"Real World" JFETs

[JDoyle]

The devices do have quite a range of IDSS, so much so that I'm 80% confident that any J201 would bias to within 5% or 10% and work.  No other active devices were used in the biasing arrangement.  OK, it does depend on your method of biasing but ways of avoiding trimmers in JFET circuits have been much discussed here.

As I was quoting percentages, and therefore statistical CHANCE, you won the bet with those 10 J201s. If you were to get a batch of 100, or 500, or 1000 J201s, the more you got, the more my confidence would rise to 100% that you would only have 20% that work.

Or you have a circuit with a small load resistor (so that changes in Id don't change the bias much) and therefore very little gain, if not a gain of less than one.

Jay Doyle

[R.G.]

The devices do have quite a range of IDSS, so much so that I'm 80% confident that any J201 would bias to within 5% or 10% and work. 

Tee hee... are you 80% confident enough to bet some cash on that? I'll supply the J201s... 

[alanlan]

The devices do have quite a range of IDSS, so much so that I'm 80% confident that any J201 would bias to within 5% or 10% and work. 

Tee hee... are you 80% confident enough to bet some cash on that? I'll supply the J201s... 

I've just come over all sweaty. 
I'm 100% confident that 80% of my statistics are at least 10% accurate. 
According to the datasheet, J201's IDSS vares from 0.2 to 1.0mA.  My batch varies from 0.4 to 0.9, so I might have trouble with a 0.2, but would I bet money on it?  Perhaps a little...

[alanlan]

The devices do have quite a range of IDSS, so much so that I'm 80% confident that any J201 would bias to within 5% or 10% and work.  No other active devices were used in the biasing arrangement.  OK, it does depend on your method of biasing but ways of avoiding trimmers in JFET circuits have been much discussed here.

As I was quoting percentages, and therefore statistical CHANCE, you won the bet with those 10 J201s. If you were to get a batch of 100, or 500, or 1000 J201s, the more you got, the more my confidence would rise to 100% that you would only have 20% that work.

Or you have a circuit with a small load resistor (so that changes in Id don't change the bias much) and therefore very little gain, if not a gain of less than one.

Jay Doyle

Well, the load resistor has very little to do with it since I was measuring ID and not VD.  And, as my batch covers over 60% of the range of IDSS  quoted by the manufacturer, I don't think I'd be way out.

[JDoyle]

Well, the load resistor has very little to do with it since I was measuring ID and not VD.  And, as my batch covers over 60% of the range of IDSS  quoted by the manufacturer, I don't think I'd be way out.

I thought you said that they 'bias up to within 5% with fixed value resistors'? Actually, you did.

I was assuming that you had them in a circuit and weren't just stating your luck with the bell curve vis a vis Idss...

And if you did have a load resistor, then yes, it would have something to do with it - a range of 0.2 mA to 1 mA makes a lot less difference in VD if you have a 1k resistor (dropping 0.4V to 0.9V in your batch, a difference of 0.5 volts, enough to be happy, but with a gm of under a mmho, a gain of less than one) versus a 10k (4V to 9V with your batch, a difference of 5 volts - I think that makes a difference in biasing, but then again, I can be a bit picky), unless you think that 'biasing up' only has to do with Id and nothing to do with Vd (and therefore Vds, headroom, gain, or the actual real world operation of a circuit)...

I'm pretty confused...and this time I'm pretty sure it's not because I'm not an EE...

I'll stand by my statement - you got lucky.

[alanlan]

Yes, I measured them all in a real circuit.  One of my own designs as it happens.

I don't get your point about drain resistors.  Percentage is percentage right?  If I design to bias nominally at 5V and it's 5% out then that sounds pretty good to me.  ie. 5% ID error = 5% VD error?

[aron]

Other than Mu/Syrrp circuits, I haven't had much luck getting Jfets to sound as good as other distortions types.

Interesting.

With very few exceptions, every circuit that I love and use has a JFET in it. Despite all the problems that have been stated, I have been able to get my "personal" circuits to sound quite good using trimmers and sometimes using fixed resistors. I use a well-worked variation of the Shaka live for years and I haven't used anything better than this circuit. No other circuit I have used has lasted this long for me. I have tried over and over to replace it with some other circuit, but the JFET rules for me.

I heard someone play my rig last night and my suspicions have been confirmed. I will leave the discussion on whether we should use them to others.

[R.G.]

You see, that's the silly thing aron - OF COURSE we should use them. Use whatever works.

I had a friend named Albert that once said "Everything should be as simple as possible - but no simpler."

The Right Way is to learn what is and is not, what the variations are, and work with them.

[aron]

I just noticed that my Pignose 150R used a trimmer on the source for the FET stage. Interesting....

http://www.diystompboxes.com/amps/pignose150r/pn1.jpg

[JDoyle]

I'm sorry that my post didn't explicitly say that I wasn't discouraging the use of JFETs in anyway - in my experiments I almost always am using JFETs - my post above was just an attempt to summarize the whys and wherefores of the difficulties involved when using JFETs - which can probably be summed up like this:

Before you build or design any circuit that uses JFETs, build or design a circuit that allows you to SORT and SELECT JFETs for specific characteristics.

