Electronics > Beginners

JFET mysteries

<< < (3/5) > >>

T3sl4co1l:
Yes, precisely: the reference terminal is the lower of the two channel terminals.

Same happens for invertible BJTs, you can swap emitter and collector because it's a symmetrical N-P-N structure (not usually actually symmetrical, due to doping differences -- it's not common, but symmetrical parts are possible), and the emitter is again the lower of the two main terminal voltages.

The only devices this isn't true for, are: thermionic tubes (no reverse conduction), the SCR (no reverse conduction), and the TRIAC (always common to MT1 -- the gate equivalent circuit is a back-to-back diode to MT1, so its voltage is never very far, and current flow in either direction will activate MT1-MT2 latching, to varying degrees of success depending).

Tim

hirada:

--- Quote ---Not too sure of your objective, whether it is to understand JFETs or high side switches.

--- End quote ---

Both. I came along JFETs by searching for a low latency high side switch solution for small signals, eventually driven by a push pull output to be used as a (not really precise) buffer or impedance changer. There will be a load in parallel to the R/LED, that is why I require high side switching.

Preferably a high side switch, that works, as one would expect: A "high" on the base/gate turns on, a low off and the voltage drop being more or less constant with variying load (within in a certain range, 1-30mA in this very case). Kind of like a relay really, but for energy reasons, I would like to stay clear of a dedicated IC switch for this presumably simple task, I've once  thought, this would be.

But with this final bastion of hope has fallen now, I've learned, this is not possible with silicium. Unless you build more complex units like a npn driving a pnp or p-mos or bootstrap/pump an n-mos  - as your graphics show as well: The p-mos is plain in the wrong  state with regards to the gate. Back in the days I was hoping to fix this with symmetrical JFETs.
Or of course by accepting a higher voltage drop I could use a npn alone. All quite unelegant solutions to the simple task of switching, but thats our choice. So my next thread will be about mosfets. And push-pull outputs.

As you've written, mosfets are more easily found for larger currents.* But then they may need a driver and have way slower turn on/switching times due to larger gate capacities (though I do not really have a need for that extra high speed, except enjoying a clean square on the scope and maybe the learning effect by taking propagation into account from the beginning). On the first glance, JFETs looked quite simple and less demanding than mosfets. I may have squinted here quite a bit.
However, I have a couple of p-mos on my list that could fit my bill to test next.


--- Quote ---the other thing is that the voltage necessary to turn a JFET off varies wildly,

--- End quote ---
Yes, I've noted that too. None of the JFETs I've tested would really turn off at 0V. The 2N3819s were the best/most consistent, but still did not turn off completely. But that was not as mysterical to me as the forward gate current, but more a matter of wrong reading of specs.


--- Quote ---For switching things like an LED MOSFETs are usually better suited.

--- End quote ---
My next step. Though quite surprising, as integrated switches and lots of other chips are made of JFETs. And a lot of those seem to be able to work fine without a negative voltage. Where I was not even able to fully turn of my few devices with 0V. But then again, I am way less complex.
 

--- Quote ---Yes, precisely: the reference terminal is the lower of the two channel terminals.

--- End quote ---
Thank for confirming this again. Though bad news. 

*[Edit]: And it is not that easy to find those that actually do work properly with logic levels, especially for p-mos. But that did not really work out well for my JFETs either.

Kleinstein:
The silicon JFETs are always depletion mode. So they are on at zero G-S voltage. As an odd point, there are SiC and GaN JFETs, that are actually enhancement mode and thus of at zero voltage. But these are expensive high voltage parts.

The silicon JFETs are usually low current (< 50 mA). The small discrete MOSFETs are similar resistance and also similar in gate capacitance, but there are larger ones available. Really small ones are rare. So the driving is not that much easier with JFETs, it's different voltage levels however.
The integrated switches usually use MOSFETs where substrate and source are separate. Such MOSFETs are also available as discrete parts, but rare. There are a few older Chips with JFETs too, but these are more like the exception.

For a normal high side power switch the obvious choice is a P-MOSFET. Driving is not that much different from an N-JFET - mainly opposite logic.

Nerull:
n-channel jfets work by choking off conduction through the n-channel by creating a depletion region, this requires reverse-biasing the PN junction, just as it does in a diode. You aren't going to find one that turns off at 0V because 0V is their fully on state, they turn off with a more negative VGS and should never be driven with a positive VGS, since this drives the PN junction into forward bias and the gate begins to conduct.

hirada:

--- Quote ---The integrated switches usually use MOSFETs

--- End quote ---
Right. I was just thinking about the TL072CN and the AD SW06, both of which I am playing with right now. But those are a minority. Thanks for the heads up.


--- Quote ---You aren't going to find one that turns off at 0V because 0V is their fully on state

--- End quote ---

I've had them on the high side. So the source was not at 0 Volt, but around 2-3V. Anyway, still not enough. Which is fine, for one model I remember the sheets specified a minimum of -1V, but up to -6V.


--- Quote ---VGS and should never be driven with a positive VGS, since this drives the PN junction into forward bias and the gate begins to conduct.

--- End quote ---

I believe, finally I've gotten it now. I suppose, this is the gate forward current mentioned in the datasheet. Makes sense now, since, after disconnecting V+ from drain, Vgs rose from 0.1V to 0.7V - and the gate current rose to the draincurrent.


Thanks very much to all for helping me out!

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod