Hello everyone out there, I am playing around with fets and need some help. When the circuit below clips on the positive halve of the output, it clips rather sharp:
Now, when I substitute part of the source resistance with a couple of schottky (or germanium) diodes, the sharp edges get smoothed:
The question is, what exactly is happening here?
Looking forward to your replies, and sorry for the rather large pictures.
Now, when I substitute part of the source resistance with a couple of schottky (or germanium) diodes, the sharp edges get smoothed:
The question is, what exactly is happening here?
Looking forward to your replies, and sorry for the rather large pictures.
Attachments
Bias the fet so it has a lower drain DC voltage, and clips more symmetrically.
Do not use diodes in the source, that will cause higher distortion.
Do not use diodes in the source, that will cause higher distortion.
I biased the jfet like that on purpose, at the moment I am only interested in what happens to the positive side of the waveform. Also, this is not intended as a hifi project, so no worries about higher distortion.
But thanks for the reply anyway.
When the FET nears clipping, it is close to cutoff and its drain and source currents become very small, which increases the dynamic resistance of the diodes, reducing the effective transconductance and thus the gain of the stage
Thanks for your answer. So if I understand you correctly, the diodes create a form of automatic gain control. Now I am curious to learn more about the dynamic resistance of diodes. Or as I once read on a professor's door:
"knowledge is an ever growing island in the ocean of the unknown, with an also ever growing coastline of questions".
As a first approximation, the dynamic resistance of a junction is ~=26/If (If being expressed in mA). It is thus inversely proportional to the forward current, or in other words it depends on -1 power of the current.
In reality, the exponent is not exactly -1, but it varies according to the technology, process, etc., but generally this approximation is sufficient for practical purposes
In reality, the exponent is not exactly -1, but it varies according to the technology, process, etc., but generally this approximation is sufficient for practical purposes
The source resistor determines the local feedback. The more feedback, the closer the response to a geometric slope vs an S curve. Soft clipping comes from distortion that happens as the fet approaches cut off. Feedback causes it to remain (more) linear to the last.
Short answer: slamming against the +V rail is more abrupt, you just slammed against a wall and can´t go no further; while at the negative peak transistor saturates, not the same, and a more gradual process.
Notice it never reaches zero volts but stops near it.
By the way, triodes do about the same thing.
So much for the theory "tubes clip different from SS"
In basic no feedback circuits such as these, not THAT different.
Notice it never reaches zero volts but stops near it.
By the way, triodes do about the same thing.
So much for the theory "tubes clip different from SS"
In basic no feedback circuits such as these, not THAT different.
Either AC couple the input and set the operating point so the clipping goes away, or use dual supplies so the input can swing in either direction without running out of headroom?
Oh, using higher voltages he will get higher Vpp but the waveforms and basic clipping action will remain the same: top flat clipping, bottom gradual saturation.
It´s in the nature of the beast.
Triodes fed from 100-200V supplies do the same.
I guess that is the background of the question, not "I get 8V RMS, I need 16V RMS" or so.
It´s in the nature of the beast.
Triodes fed from 100-200V supplies do the same.
I guess that is the background of the question, not "I get 8V RMS, I need 16V RMS" or so.
The background of the question is how to make a jfet clip like a triode, for now I am focussed on the positive side, and I know a way towards this is to put some diodes between source and ground. But I want to know what I am doing, how it exactly works.
Thank you all for your replies so far!
Thank you all for your replies so far!
Germanium diodes open around 150mv but you can't really use them to autobias every j fet as the low noise ones have very low vgs in normal operation range just around one germanium diode junction voltage . With jfets you can also use the subtreshold region properties to play with using them naturally at very low currents and pretty high value source resistors that will create feedback, but won't be very effective in that area either.Never done it but I've heard a few people talking about reversing jfet drain and source or bipolar trz collector with emitter and see more interesting effects when hfe or transconductance falls.