Jfet Cut-Off etc

[sbirkenstock]

Hi Everybody,

I tested the Vgs Cut-Off voltage on a couple of 2n5457.
I can measure -0,35V with a drain resistor of 3,9k.
With a drain resistor of 10k, it actually goes up (or down...) to -0,45V.
I don´t quite understand that difference.
The data sheet of On Semiconductor says -0,5 to -5V.

Does everybody know a Jfet that has a minimum cut off of -1V or -2v or so,
for general amplification (like a booster).
I had some 2n5457 that would "bias" the drain at about 4,5V (9V V+) with a 10K drain resistor and a 1,5k source resistor.
That would leave the Jfet self biased and the gate grounded via a 1meg resistor.
No bias voltage required and no cap on the input therefore.

I tested 5 2N5457 by the way, all resulting in almost the exact cut-off voltage.
I read that even one production run can vary extremely, my examples do not.


Also somewhere I read that Drain and Source can be exchanged (at least on certain Jfets???).
Is that correct?

Best regards,

Stephan

[Rob Strand]

The simple view of a BJT is it off when the base-emitter voltage is less than say 0.6V and on when the voltage is greater than 0.6V. 

A JFET doesn't work like this at all.   As the gate-source voltage becomes more negative the JFET conducts less and less.   The relationship between current and voltage is a kind of sloppy curve which you can see on some datasheets.   Compare at to 10k a 3.9k resistor operates at a higher current so the Vgs voltage is closer to zero (ie. a smaller absolute magnitude).

Neither the 10k or 3.9k resistor case has a Vgs voltage called the cut-off.  If you put a very large resistor in there, say 1MEG or 10MEG, then the current will be small and the Vgs voltage will be closer to the cut-off voltage.

A Vgs voltage near zero will conduct high current and a Vgs voltage at the Cut-off voltage will conduct, in theory, no current.

(Now,  in reality, the the base-emitter voltage of a BJT also depends on current.  It is higher at high currents than it is at small currents.  However if you change the current by a factor of 10 the base emitter voltage will only shift by 0.06V or so.  Quite small, so we normally ignore this shift.)

[anotherjim]

J201 is usually easier to bias in 9v circuits - though you may still get some of those that don't play so nice.

On datasheets, you may see that the SMD packaged one actually states that drain & source can be swapped. It's the same spec as the TO92 so they are all interchangable.

UK, I notice BitsBox are selling J201 smd pre-mounted on an adapter for reasonable money.

[PRR]

> somewhere I read that Drain and Source can be exchanged (at least on certain Jfets???).

It says right ON the 2N5457 datasheet "Drain and Source Interchangeable".

> measure -0,35V with a drain resistor of 3,9k.

That is an unusual way to measure Vgs(off).

"Off" means off; or because "true off" is hard to measure, "a very-very small current". Often 50uA for small JFETs, though 2N5457 uses 10nA.

Vgs also matters, should be constant at the value on the datasheet. 2N5457 uses 15V. But we see that the drain curves are fairly flat, so a somewhat lower voltage is OK for quick checking. The much lower voltage you get with a large Drain resistor is NOT how you measure Vgs(off).

Set up a grounded Gate Source follower. Drain to +9V. We expect a Vgs(off) of 0.5V to 6V, take 3V as the average. The specified 10nA implies a 300Meg(!!) Source "bias" resistor. You can't measure that. So just put your 10Meg DVM in the Source circuit and measure at 300nA.

[TejfolvonDanone]

No bias voltage required and no cap on the input therefore.

I feel i need to point out that if you don't use a coupling cap and there is a faulty equipment connected (like a leaky output cap) to the input can make some trouble. In some cases it may even damage the JFET.

Does everybody know a Jfet that has a minimum cut off of -1V or -2v or so,
for general amplification (like a booster).

Unfortunately JFETs have really large tolerances in terms of Vgs(off) and Idss. So max specifications may be 5-10 times the min specs.
There are several techniques to bias JFETs: link

Also somewhere I read that Drain and Source can be exchanged (at least on certain Jfets???).
Is that correct?

All the D and S are interchangeable because of the JFET's internal structure. Some manufacturer states this on the datasheet some don't.

[PRR]

> D and S are interchangeable because of the JFET's internal structure

Quite true for anything anybody here needs.

For the *highest* radio frequencies, a few types are "asymmetrical". Where a common symmetric JFET has 3pFd G-S and G-D, an asymmetric JFET may have 4pFd G-S but 2pFd G-D. At high radio frequencies (many hundred MHz), this 1pFd difference may add another third-octave of response. And installed "backward" would shave a good bit of bandwidth.

In audio we have dozens of pFd everywhere so a lousy 1pFd difference is no difference at all. Even if you got an asymmetric JFET you could not tell or care which was S or D.

> faulty equipment ...can make some trouble. In some cases it may even damage the JFET.

It would be Good Practice to always have 33K or so series resistance to a naked Gate. A leaky source would still mis-bias the FET out of happy-zone. However 33K will give significant reduction of radio signal detection. And since most JFETs can stand 10mA, you can stuff in 330V either way and the JFET may survive.

[Rob Strand]

Many moons ago I measured the parameters in each direction of a handful of JFETs.
I can't quite remember all the details but most were symmetric. 
It may very well be that the RF types PRR mentioned were asymmetrical ones.
The JFET parameters of the symmetric types were very close in each direction - which you would expect.

[PRR]

The RunOffGroove JFET tester is a good simple reference test.

