PCB layout tips for high gain FET circuits?

[John Lyons]

In laying out a PCB for a multi-stage FET high gain circuit at 9v what are my concerns as far as proximity to avoid noise/oscillation etc?

In tube amps you don't want the coupling caps (caps between stages) too close to the next stages plate (or drain in FETS) because of high voltage energy fields getting into the Grids (or Gates in FETs) . The stages should be spaced out a bit and wires should not run parallel or even cross if you can help it.

Do low voltage circuits suffer from this as much? Should I take care to keep the coupling caps and Gates away from the Drains?

Are there other pitfalls to look out for with high gain circuits?

Some things I do are:
Use shielded input and output wires to the switch and board and space these wires as far as possible apart.
Add 100uf filtering and parallel .1uf  from  +v to ground to help with the electrolytic caps slow response to high Frequencies.

John

[MartyMart]

I think that what you're doing now is fine John, so long as cables are as short as poss
and dont cross over each other it should be fine.
I can't see that the same thing as tube amps applies here with coupling caps and the plate !
I try not to "cram" my builds too close together and rarely get oscillation, I've never used
shielded cables !!
My "Mr Mostorto" is possibly the highest gain circuit that I have and has ZERO noise issues

MM.

[markm]

Same here Marty.
I don't really do the high-gain thing but, I haven't had any major noise issues at all.

[John Lyons]

Yeah, it seems that some high gain layouts are pretty close together and they work well as far as I know (have not built them...yet)
Example: Meteor with doug H's PCB layout, Thunder chief, Plexizer, MrMostorto, BSIAB.
But then you have the Drboogie which I had to wrestle with to get it under control and squeal free. Many "band aids" in place with that one.
I had to lower the gain in 3 places and use a 500K gain pot. Plus there is always a good amount of  pink noise/snow even with the gain turned all the way down.

Any ideas on what the main oscillation issue is with that one? Mine is fine now but it would be nice to know which areas are problematic (other than the 220pf miller responce caps added across the gate and source of Q1,2 and 4) I wonder if it's a matter of the tone controls too close to the gain pot wiring? I really want to bread board that one and disect those gain blocks and play with alternative values with the cold biased clipping section of Q3.

Anyone else have any thoughts?

John

[R.G.]

Oscillation requires two things:(1) feedback exceeding unity gain through the loop from input to output and back through the feedback path to the input and (2) a phase shift adding up to 360 degrees at some frequency.

For systems with overall positive, noninverting response from input to output, all frequencies are in phase, so all they need is enough signal getting back to sustain unity loop gain or more.

For systems with overall negative, inverting response from input to output, the oscillation will be a distorted sine wave at the freuquencies where the phase shifts by another 180 degrees.

What people don't usually understand is that in a pedal with a huge gain, the amount of feedback can be tiny and there can still be enough to sustain oscillation. Capacitive feedback is a good one - the input to a good JFET gate is at least 100M ohm, often thousands of megohms. Put an inductive gate pulldown resistor on the gate and this baby can oscillate at the frequency of the resistor - and wire leads' - inductance with a feedback capacitance of two wires crossing at right angles half an inch apart.

With enough gain, the resistance of the copper wires carrying the sewer-ground current makes a feedback signal that will make it sing.

With enough gain, the wires in the power supply will cause oscillations that get worse when you add bypassing caps. The bypassing caps only lower the frequency of the oscillation because there's not enough damping on the resonant (!?) power supply leads to eat the oscillation.

For how to do a layout to avoid oscillation, get an old copy of the ARRL handbook and read how to lay out RF circuits. Same things apply.

1. physically separate input and output; then don't throw it all away with six inches of wire from the input to the bypass switch twisted up with the output leads.
2. arrange your circuit so the signal flows in one side of the board and out the other as far from the input as you can get it.
3. bypass your power supply, but introduce some resistance into the power leads and put on not only bulk caps of 10-22uF but also 0.1uF ceramic and some 0.01uF ceramic with short leads
4. short leads; a straight wire has stray inductance and capacitance all by itself; keep the wires as short and straight as you can get them
5. visualize where the ground currents flow and what they look like; separate reference ground (also known as high quality or clean ground) from the power supply sewers which drain the used electricity back to the power supply; keep the sewer currents from contaminating your clean reference ground.
6. be very, very careful with high impedance points; know what the input impedance of each active device is on your circuit board and lay out accordingly; a high impedance can be fed into by a vanishingly low capacitance, especially if you have a lot of gain; there are reasons why RF circuits are all done at impedances which are in the 50-377 ohm range.

If you put in enough gain *anything* will oscillate.

[MartyMart]

John, I read something recently that said when comparing the setups of valve circuits
with Jfets, should take into account the bias and how this equates from one to the other.
I wish I could remember the thread, but the bottom line was, biasing for 4.5v as we seem
to do, was NOT recommended, to be accurate to the "valve" version and that it should be
more like a  "6 volt" bias at the drain !!
B+ on a valve amp perhaps is say 300 volts, you always have more than "half" that at the plate
so dividing 9v down to 4.5 seemed wrong to the poster ......
This should keep the high gainers a bit more " tame "  !!
Also Gus keeps reminding us to use calculations for bias from the "source" rather than just
a huge trimpot at the drain, though i thought the two were pretty much "symetrical and exchangable"
myself .

Good stuff RG, in response, here's how all my "stripboard" layouts get layed up :

This is more "follow the schem" than "lets twist this into oblivion" !

1 - I layout from left to right, input to output and try to keep traces separate
where possible.

2 - Power and ground strips are normally top and bottom of the vero, so as
far apart as possible.

3 -  I dont always take "off board" connections to the edge "holes" why not
use a shorter distance to the "node" .... keep wires from gain/tone/volume
from crossing and as short as possible.

4 - I use 100uf and 100n bypass caps with a reverse protection diode on every build
and often a 100ohm resistor across the 9V line.

5 - I try to keep components straight without all the "lead" spread over the board
Dont twist leads or cross components over others, in particular the gain stages !

MM

[John Lyons]

Thanks much for the insight RG and Marty!
I'll work on this a while and see what I can come up with.
Lot of good info there.

So board component proximity isn't of much concern, assuming they are layed out linearly?

John

[Bucksears]

I felt the very same way a while ago in that I felt like my 'gainier' circuit boards have layouts that are too close together. My  BSIAB II (same one as GGG, but I pulled everything in a little closer together in Illustrator), Big Muff, May Queen and Guvnor are all fine, noisewise.
My Dr. Boogey is a bit noisy, so I designed another one with the parts spread out a little more. The size difference is negligible. I haven't gotten around to doing a board for that one yet, but it's in the works.

- Buck

[John Lyons]

I have a Dr boogie layout pretty much finished with the added pf caps and some slight gain changes (lowered) but if a tight layout won't make any difference theoretically then I can tighten mine up a good amount for board size reasons. Mine is spaced a good bit to give the fets some breathing room... at least for the gates and coupling caps.

John