I also have one addendum to the original post - my comments about the Idss suffixes for Japanese JFETs only included those used by Toshiba (O, Y, GR, BL, V in order of increasing Idss) - other Japanese manufacturers, such as Sony, use a system similar to that used by the Europeans (who use both the 2NXXXX system as well as the BFXXX system) which ranks increasing Idss in alphabetical order (A, B, C, etc.). There are also situations that are a mix, but are normally a selection process for noise, and this is used mostly for older American AND Japanese JFET styles - some part numbers, like the 2SK68, also have a low noise version which has an additional 'A' suffix, unfortunately. So the 2SK68 is described as a 'low frequency amplifier', with the 2SK68A as a 'low noise, low frequency amplifier'.

And to bring it all back around to the question about the differences between the 2N5485/86 the following is a list of all of the JFETs that are made using the same process, a process designed mostly for VHF/UHF amplifiers (this is the Siliconix list from 1989):

TO92 Package: 2N3819 (a classic), 2N5484, 2N5485, 2N5486, BF244A, BF244B, BF244C, BF245A, BF245B, BF245C, J304, J305, PN4416
TO72 Package: 2N4416, 2N4416A (a similar situation to the 2SK68/2SK68A mentioned above, the A suffix is a lower noise version)
Surface Mount Package (SOT-23): SST4416

Again, I completely encourage the use of JFETs, I just wanted to point out why they are difficult to work with, at least in terms of consistency.

Regards,

Jay Doyle

[DougH]

Yes, "Jay Fet Doyle" loves him some FETs...

[JDoyle]

Yes, "Jay Fet Doyle" loves him some FETs...

As corny and weird as the following may seem, I do have a childlike fascination with the semi-mystery of JFETs .

Here we have something that by every analogy available is easy to understand, the gate is a faucet and the channel the pipe; or the drain-source channel is a garden hose and the gate is your hand tightening around it as the voltage goes down - so it would seem that it should translate so easily to real world parts that work just like that - which they do, just that they are all, in a way, unique.

Then you have a part, the BJT, which is really hard to wrap your mind around - there isn't any real world analog; the best has to be H&H's 'Transistor Man' in The Art of Electronics but that requires us to give nature/current some sort of thought process or sentient ability, which is probably the most revealing fact of all... But either way, in comparision to anything in our physical world, the way a BJT works is more akin to magic.

It's as if the only way to turn the faucet on harder is to stick a tiny hose, or squirt a squirt gun, into a small hole on the faucet itself - and that action then opens up some watery 'trap door' in the faucet that allows the release of a larger flow of water which is proportional to the water INTO the faucet from the hose/squirt gun (but in reality in proportion to the pressure behind the water injected!).

Yet the above completely foreign concept, when applied to semiconductors, allows the creation of a useful, real world part that can be reproduced again and again with almost the exact same result in every part.

To me it is both a window into the weird and backwards world of Quantuum Physics and real world proof of how strong the elemental function of electricity is - the sheer overwhelming strength of current's 'need' to get to ground means that, when it comes to semiconductors, a forward biased diode, which is an open door for current to get to ground, or at least closer to it, nearly always takes the same form and overall function. In fact, all it takes is a little, little bit of that current that biases a diode on (from the base to emitter) and a seperate and comparatively massive amount can exploit that path and come crashing through (from the collector to the emitter) and AROUND another adjacent and attached diode (from base to collector), which by all accounts should remain a barricade in its path.

But if one tries to STOP current from getting to ground, then that same current's 'need' to get past our attempts to stop it from getting there mean that anything and everything we put in it's way, from the difference in voltage, to the electron/hole concentration on each side of the pn junction, to the real world and miniscule impurities in the path to ground itself - the semiconductor material - will have an effect on your ability to do so - and therefore make the task and the results of our actions difficult to predict from one attempt to the next.

Each time we fire up a circuit we are in control of our own microcosm of a part of nature that doesn't work like ANYTHING else we experience in our walk on this earth in any tangible manner, yet at the same time is essential to everything around us.

And instead of us choosing to hitch a ride with nature, like the current we put through the base of a BJT, a JFET allows us an opportunity to personally grab ahold of nature and use itself to control itself; to try wrestle it to do our bidding.

I guess if we have the hubris to believe that we can actually do that, we almost ought to EXPECT it to be difficult to duplicate and be happy that we can even do it at all!

However cheesy that may seem, I think it is pretty damn cool.

Not to mention, JFETs DO sound pretty damn could when you slap them around with a guitar signal. 

Regards,

Jay Doyle

[WGTP]

http://diystompboxes.com/pedals/schems/Shaka%20Discrete.pdf   

[MicFarlow77]

I had to bookmark this one. So much good info here that for me, understanding it will take more time, study and experience.

Thanks all for your input and for sharing!

Mick