Virtues Of Paralleled FETs?

[Paul Marossy]

On the Cranked AC schematic on page 7 HERE, there are paralleled J201s. I've been wondering about this for a while, what benefit is there to having the FETs paralleled in such a manner?

[Gus]

http://www.fairchildsemi.com/ds/MM/MMBFJ201.pdf  Note J201 J202 J203 are in the datasheet
One J202 might be better look at the Yfs specs
Parallel devices often need to be matched and sometimes need separate source or emitter or cathode resistors to help make sure one device is not doing all the work.

[Paul Marossy]

I was thinking maybe it was an attempt at doing a paralleled triode thing like Matchless does. Or because it was thought that devices with a high degree of variability would work better if in parallel.

[Gus]

Or because it was thought that devices with a high degree of variability would work better if in parallel.

Where did you read that?

[Paul Marossy]

Nowhere, I'm just making guesses as to the reasoning behind doing that.

[Gus]

Don't quess do the math.

Is the IDSS higher than the max drain current in the stage J201 is .2ma to 1ma. Some fast math 8VDC/.2ma = 40K   8VDC/15K=.5ma  SO maybe it could be they bought a lot of J201s and used two to make sure the IDSS was greater than the Drain current
J202 IDSS .9ma to 4.5ma

Do you want more input capacitance? maybe use another Jfet like a J202 instead of a J201
Do you want more transconductance use another JFET., a J202 is speced at 1000Yfs vs the J201 500Yfs

The above is some guesses based on the specsheet and schematic.
Maybe ask the DESIGNER.

[Paul Marossy]

Maybe ask the DESIGNER.

I guess I should. I know Brian (Wampler) loves J201s, he used them in many of his designs. But the Cranked AC is the only one I've seen where he had paralleled J201s like that.

[Bill Mountain]

I've heard paralleling lowers noise in BJT's.  Maybe it's the same with JFETs???

In a perfect world all JFETs would be on spec.

In a less than perfect world you could parallel to average the performance.

In our world?  It makes little sense in effects circuits.

[Paul Marossy]

In a less than perfect world you could parallel to average the performance.

That was my thinking, although it seems to have been misinterpreted as something else.

[Bill Mountain]

In a less than perfect world you could parallel to average the performance.

That was my thinking, although it seems to have been misinterpreted as something else.

I meant that it would be cool if it worked like that but it doesn't.

I don't see the point because it has no noticeable effect in the ways we use it.  You'd still have to bias it correctly and one of the JFETs would end up working harder than the other.  There would be no sharing of responsibilities.  This is not an informed/schooled opinion.  This is based on experiments I have done myself.  One tranny or two, it makes no noticeable difference to me.

I do know that you can increase current capabilities but I think a you could always find a single transistor that would work as well as several paralleled JFETs.

[Gus]

did read this?
http://www.fairchildsemi.com/ds/MM/MMBFJ201.pdf

Did you read my post that you should match || devices otherwise one device does most/all of the work.  If you look at power amps with multiple BJT output devices you will see emitter resistors to help force the device to share power delivery.

Thing to think about and search for
Would one J201 with an IDSS of .2ma work in this type gain circuit.  What is the max Id (drain current) of each stage?  How will J201 with different Vgs and/or Yfs measured parameters work together?

[Paul Marossy]

Yes, I know that they should be matched. I found that much from a quick google search. But I can't find anything that applies to stompboxes, and hence my question. I'm hoping Brian will answer my email so I can see what his thoughts were on it.

[Bill Mountain]

Yes, I know that they should be matched. I found that much from a quick google search. But I can't find anything that applies to stompboxes, and hence my question. I'm hoping Brian will answer my email so I can see what his thoughts were on it.

I think I read somewhere (maybe the other forum) that he liked the sound of parallel JFETs.  I've learned over the years that not every design is perfectly electronically sound.  Sometimes the designer will do something simply because they think it sounds good even when it shouldn't make a difference.  Hence MOJO

[Paul Marossy]

That probably is the case I guess. Hope to get it straight from the designer though.

[Gus]

liking the sound of || jfets means nothing to me.

Try building this with one selected J202 per stage

Somethings else about the design
A 39K or greater drain resistor should work with one J201. 
Note the gain goes down as drain current goes up
The source resistors are cap bypassed for max gain with 15K drain resistor and two devices
So one device with a cap bypassed different value source resistor would something to try
You will need to adjust Rs and Cs between stages for the same frequency response because of the different drain resistors

[Paul Marossy]

I wasn't planning on building it, but I've played one before. They sound pretty good. But I have always wondered about this parallel J201 business.

[amptramp]

You don't have the same problem with paralleling JFET's as you do with bipolar transistors.  With a bipolar transistor, one slightly stronger transistor will hog the current because the higher temperature from more current flow will reduce Vbe and therefore, devices that start with nearly equal current will shift entirely to the transistor with lower Vbe.  This can only be corrected by separate equal emitter resistors.  There is no such effect with a JFET or MOSFET - the channel is modelled as a resistance with a positive tempco of resistance and if one starts to heat up, its resistance increases and it tends to share the current without the need for source resistors.  This of course implies the Vgs on voltages are equal to start with.

[Paul Marossy]

You don't have the same problem with paralleling JFET's as you do with bipolar transistors.  With a bipolar transistor, one slightly stronger transistor will hog the current because the higher temperature from more current flow will reduce Vbe and therefore, devices that start with nearly equal current will shift entirely to the transistor with lower Vbe.  This can only be corrected by separate equal emitter resistors.  There is no such effect with a JFET or MOSFET - the channel is modelled as a resistance with a positive tempco of resistance and if one starts to heat up, its resistance increases and it tends to share the current without the need for source resistors.  This of course implies the Vgs on voltages are equal to start with.

Ineteresting. So if you're using FETs from the same batch, the Vgs probably won't be that far off from one to the next. But still the question remains, why even do that in the first place? The only thing I can think of is to try and emulate parallel triodes with FETs. Matchless liked to do the paralleled triode on the input tube, I've always wondered about why that was done as well.

EDIT: So it looks like at least the input tube on a Vox AC30 is paralleled, because two in parallel will give half the noise and double the gm (gain). That much at least makes sense on the Cranked AC schematic if it's supposed to emulate an AC30 sound.

[Gus]

Again  IDSS needs to be higher than Id (drain current)

Read my posts again I did some fast math Ohms law with the use of a 15K drain resistor.  One J201 might not work if its IDSS is at the lower part of the range.
A J202 has X2 the transconductance as I pointed out if you look at the data sheet

One selected jfet should work the same as two || jfets

[amptramp]

Ineteresting. So if you're using FETs from the same batch, the Vgs probably won't be that far off from one to the next. But still the question remains, why even do that in the first place? The only thing I can think of is to try and emulate parallel triodes with FETs. Matchless liked to do the paralleled triode on the input tube, I've always wondered about why that was done as well.

EDIT: So it looks like at least the input tube on a Vox AC30 is paralleled, because two in parallel will give half the noise and double the gm (gain). That much at least makes sense on the Cranked AC schematic if it's supposed to emulate an AC30 sound.

Large MOSFET's are usually designed around a very consistent geometry and oxide deposition thickness, so various parts of the FET act like separate devices connected in parallel.  If you use two devices in parallel, the noise does not go to half, it goes to 1/root 2 or 0.707.  You need four devices in parallel to get half the noise.