there is no established depletion region until it hits the 330oh resistor which should then drop all its voltage over it to ground.

It seems as if there is a "voltage divider" established that "burns" off a tonne of voltage before the 330oh resistor, is it an internal resistance from the Jfet itself or am I very misguided?



Thanks,

Edit: is it due to the Idss value?

[bluebunny]

Paul - when are you going to write a book to house all these priceless nuggets of brilliant explanation?  When I say "write", I mean "draw wonderful pictures".  The illustrations and analogies are to die for.  Never mind that Horowitz and Hill stuff...

[duck_arse]

the elephant will provide a large load.

the drain-source resistance Rds is often presented graphically in the manufacturers datasheet. although, fet datasheets are very inconsistent with the way they present their specifications.

[chptunes]

If there's a "Post of the Year Award", I nominate Paul R.

[PRR]

> what internal resistance is generated by the JFET itself?
> there is a "voltage divider" established that "burns" off a tonne of voltage before the 330oh resistor, is it an internal resistance from the Jfet itself

Figure a random JFET will be around 1K Ohms.

So yes, 330 ohms is a HEAVY load.

You can get somewhat fatter JFETs.

There is also a non-ohmic limit, Idss.

IIRC, J201 is a small little beast made for low-current work. Idss may even be specced at 1mA, which echos the 914uA numbers on your drawing.

*Why did you pick 330 Ohms??*

In general audio design, you pick the DC load resistor 2X to 5X smaller than your lowest expected audio load. So 330 Ohms implies 700 to 1K6 loads. Even studio gear does not run this low these days. =>10K for studio. =>50K for guitar-cord work. The two fields are merging, take 10K. That suggests 5K to 2K, NOT 0.33K of DC load.

Try again with say 3K3 and you will probably get decent swing.

Assume that a low-current 9V JFET buffer is really aiming at classic guitar-cord interfaces, 50K up, so a 33K DC load may be suitable.

Note also that IF you could get that JFET biased to around half supply, 4.5V at Source, that is near 14mA of power drain. In days of battery we would HAVE to question such power consumption. With 3K3 DC load only 1.4mA, which still would be questioned, but doesn't stand-out like a hole in the wall.

[PRR]

> Rds is often presented graphically in the manufacturers datasheet. although, fet datasheets are very inconsistent

There's two basic uses for JFETs, and thus two different ways the data are given.

* Amplifiers
* Switches

The actual JFETs may be the *same* under the different part numbers and datasheets. Certainly they are all selected from the same barrel.

Amplifier favors a low Vto while Idss may not be so critical. Rds is often omitted, but Gm (Gfs or Yfs) is kinda-nearly the inverse of Rds. For J201 the Yfs (at 1KHz) is given as 500 micro-Mho. Which is 1,000,000ohms/500, or 2K Ohms. (Which won't drive 330 Ohms with authority.)

Switch favors a high Idss and a Vto which is not low (but not high). Rds is often given directly.

Both come out of the same kettle, the way a baking sheet of cookies yields some soft and some hard.

There's also various "sizes" of dough they put in the oven. If you need a very sensitive gate (low leak current or low capacitance) they start with a small die. For very high current work they start with a huge die, and the resulting FETs will not be as easy to drive (though rarely an issue in audio; condenser mikes the marginal case).

If you "must" drive 330 Ohms, you want a JFET 20 times fatter than little J201. (Or twenty J201s parallel.)

If you must drive guitar-cord loads, make the 330 Ohms about twenty times bigger, 5K or 10K. That's a good compromize between the ~~2K of J201 and the ~~50K audio load.