Common-cathode triode vs source-follower Hexfet.

[Circlotron]

People crow about how triodes sound better than pentodes, (I refuse to say "penthodes" ) and indeed the plate curves are different. The pentode ones are more or less constant current for a wide range of anode voltage. That means the curve is a nearly flat horizontal line. On the graph, a change in horizontal voltage doesn't make much change in vertical current. This means of course that the pentode has a high output impedance, i.e. it is rotten for generating a specific output voltage but good for generating a specific output current.

Triodes on the other hand have a plate curve that starts off gradual and then slopes upward at a fairly constant rate. You pull the anode voltage across X volts and the anode current increases up by Y milliamps. Relatively low output impedance.

A Hexfet has a drain curve pretty similar to a pentode and so if we use it in a common source cct then it will exhibit current sink properties. But look a bit closer at the drain current vs gate voltage - the transconductance curve I think it's called. Doesn't it remind you of a triode plate curve? So then, how do we make use of that? If you have a Class A *source follower* driving your speaker then I expect it will mimic to a certain extent the characteristics of a single ended triode power amp. On positive excursions as the drain current increases it progressively follows the gate voltage more accurately (up to a certain limit anyway) and so our sinewave gets a pointy top. On negative excursions the opposite happens as the drain current decreases and as the curve starts to bend more horizontal our sinewave gets a bit flattened on the bottom. Pointy top and flat bottom = second harmonic distortion. Only a single ended source follower will make this. A common source will too but the high output impedance means the output is at the mercy of the load impedance and the voltage is all over the place. Complementary source followers (that means 99% of all amps) will make symmetrical distortion, the same thing on the top half and the bottom half of the waveform, which means *odd* harmonic distortion - not good for the ears.

Seems like the humble source follower without global NFB has something going for it then. 100% local NFB, gain of almost 1 so no stability problems provided you use a stopper resistor in series with the gate, light-speed risetime and frequency response. Dead simple.

I've got global feedback in my class A; I'll have to disconnect it and see how it sounds with a bit of second harmonic. Not much feedback though. Open loop voltage gain is 100, closed loop voltage gain is nearly 10. Is that 10dB or 20dB feedback? I always forget...

[SY]

You'll get the same advantages with a pentode cathode follower (take a look at the equivalent ip vs. eg curves). And a bipolar emitter follower will have even lower distortion and source impedance than a source follower because of the higher transconductance. Rise time in all cases will NOT be "light speed."

Design choices are based on other criteria.

[EC8010]

Power MOSFETs look terribly attractive until you look at the curves for gate capacitance. All of a sudden, they're really hard to drive...

[Circlotron]

That's true. I would expect gate capacitance is more of an issue in a common source cct than in a source follower cct. I suppose it all depends on just how much you swing the gate voltage with respect to cathode. And how fast.

[Sch3mat1c]

When you take a pentode and put it in CF mode, that doesn't make it a triode. Likewise, when you put a FET in SF mode, that doesn't make it a triode.

No offense, but that 2nd/3rd harmonic you mentioned is pointless, in that 1. yes, SE stages make 2nd harmonic as a result of the incremental Gm inherent in the device, and 2. 3rd harmonic is the predominant distortion product because 2nd is canceled due to the PP action of the output, which also as a result makes total distortion less. Kinda apples to oranges, I *think*..

Tim

[Tube_Dude]

In the source folower configuration the input capacitance is much reduced in relation of the comun source.

Because in the source folower the input capacitance is efectively bootstraped and we will end with a much less capacitance!

Regards

Jorge Santos

[Zelko]

I tested IRFP240 as a source follower and it works well untill I have made it work into capacitive load i.e.8ohms+1uF. It exibited a lot of ringing (overshoot).
Just for fun I tryed output trasnsformer coupling with simple 1:1 bifilar prim/sec over ordinary iron for mains transformers.
Surprised I was! Almost no ringing, about the same as with best SE300B with high price output trafo!
7Hz...55 kHz, -3dB bandwidth with simple trafo! Distorsion is 0.1% up to 10W, second harmonics with traces of third.
I am building second channel to do some listening tests.
Will be back as soon I get some results...

[Circlotron]

Originally posted by Zelko
7Hz...55 kHz, -3dB bandwidth with simple trafo! Distorsion is 0.1% up to 10W, second harmonics with traces of third.

This is interesting! Did you have the primary of the transformer as the only thing in the source lead and therefore have DC flowing in the primary? If you did, was there an airgap in the iron core? What power level did you measure the frequency at? Was it an EI or toroid transformer. Have I asked one too many questions?

[Zelko]

Jes, it is EI core 96x82x33mm outer dimensions, approx 9 cm sq,
bifilar wound, with insulated copper wire 1mm dia, connected between source and ground, secondary to speakers, no feed-back.

[Zelko]

Ooops, sorry, forgoth to mention the air-gap is 0.5mm,