PCB design - Ground planes and digital/analog ground

[wcampagner]

Hello,

I'm trying to make my own pcb's in Eagle...
They are double-sized...

I've breadboard a Tubescreamer and made all the mods i wanted, then i started to make the PCB...
It's my second prototype now... and i've got the same problem: RFI interference from an AM Radio Station near my house...

The first PCB was just an autorouter from the schematic...
But the second i was carefull with some aspects:

- put small value ceramic caps near all the digital CI's,
- use start ground on the filter caps. (GND, VCC and VCC/2),
- separate digital and analog signals (GND, VCC),
- ground to the chassis only in one point... etc... etc...

But the problem is still there... i have an Sansamp GT-2 that doesn't suffer this RFI interference when i use it in my house... so i want my stompbox not to suffer from this too...

Now what i want to do is use a ground plane on the bottom of the double-size PCB to make another try...
Can someone help me how to do it?? All the bottom area will be an entire ground plane?? Or just some areas??
How can i separate the digital and analog GND on the ground plane?? I'm using electronic bypass with Flip-Flop and FET's...
I'm thinking about two ground planes on the bottom of the board... one for the analog ground and another for the digital ground... connecting them with just one point... is it correct??

Thanks a lot in advance,
Wagner.

[hank reynolds 3rd]

this sounds obvious, but i guess you could use the whole of the other side, except for the areas where the copper will touch the parts (unless the corresponding side of the res/cap etc, is at ground)

Is it in a box yet?? that might cause it to pick up RF

[d95err]

The RF interference may not have anything to do with the PCB design. It could be the shielding (or lack of it).

The generic solution to RF interference is to put a small (e.g. 47p) cap from input to ground. Even better is a ferrite bead in series with the input, followed by a ferrite bead. You could also try to run a shielded wire from input to the PCB (shield connected to ground only at one end).

[wcampagner]

Thanks hank reynolds 3rd and d95err,

The stompbox is in the box... and the RF interference is still there... as i told before... there is only one connection from the PCB ground to the aluminum box... one thing i notice is that when i touch the string the RF interference gets a little louder... so that's why i'm suspecting of some ground problem... it seems that the ground signal is in some way getting on the stompbox circuit... but that's is what i'm suspecting... i don't know if this is the problem...

I didn't tried the ferrite beads yet... that's because i'm usign my original GT-2 as an example... and the GT-2 there is no ferrite bead inside and there is no RF interference on it... so i don't know what it is...

I tried the small cap from VCC to Ground too... but it didn't solve the problem too... i tested it with normal Power Supply and with battery too... and in both tests i get the interference... so i think the interference is not comming from the Power Supply.

Do you have another suggestions for me?

Thanks again,
Wagner.

[wcampagner]

oh, just another consideration:

All my components are onboard... so i don't use offboard wiring... that's why i'm not using shielded wires in the input... the input and output jacks are those plastic normaled jack... none of them get contact with the metal chassis... the grounding of the jacks are made on the PCB ground trace.

Thanks,
Wagner.

[d95err]

one thing i notice is that when i touch the string the RF interference gets a little louder... so that's why i'm suspecting of some ground problem...

I don't think it's a ground issue. When you touch the strings, the signal will pick up RF interference from your body as well (it works as an extended antenna). So, the RF gets into your signal through the guitar pickups. Shielding won't help, since the RF is already in the signal when it enters the box.

Try the 47p cap from input to ground, I think it will cure the issue. A small resistor instead of a ferrite bead can help too.

[wcampagner]

Thanks d95err,

I'll try it at home this night and i'll post the results tomorrow.

Thanks again for your help.
Wagner.

[Transmogrifox]

I would second any recommendation to deal with the external before making a new PCB.  After enough of those the price of your pedal goes sky-high.  You may even benefit from sticking a small-valued capacitor in parallel with your pickups in the guitar itself.  Try some things like turning the volume pot all the way off.  Another thing to try is volume at max, tone at minimum (most dull setting).  See how messing with such knobs and switching pickups effects the sound.  Even try hooking a CD player into the pedal input to see if there is any RF coming from the CD player output.  These will be easier things to try first to give yourself a better idea of the real problem before you do expensive and time consuming things like making a new PCB.

