PCB Ground Plane Question

[Paul Marossy]

Is it possible for a ground plane on the back side of a PCB to cause oscillation problems in something like a ring modulator circuit?

[R.G.]

In some cases yes. It would be really an unusual set of circumstances for low frequency audio or an LFO.

[Paul Marossy]

OK so how about if you have a situation where you have +/- voltages created by a half wave rectifier with the 16V filter caps seeing about 3V over their voltage rating and everything sharing the same ground? Can you get a ripple/oscillation on the ground that would cause certain types of IC chips to malfunction? And would a ground plane make that worse?

[R.G.]

OK so how about if you have a situation where you have +/- voltages created by a half wave rectifier with the 16V filter caps seeing about 3V over their voltage rating and everything sharing the same ground?

First off, don't do that unless you're already stuck with it and can't change, and for reasons unrelated to ground planes. Rectifier/filters need special wiring on the grounds, and ground planes are not the way to do it. Rectifier/filters have relatively huge pulses of current on the nominally ground side, and this stuff simply must be on its own wires from transformer to rectifier to filter caps to avoid hum issues. In addition, these pulses can make the wiring to them ring at RF, and a ground plane setup may in fact carry the RF to sensitive places.  Half-wave makes this worse because the pulses are less frequent, and therefore higher amplitude. And filter caps should never be operated at over their voltage rating unless you really have no options. If at all possible, trade those out for the next higher voltage rating.

If you're stuck with it, there may be ways to cope.

Can you get a ripple/oscillation on the ground that would cause certain types of IC chips to malfunction? And would a ground plane make that worse?

All that being said, causing certain types of ICs to malfunction would be unusual just from the ground plane; but possible, as a ground plane is, among other things, a tuned high frequency transmission line to everywhere in the circuit. The tuning and other issues get critical, as does the specific IC. But I think that other issues are your problem instead of there just being a ground plane. Details always matter, though.

[Paul Marossy]

Interesting. I'm actually troubleshooting a PCB someone else designed and it just ain't working right. After double checking the schematic and stuff, and correcting the mistakes on it I just can't get it to work without oscillation problems. So my current theory is that the power supply grounding is somehow part of the problem. Will find out if I'm right soon enough...

[R.G.]

Interesting. I'm actually troubleshooting a PCB someone else designed and it just ain't working right. After double checking the schematic and stuff, and correcting the mistakes on it I just can't get it to work without oscillation problems. So my current theory is that the power supply grounding is somehow part of the problem. Will find out if I'm right soon enough...

What does the power supply decoupling look like? At least 10uF + 0.1uFceramic on each power pin of the IC(s)? Vanishingly short traces from IC power pins to the decoupling caps? This is usually far more important at audio frequencies than ground planing.

[alanlan]

I would replace the 16V caps with 25V rated caps and then re-evaluate.

[Paul Marossy]

What does the power supply decoupling look like? At least 10uF + 0.1uFceramic on each power pin of the IC(s)? Vanishingly short traces from IC power pins to the decoupling caps? This is usually far more important at audio frequencies than ground planing.

The IC chips all have a minimum of 10uF electro cap, but not any 0.1 ceramics. All of the IC chips are powered thru voltage regulators, for the negative supply it's 79L05s & 79L12s and for the positive supply it's 78L05s and 78L12s. All the power supply traces are pretty much as direct as possible except for one +12V track that makes a 1/2 trip around the board to the output IC (but that's not the IC I am having trouble with). On the negative supply there is a 1000uF and a 470uF filter cap right after the 1N4007 and on the positive side there is a 2200uF, a 1000uF and a 470uF after the 1N4007. The whole bottom side of the PCB is one big ground plane.

I would replace the 16V caps with 25V rated caps and then re-evaluate.

That's exactly what I'm doing. I'm using 35V caps and waiting for a couple of parts to complete a second board populated by myself just to make sure there isn't one wrong part on the board somewhere on the pre-populated one that was sent to me. But all the filter caps need to be at least 25V regardless.

[R.G.]

The IC chips all have a minimum of 10uF electro cap, but not any 0.1 ceramics. All of the IC chips are powered thru voltage regulators, for the negative supply it's 79L05s & 79L12s and for the positive supply it's 78L05s and 78L12s. All the power supply traces are pretty much as direct as possible except for one +12V track that makes a 1/2 trip around the board to the output IC (but that's not the IC I am having trouble with). On the negative supply there is a 1000uF and a 470uF filter cap right after the 1N4007 and on the positive side there is a 2200uF, a 1000uF and a 470uF after the 1N4007. The whole bottom side of the PCB is one big ground plane.

Before you go any further, get a batch of 0.1uF/50V dipped epoxy caps. If you've been at building long enough, you have probably already bought a bag of a few hundred of these for under $10 and use them like candy party favors. I do. For instance, look at these:
http://www.mouser.com/ProductDetail/Vishay-BC-Components/K104K15X7RF5TH5/?qs=sGAEpiMZZMuAYrNc52CMZNaBwupcnfl2GHhuVIbtnIM%3d
A baggie of 100 of them will cost you $4.40 plus shipping in your next Mouser order.

Now, find every V+ and V- power pin on every IC on the bottom of the PCB, and solder a 0.1uF cap from the pin to the ground plane, scraping a hole in the ground place to solder the cap lead to. This should be as close to the power pin as possible, and the cap leads should be as short as you can make them and still get them soldered. If you're conservative with your time and effort, start with the ICs that are giving you problems first. Include as a first effort the input and output pins of the voltage regulators.

This may not be THE problem, but it's certain that it will not make things worse, and it may be all of the the problem.

[Paul Marossy]

The addition of the .01 caps to ground on the +/- pins of the ICs seems to have helped to make things more stable. At least now I can get it to where it doesn't just spontaneously oscillate when it's not supposed to. Before I couldn't get the carrier noise to an acceptable level it just wanted to oscillate like a maniac when there was no signal. It would start out OK, or so I thought, but it would always start to oscillate no matter how I tweaked the trimpots (which I did for a few hours). The decoupling of the IC power supplies seems to have been part of the problem. Hopefully I can figure out the rest of this on Monday or Tuesday when my regulators arrive...

[Paul Marossy]

Well, after many hours of messing around and coming to dead ends, I eventually found that out that the schematic they provided was missing two very strategically located resistors whose omission screwed up the whole functionality of the circuit. Argh.