Strange artifacts in MOSFET preamp. What are they?

[marcelomd]

Hi,
I've been playing with simulating a MOSFET preamp when I found these artifacts in the node after the tone stack, on either side of the 100n cap. Strangely, they are not present in the output of the MOSFET. Lowering the resistors make them go away, so I guess it has to do with some impedance mismatch.

What causes it?

Thanks!

[Rob Strand]

It's a non-linear effect so it's going to be due to the MOSFET.   If you are doing a simulation it's probably the MOSFET model.   What drain voltage is your ckt biasing at?  Maybe a low voltage?



Do a search on this forum,  I posted some MOSFET models for rankot some years ago. They are tuned for the low currents you get in pedals.

This thread has some models,
https://www.diystompboxes.com/smfforum/index.php?topic=52905.msg1142272#msg1142272

Lowering the resistors make them go away, so I guess it has to do with some impedance mismatch.

This is a sign the model isn't good for low currents.

[merlinb]

My guess is the first few cycles of audio are driving the MOSFET to saturation, causing it's feedback loop to collapse on the peaks. Eventually the caps charge up to a new value that biases the MOSFET to a more stable condition. I'm just spitballing here.

[marcelomd]

Look at that. Rob's model gives a very different response. Amazing.

No more artifacts:


Thanks a lot, guys!

[marcelomd]

Ah, one more question.

This is the original circuit. 2 MOSFET stages:


Using the original model, I get slight clipping in the first stage. 100mV. The updated model is clean for the same signal level.

Thing is, I've built this and it does break up easily. Is my hand that heavy that I consistently produce 100mV signals or is there more to it?

[Rob Strand]

Using the original model, I get slight clipping in the first stage. 100mV. The updated model is clean for the same signal level.

Thing is, I've built this and it does break up easily. Is my hand that heavy that I consistently produce 100mV signals or is there more to it?

You may need to do some tweaks.

The original model you used with the artifacts has a very high KP, that's what gives it more gain and is part of the reason the DC bias get screwed up (IIRC).

Real devices (at low current) correspond to parameters close to my model.   There's a few variants of those MOSFETs and I'm not exactly which one the model I posted is for.   You might need to increase KP to match your devices, say double KP to KP=52e-3 - off hand no more than that.  If you really want to get it right you can put a sinewave into the real circuit then tune KP in spice to get exactly the same AC gain.  You need to make sure the signal generator impedance is put into spice.   Just make sure the drain voltage on the real circuit biases to the same voltage as the simulation. 

[marcelomd]

Interesting.

KP=52e-3 makes it clip more or less like the original model.

What I'm actually doing is trying to tune a preamp to have light distortion/start to break up at half gain. I also don't have any tools or components here. I'm living in the US for a few months, which complicates any physical tinkering for now.

So, I don't need things to be exact, close enough is good enough =)

[PRR]

This is the original circuit.

Do you really want one hundred uFd input caps, or did your keyboard mess-up?

[marcelomd]

This is the original circuit.

Do you really want one hundred uFd input caps, or did your keyboard mess-up?

Wow. Good catch.

Looks like an artifact of the image compression. Case in point: the treble capacitor also looks like 220u (it's supposed to be 220p).


This is the same circuit zoomed in:


It's been a few days and it's also entirely possible I've made a mistake and corrected it later... who knows...