Also, separating the "digital" an analog circuit paths in this case is not so critical.  If you end up adding a ground plane, don't worry about using the same ground plane for the switching and analog because it's very low impedance and the probability of introducing switch pop through a ground plane is very small.  Just give it a separate power supply trace and a little additional capacitance if you're worried about switch pop.  It can be a little more critical when your "digital" operates in the hundreds of MHz range where effects that are negligible at audio to lower RF become serious issues that can't be swept under the "negligibility rug".

I still don't think it's a PCB problem.  If the RF goes away when you short the input to ground, then you know it's getting in on the signal path--and the shunt 47pF capacitor would help immensely.

[Processaurus]

x... one thing i notice is that when i touch the string the RF interference gets a little louder...

Usually RF picked up through pickups gets quieter when you touch your strings, because they are grounded and doing that grounds your body (which is acting as an antenae).  You might want to check your strings are actually connected to the shield on your guitar cable (they should be).

Another thing is to check with a multimeter or beeper that all the external metal on your box, including pots, switches, jacks, and both the top and bottom of the enclosure are all grounded together, rather than a metal part not being grounded and acting as an antennae.  paint can act as an insulator, so can clear coats on bare metal.  If you don't have a beeper you could just attach a wire to ground, and then push it against different metal components on the outside of your enclosure and listen to see if the noise goes away or gets better.

I still don't think it's a PCB problem.  If the RF goes away when you short the input to ground, then you know it's getting in on the signal path--and the shunt 47pF capacitor would help immensely.

also a good suggestion.

[wcampagner]

Hello,

Thanks all for the replyes...
Yesterday i tried to test it again... but i don't know what i happened, but there was no RF interference...
Maybe it was because it had just rained... i don't know...
I'll try to test it again today when i get home...


Another problem that i have is with the switching system... it works great... very silent... but when the Stompbox Volume Control is in the max... i hear switching POP's... very loud and it seems it is a high frequency POP... but the frequency changes when i change the Tone control too...

The stompbox that i am making is the tube screamer original version from:
http://www.generalguitargadgets.com/diagrams/ts-808.jpg

The only difference is that i'm using Flip-Flop switching instead of Transistor switching...

But when i do the modifications proposed in this schematic:
http://forum.musikding.de/attachements/Boss_SD-1_mod_c.gif
The problem is solved... ading another JFET and a capacitor in the first Amp-Op loop solve the problem...

Do you think there is any problem using it?? I mean, will it change in any way the behavior of the stompbox??
Why is it happening?? I think the orinal tube screamer didn't have this additinal FET and it worked allright...

Thanks in advance again,
Wagner.

[R.G.]

It' s likely that the PCB is not your problem at all, as the others have noted.

Ground plane construction is essentially useless at DC and low audio, and gets more useful as frequency increases until somewhere in the UF band it's impractical to get things to work without it. It's utterly crucial with high speed digital logic signals where the fundamental frequencies of the leading edges of pulses may be in the GHz range. But at audio and low AM band, it's noncritical to nonfunctional except as shielding.

What kills you on RF is the antennas. That 10-20 foot guitar cord is a good one. And living near an AM radio station can produce field strengths that will nearly boil water in a cup. The RF comes in on any kind of wire antenna, and it gets rectified by any nonlinearity - usually the first semiconductor junction it hits.

To kill RF
(1) keep it out of the box; that means a conductive enclosure all around the circuit, connected to signal ground. This forms a Faraday cage that forces the RF to the outside of the box. The more conductive the box, the better the Faraday cage and the less RF that gets in.
(2) choke it to death when it gets in; simply putting an RF bypass capacitor at the input of your circuit is not enough. That solution is usually resonant at some frequency, and will actually act as a tuner for that frequency. Instead, insert some series impedance in the input lead, such as a resistor or ... better... a ferrite bead, which gives you some inductance and a second lowpass pole. Then put a high quality ceramic capacitor to ground after that.
(3) ignore it; you do this by making your circuit have little or no response above, say, 100kHz. For transistors, feedback capacitance such as collector-base or drain-gate will do it. You only need a little - 10pF feedback is usually enough. Feedback caps on opamps are OK, but you ahve to check closer. They can have unintended side effects.
(4) stay well grounded; in extreme situations, locally bypass the power supply with high quality ceramic caps to shunt power supply RF signals.

Digital logic in footswitch setups hardly qualifies as logic. You really don't need to preserve fast edges in that.

And save the ground planes for your multi-MHz projects.

[Dave_B]

I'm glad I read this thread.  Thanks, R.G.  I've got a Weak Joe compressor on my build list but the double-sided board (mostly serving as a ground plane) was really keeping it back-burnered.