Replacing a BJT with a FET

Thread Starter

jaydnul

Joined Apr 2, 2015
151
In a simple ideal BJT amplifier the input signal \(I_b\) is linearly amplified: \(I_{c}=\beta I_{b}\)

But if you replace the BJT with a FET, the input signal \(V_{gs}\) is amplified by the square: \(I_d=K(V_{gs}-V_{th})^2\)

So really you can't just replace a BJT with a FET, right? Do FET amplifiers all amplify by the square of the input signal? Is there a way to avoid that or is it just not that noticeable?

Thanks
 

KL7AJ

Joined Nov 4, 2008
2,208
In MANY cases, it's just not noticeable. However, this is not the only thing that makes it a bit difficult to just "plug" an FET into a BJT's socket. The impedance and bias levels are generally quite different.
 

ian field

Joined Oct 27, 2012
6,543
In MANY cases, it's just not noticeable. However, this is not the only thing that makes it a bit difficult to just "plug" an FET into a BJT's socket. The impedance and bias levels are generally quite different.
A JFET is easier to bias than a BJT - biasing a MOSFET in linear mode can take considerably more effort.
 

#12

Joined Nov 30, 2010
18,076
This is why negative feedback was invented. Even a bjt has a small difference in Vbe/Ic which causes a small distortion because it is of different magnitude depending on whether the input current is becoming more or less than the bias point. You can add another stage of amplification and feed that result to the emitter so the Vbe never changes. You can use a differential pair with feedback on the second base so the Vbe differences cancel each other.

Why would you want to bias a FET into linear mode? Shouldn't it be avoided?
Apparently you have never seen an operational amplifier built with mosfets.
 

AnalogKid

Joined Aug 1, 2013
8,103
Why would you want to bias a FET into linear mode?
Because deep down inside, all FETs are linear amplifiers. The boys at Bell Labs who invented the bipolar transistor were supposed to be working on a FET, but they couldn't get it to work so they tried something else. The FET was their boss's pet project, and dumping it for the BJT really pissed him off.

ak
 

KL7AJ

Joined Nov 4, 2008
2,208
Because deep down inside, all FETs are linear amplifiers. The boys at Bell Labs who invented the bipolar transistor were supposed to be working on a FET, but they couldn't get it to work so they tried something else. The FET was their boss's pet project, and dumping it for the BJT really pissed him off.

ak
This is a fascinating snippet of technological history. I've often wondered how things would have changed if the FET squeaky wheel got greased first. :)
 

#12

Joined Nov 30, 2010
18,076
This is a fascinating snippet of technological history. I've often wondered how things would have changed if the FET squeaky wheel got greased first. :)
It seems predictable. Engineers were accustomed to thinking in terms of voltage amplifiers (vacuum tubes) so accepting a solid state vacuum tube would have been very easy. The bjt would probably have been delayed for years by the ease of adapting to FETs. There is room to speculate that the bjt would never be considered necessary, but I believe mere curiosity would have outed it eventually. Just consider the strange devices that were invented (or designed) like the UJT and the SCR, tunneling diodes, Shottky diodes, and zener diodes. Did people think of a need and then design them, or just math around and experiment until something useful was discovered? Probably some of each.
 

GopherT

Joined Nov 23, 2012
7,983
Did people think of a need and then design them, or just math around and experiment until something useful was discovered? Probably some of each.
For the diodes, I am betting that they just made a bunch of devices, tested responses and they became what they are known as today. Some tweaking here or there to see how to improve the response they did get - tuning Zeners and so on.

The SCR is a 4 layer device so it was probably designed with some thought. No idea on the UJT.
 

