JFET brain teaser!

On my coffee break this morning, I wired up two JFET amplifier circuits.



Riddle me this, Batman: which of these amplifiers is "better", and why?

> the first is ...better for manufacturers because it uses less components. But the second is ...more stable.

Right, except... version A will fail to work for some FETs, leading to warranty returns (or testing/rework) costs typically greater than the two added resistors.

The audio difference is nil. The 2dB difference in predicted gain will vanish in FET variation, especially for version A.

If sims are to be believed the first (left-hand) circuit has significantly more distortion above second than the other. Both are low by a bunch relative to the second harmonic.

I don't have the J201 in the library, but the Process 52 2N3686 seems to be a good match for your d.c. levels.

Just add an emitterfollower and you can drive inputs with low impedance.

Yep, I know you guys were too smart to take things at face value :thumb: The circuit on the left looks superficially "better" because of its lower parts count and marginally higher gain, but the circuit on the right does indeed give more predictable performance over the Vgs spread that can be expected with a JFET.

By the way, these circuits were actually built, not simulated.

Buttachunk: Unless I'm really missing something, I don't see the point of increasing Vgg and putting a cap across R5. Increasing Vgg means a higher Vs and therefore a further reduction of Vds (and maximum output swing). And R1+(R4||R5) appear in shunt with the input signal, not in series, so I don't see any benefit in AC bypassing of (R4||R5).

(Vgs= gate-source voltage, Vgg= gate voltage relative to ground, Vs= source voltage relative to ground, Vds= drain-source voltage).

The DI-input of the 'Great River N*v*' does this as well: three resistor-biasing network without a cap (3*1M).

Regards,

Peter

[quote author="clintrubber"]The DI-input of the 'Great River N*v*' does this as well: three resistor-biasing network without a cap (3*1M).

Regards,

Peter[/quote]

I think the main reason to bypass the divider is to reduce the effects of power supply noise. If the coupling cap-gate resistor time constant is large and the source impedance low, this is not as much of a problem.

Note that you can tweak the values of the unbypassed divider to effect some cancellation of power supply noise, if the signal off the drain is referred to ground---as it is in this example. This will depend on the source impedance. If the signal source can be referenced to the divider tap, and then if the divider has about the same attentuation as the stage has gain, there is cancellation, since the drain impedance is typically fairly high relative to the load R.

At higher frequencies depending on impedances the drain-gate capacitance will start to dominate and spoil things. You can pick up 30-40dB of reduction midband this way for high Z sources or floating sources (like transformer secondaries) referenced to the divider. The divider impedance should be low to make the thermal noise contribution small.

EDIT: Also, with a floating signal source with d.c. conductivity the coupling cap and gate resistor could be eliminated altogether.

Just wanted to mention that I fooled around with amp #2 a bit more today and found that Vgg could be optimized further. With R4=100K and R5=15K, Vs=1.57V, Vgg=1.2V, Vd=5.7V and the max output before the waveform starts to get "bent" is 6V pk-pk. The gain and output impedance are pretty much the same.

Amp #2 is like the Oktava MK-219/319 head amp with different FET and bias.

What value for C1 ? 10uF ?

From Siliconix App Note #102:



I guess this is the same bias scheme without the source bypass cap and drain load resistor.

I would guess that the one that sounds better is the better one.

which one sounds cooler? Can you hear a difference using the same components (that are common to each)?

do I get to check them out?

dave

Dave,

I'm planning to use something along the lines of circuit #2 in a booster pedal I'm going to build--hence the 9V supply. This is to goose the front end of my 2-watt guitar amp when I want really insane overdrive. :grin:

I'll use the same basic config--with a higher supply voltage--in a mic pre or whatever, one of these days. But I'm knee-deep in vacuum tube projects for the forseeable future. There's so much shit I wanna build; I hope to get most of it done in this lifetime!

http://www.vishay.com/fets-small-signal/ssfanp/

an102 is very good. This is some of the best stuff for fets on the web all the ANS about fets there are good.

I´m very curious about the duals from VISAHY. Part U401, 404 and 406. Anyone tried it for a discreet opamp input stage?