Groundplane vs. star ground

[Devil_H@ck]

I am in the process of routing (well, trying to...) a PCB for a preamp I'm making. Although it will be double sided, I think I'm going to have a lot of trouble if I need to create a star ground, since there are 12 opamps on it, 4 buffers, etc.

Now, if I made a ground plane, routing would be a lot easier, but I'm wondering if it will do anything to the sound quality. I could make the groundplane double sided if that's better.

So, what influence would a groundplane have compared to a star ground & would a double sided ground plane be any good?

[MWP]

Star ground is always best, but if you cant do it, you cant do it.

Ground plane will be fine if its done properly.
Use both sides, and place vias between the ground planes quite often.

[Yoghourt]

You can also try tree-like grounding.
It's easier to route than star-grounding, at least. If tree-ground matches signal pathes, you may achieve even better results than star-ground.
(IMHO)

[Devil_H@ck]

Originally posted by Yoghourt
You can also try tree-like grounding.
It's easier to route than star-grounding, at least. If tree-ground matches signal pathes, you may achieve even better results than star-ground.
(IMHO)

Well, that's what I had in mind when I was talking about the star routing, since the board has 5 separate LM317/337 regulators, one for every "stage" in the preamp. If I were to do star grounding, I would route the ground trace to all those regulators and then start again from there.

I'll see if I can do it, otherwise I'll make a double ground plane.

Would the difference be really audible?

[Upupa Epops]

Star ground is better ( I agree with MWP ), if you know, how to make it. Little off topic notice : I was looking on your soft start module : you have there too near distance of holes from mains, so use there only plastic distance columns ( not with thrue hole ) !

[daatkins]

I recently made a seven channel amp using five TDA2050 and two LM3886 on a double sided PCB with the top layer as a ground plane. A couple of pictures of the PCB can be found in this thread:
Problem controlling a PGA2310

I ended up scrapping the original layout because of hum. The current spikes that charge the local reservoir capacitors were causing a 100Hz ripple voltage in the ground plane. It was only a few milivolts in amplitude but that was enough, when amplified, to cause an annoying hum to come through the speakers.

There was no way I could route PCB tracking into a star ground so I took the easy way out! The second revision of my PCB used thick insulated cable soldered under the board to link each reservoir cap to the star ground on the main power supply. That removed the hum completely. I retained the top layer as a ground plane but it was only used as a signal ground.

Edit: Just re-read your message and realised you’re making a pre-amp. That changes what I’ve written above! I made a combined pre-amp and three way active crossover that uses a single layer ground plane and it sounds wonderful – no hum at all. For a pre-amp, a ground plane should be fine as long as there are no large currents in it.

Nice one,
David.

[Devil_H@ck]

Originally posted by daatkins
Edit: Just re-read your message and realised you’re making a pre-amp. That changes what I’ve written above! I made a combined pre-amp and three way active crossover that uses a single layer ground plane and it sounds wonderful – no hum at all. For a pre-amp, a ground plane should be fine as long as there are no large currents in it.

Nice one,
David.

Well, what are "large" currents? How much are we talking about? 500mA, 1A, 2A? There are 11 opamps, 4 BUF634 (terminated with 50 ohms, so they are always on, even if no cable is connected to them), 8 SSM2404 (of which at any time, only 4 of the 32 switches are engaged) and 2 more chips. I don't think the total current consumption will be higher than 500mA.

[daatkins]

The maximum current depends on the size of your ground plane and the thickness of the copper as this will determine it’s resistance and therefore what voltage will develop across it when the current flows. I assume you’re powering the op-amps and analogue switches from +/-15v so there won’t be any power supply currents in the ground plane (perhaps a little from the 2204, but not enough to worry about).

The only possible problem I can see is the 50R terminations on the BUF634s. What signal level will these see? For example, if you’re driving +/-10v from the BUF, that will give 200mA. If the ground plane has a resistance of 1mOhm from the termination back to the power supply then the ground voltage at the termination will rise/fall by 200uV. A nearby op-amp (in, say, inverting mode) will see it’s non-inverting input (which is connected to ground) also rise/fall by 200uV so this signal will be amplified by the closed loop gain and appear on the op-amp’s output.

The lesson I learned was to avoid the short, high current spikes that occur when a power supply cap charges. I don’t think you have anything to worry about.

Nice one,
David.