GopherT

Joined Nov 23, 2012
7,983
Because deep down inside, all FETs are linear amplifiers. The boys at Bell Labs who invented the bipolar transistor were supposed to be working on a FET, but they couldn't get it to work so they tried something else. The FET was their boss's pet project, and dumping it for the BJT really pissed him off.

ak

I heard that the manager doesn't spend any effort dwelling on it today.
 

ian field

Joined Oct 27, 2012
6,543
Why would you want to bias a FET into linear mode? Shouldn't it be avoided?
They've got huge gain compared to a JFET, but you need a BJT added as a current control to make it stable - cutting out the AC nfb is tricky, but do-able.
 

recklessrog

Joined May 23, 2013
988
Because deep down inside, all FETs are linear amplifiers. The boys at Bell Labs who invented the bipolar transistor were supposed to be working on a FET, but they couldn't get it to work so they tried something else. The FET was their boss's pet project, and dumping it for the BJT really pissed him off.

ak
I heard something along those lines when I was apprentice with a British semiconductor manufacturer but no-one was ever able to confirm it as true. Interesting to have it verified 50 years after I first heard it!
 

recklessrog

Joined May 23, 2013
988
It seems predictable. Engineers were accustomed to thinking in terms of voltage amplifiers (vacuum tubes) so accepting a solid state vacuum tube would have been very easy. The bjt would probably have been delayed for years by the ease of adapting to FETs. There is room to speculate that the bjt would never be considered necessary, but I believe mere curiosity would have outed it eventually. Just consider the strange devices that were invented (or designed) like the UJT and the SCR, tunneling diodes, Shottky diodes, and zener diodes. Did people think of a need and then design them, or just math around and experiment until something useful was discovered? Probably some of each.

A bit like the discovery of Laser, I heard that even though they had produced it, initially they didn't know what to use it for!! But maybe that was a myth. I well remember seeing holes being punched in razor blades and bricks being blasted by "invisible heat rays" It seemed that the "Holy Grail" of a death ray had been found.
Seems we do have some things to thank the cold war for!
 

AnalogKid

Joined Aug 1, 2013
8,103
I heard that the manager doesn't spend any effort dwelling on it today.
Probably not - he died in 1989.

The transistor made the team very famous very fast. Shockley continued development of the FET in secret, got it working, and still could not patent it because some little Canadian weasel named Julius Lilienfield patented something close to it - 20 years earlier.

Also, the original Bell patent applications for the transistor did not have Shockley's name on them, which honked him off even more. He wanted them re-written with his name only. The famous trio was a compromise.

ak
 
Last edited:

OBW0549

Joined Mar 2, 2015
2,984
Why would you want to bias a FET into linear mode? Shouldn't it be avoided?
Not at all. For some linear applications, MOSFETs are ideally suited and they work splendidly as linear devices. They are somewhat tricky to use, though, in part because they just love to oscillate at RF frequencies if you give them half a chance. Having lots and lots of gain at high frequencies is not always "a good thing."
 

crutschow

Joined Mar 14, 2008
23,349
..................
So really you can't just replace a BJT with a FET, right? Do FET amplifiers all amplify by the square of the input signal? Is there a way to avoid that or is it just not that noticeable?
The square-law function of the FET will cause noticeable distortion in a linear amplifier but a sufficient amount of negative feedback will make that distortion tolerable, and such feedback is always the case when FETs are used as amps.

BJTs often are used as a transimpedance amp (voltage-controlled current-source) for AC amplifiers and that also has a square-law transfer function.
 

ian field

Joined Oct 27, 2012
6,543
Not at all. For some linear applications, MOSFETs are ideally suited and they work splendidly as linear devices. They are somewhat tricky to use, though, in part because they just love to oscillate at RF frequencies if you give them half a chance. Having lots and lots of gain at high frequencies is not always "a good thing."
In most cases; the tendency to oscillate can be subdued by adding a "gate stopper" resistor.
 

OBW0549

Joined Mar 2, 2015
2,984
In most cases; the tendency to oscillate can be subdued by adding a "gate stopper" resistor.
Yup. That's what I meant by "tricky." Most of the time it seems like 100Ω is adequate to squelch the RF oscillations.
 
Top