Bottom of http://www.runoffgroove.com/fetzervalve.html

[Eb7+9]

10Meg is useless (un-necessary) value if meter's internal resistance is not at least 100Meg in voltage mode
most DMM's don't do much better than 1Meg ...

just use the meter's resistance directly, no external/circuit resistor needed,
as shown in the Paradigm Shifter thread ...

http://www.lynx.net/~jc/jFETVpIdss.jpg

[merlinb]

just use the meter's resistance directly, no external/circuit resistor needed,

+1. Most of the JFET testers floating around this forum are way over engineered.

(Set meter to milliamps to measure Idss)

[PRR]

> way over engineered.

Or just not measuring the right thing. Like the 3.9K-10K in the Drain plan.

Whether it has one or two extra 12-cent resistors is probably not a real problem.

[Eb7+9]

+1. Most of the JFET testers floating around this forum are way over engineered.

... or, plainly whimsical ... like hard driving a 400uA current thru a jFET and measuring its Vgs voltage as a means of "matching" // ... then referring to devices who's Idss spans 200uA-to-1000uA as "funny"

[R.G.]

... or, plainly whimsical ... like hard driving a 400uA current thru a jFET and measuring its Vgs voltage as a means of "matching" // ... then referring to devices who's Idss spans 200uA-to-1000uA as "funny"

Just can't leave it alone, can you J.C.?

This would be the 157th last time you comment? Or 158th?

[R.G.]

just use the meter's resistance directly, no external/circuit resistor needed,

+1. Most of the JFET testers floating around this forum are way over engineered.
(Set meter to milliamps to measure Idss)

Nota bene - what's actually being measured in the "most of the JFET testers floating around this forum" is Vp.

[Eb7+9]

... what's actually being measured in the "most of the JFET testers floating around this forum" is Vp.

classic ... more snow from the master of arithmetic

This would be the 157th last time you comment? Or 158th?

[R.G.]

classic ... more snow from the master of arithmetic

OK - let's just say that's 159. I figure less than a day to 160. 

Shall I replay your comment about leaving this alone to you? I have a copy of that if you've forgotten. 

[merlinb]

OK - let's just say that's 159. I figure less than a day to 160. 

Have I just stumbled into some kind of feud?! 

[EBK]

OK - let's just say that's 159. I figure less than a day to 160. 

Have I just stumbled into some kind of feud?! 

Entire wars are fought over JFET testing around here.  It's the reason I will never build a phaser -- too dangerous. 

[Cozybuilder]

OK - let's just say that's 159. I figure less than a day to 160. 

Have I just stumbled into some kind of feud?! 

Entire wars are fought over JFET testing around here.  It's the reason I will never build a phaser -- too dangerous. 

[R.G.]

Have I just stumbled into some kind of feud?! 

Yes, you have.

J.C. seems to be offended by the JFET tester I put up at geofex about 14 years ago. From time to time, J.C. will pop in here with some remarks intended to be scathing about how the tester isn't good enough in some way, doesn't measure what it says it does, can't work well, and any other criticisms he can think of at the moment.

Back when this got started - shoot, I don't know, maybe ten or more years ago - I explained to J.C. and anyone else that would read it that the JFET tester was by no means perfect, but it did a really fast and decent job of sorting JFETs into bins for use in phasers. And it does, as evidenced by whole generations of effects newbies using it or one of the meeeee too improvements that popped up. Near as I can tell, J.C. may be the only person who's offended by the circuit.

The circuit, by the way, is an adaptation of a Vp tester that is old, but that I found in "Applications of Operational Amplifiers" by Jerald Graeme. The opamp is to remove errors in the 'meter only' way of measuring this.

[And for J.C.'s benefit, as he'll certainly be reading this, yes, I know that Vp and Vgsoff are not technically the same, just very close to one another in value. ]

Having explained this gently at first to J.C. in public several times, I gave up and became offended too.    So I'm not too polite when he feels the need to have another run at it.

Historically, the scenario unfolds like this. J.C. is quiet and leaves this alone for some longish period of time, months or maybe a year at a time. I'm fine with leaving it alone, so I don't mention the JFET matcher, other than to answer questions people have about making it work. Then J.C. will post a new blast of criticism of the matcher. I respond. We trade a few messages, and when J.C. drops the issue, I do too. The process is reset.

After each of these cycles, I have some hope that J.C. will just leave it alone. It's a circuit, for goodness' sake. It either works or not, and history says that this one does. In the past, J.C. will be quiet for a while, then - something happens, the pressure builds up, whatever, and he's back here with another blast. The cycle is long enough that a whole generation of effects newbies may pass through before he erupts again.

I have encouraged J.C. to do a better matcher that's more perfect in some way that pleases him, just so he'll leave it alone. The "paradigm shifter" seems to be the result of some of that, and I wish him well with whatever that turns out to be. But as you can tell from his sniping, that's apparently not enough.

It does seem to encourage J.C. if he can get someone to agree than the JFET tester could be better. Currently he's on a run with you and Paul making comments about JFETs. Getting any kind of public support seems to keep him running on longer.

I wish the guy would just leave it alone. He even promises to do that, as you could see from the most recent eruption sequence. He went so far as to say things to the effect that 'this is the last I'll say about this', which I thought was great, but didn't really hold much hope for. And sure enough, he is still on the latest rant. That's the origin of counting the number of "last times" thing I say.

So yes, it's a bit of a feud. I'd be happy to let it rest. But I think we're going to have to let J.C. get this most recent episode out of his system. It'll be over a lot faster if people don't feed the troll.