Planes, layers, ground, power, and two-sided PCBs

[amptramp]

Sometimes, as with many tube amplifiers, there is a need for a power converter on the board.  It would pay to get the Unitrode designer's book because different topologies have different noise spectra.  The flyback converter has a steadily rising current drain until it shuts off, giving fundamental and all even and odd harmonics.  The current-fed converter is a strange one that is used where nuclear weapons are being used but it has relatively low noise as it has a triangular current drain and can stand all semiconductor devices turning on at the same time due to nuclear photocurrents.  The resonant converter is designed for situations where noise is the prime concern because it uses a resonant circuit to limit much of the noise to one frequency.  Push-pull converters have a number of harmonics that vary with the current output.  The 7660 capacitive switching devices have plenty of noise and I am surprised we don't get more complaints about them.

Pushing the fundamental frequency of the convertor well beyond acoustic response is one way to help cut noise.  Even if it is not audible, it can sometimes limit the dynamic range of a circuit by lowering the supply voltage during parts of its waveform.  If you use a power converter, it pays to investigate these things.

[GGBB]

My approach for basic through hole pedals is to try and route everything on the bottom copper layer. Use the top layer for "links" i.e. short straight traces for when you need to jump over tracks. This leaves 99% of the top layer of copper available for ground. Once you've done the copper pour, take a good look and make sure there aren't any breaks, gaps, cuts or otherwise circuitous or high indutance paths that need to be addressed. You can then fill the remaining bottom layer with ground pour and couple it to the top layer with vias. This won't be a ground plane on the bottom layer, it's too fragmented for that, but it can provide a nice electrostatic shield. You might not want this near to guitar input signals, MOSFET gates or otherwise high impedance nodes where the distributed capacitance will kill the treble.

Thanks.

When avoiding the high impedance nodes, do you use a cutout (hole) in the plane that follows the input traces? I would love to see an example of that. And do you do the same for both planes/layers or just one (which one)?

What counts as a high impedance node? I would guess any FET input device, but I'm only guessing and want to be sure. Not looking for an exhaustive list - just the sort of things we find in DIY dirt pedal clones for example. Does it matter if the input is for the guitar/outside world or if it's internal such as the second stage of a dual op-amp overdrive? What about non-FET devices that have a high input impedance set by bias or resistors?

[jonny.reckless]

My approach for basic through hole pedals is to try and route everything on the bottom copper layer. Use the top layer for "links" i.e. short straight traces for when you need to jump over tracks. This leaves 99% of the top layer of copper available for ground. Once you've done the copper pour, take a good look and make sure there aren't any breaks, gaps, cuts or otherwise circuitous or high indutance paths that need to be addressed. You can then fill the remaining bottom layer with ground pour and couple it to the top layer with vias. This won't be a ground plane on the bottom layer, it's too fragmented for that, but it can provide a nice electrostatic shield. You might not want this near to guitar input signals, MOSFET gates or otherwise high impedance nodes where the distributed capacitance will kill the treble.

Thanks.

When avoiding the high impedance nodes, do you use a cutout (hole) in the plane that follows the input traces? I would love to see an example of that. And do you do the same for both planes/layers or just one (which one)?

What counts as a high impedance node? I would guess any FET input device, but I'm only guessing and want to be sure. Not looking for an exhaustive list - just the sort of things we find in DIY dirt pedal clones for example. Does it matter if the input is for the guitar/outside world or if it's internal such as the second stage of a dual op-amp overdrive? What about non-FET devices that have a high input impedance set by bias or resistors?

You can remove the copper pour around high impedance nodes, on both layers. Of course what defines high impedance depends on the signal frequency and the impedance of the rest of the circuit. A 1M guitar input is high impedance, but you only really care about frequencies up to around 8kHz, and a typical pickup might look like a 10k resistor in series with a 2H inductor at HF. The gate of a MOSFET in a switch mode power supply operating at 1MHz is a very different proposition. Whilst it's high impedance at DC, at the switching frequency the internal gate capacitance swamps anything the local ground plane is doing.

Another approach is to shield high impedance nodes with a "guard ring" or "guard plane" - something which is at the same voltage but a much lower impedance. This is typically done in very high input impedance instrumentation amplifiers. Build (for example) a JFET input unity gain follower, and use that to drive the guard ring / plane around the high impedance input node. This effectively bootstraps the capacitance so it doesn't leak electrostatically, and extends the input impedance bandwidth. I see this trick a lot in commercial instrumentation, measurement and lab equipment. High bandwidth FET scope probes need to worry about a picofarad of stray capacitance - you probably don't.

It all depends on what you are trying to achieve. For most guitar frequency circuits I wouldn't worry too much about it. I just try to keep the trace from the input jack to the first stage fairly short, and often start with a simple JFET source follower to avoid Miller effect. You're probably fine with an op amp first gain stage too. Don't worry too much about it for guitar signals.

One piece of advice though - don't use BS170 or 2N7000 MOSFETS in a high gain stage connected to the input of a guitar effect (although I've seen this done in commercial designs. maybe for deliberate tone shaping effect). These MOSFETs are quite high input capacitance devices, and quite high gain. Together the Miller capacitance will totally kill the treble of a guitar pickup. This is why I used a JFET source follower on the input of the Little Jim MOSFET distortion circuit. Without it, the first gain stage started to roll off around 1kHz or even less. If you fully bypass the source of a 2N7000 common source gain stage with a typical stratocaster pickup you'll be lucky to get a 3dB bandwidth of 800Hz.

For most guitar circuits, you care much more about 120Hz power supply injection, noise from a switched capacitor inverter / booster, or inter-channel crosstalk through power supply or ground coupling, than you do about treble loss due to ground planes. I've built high gain tube amps with and without ground copper pours. There might be a bit less top end on the ones with a lot of solid copper flooding, but they're generally quieter, better behaved and less likely to burst into unwanted oscillation at high gains

[jonny.reckless]

Sometimes, as with many tube amplifiers, there is a need for a power converter on the board.  It would pay to get the Unitrode designer's book because different topologies have different noise spectra.  The flyback converter has a steadily rising current drain until it shuts off, giving fundamental and all even and odd harmonics.  The current-fed converter is a strange one that is used where nuclear weapons are being used but it has relatively low noise as it has a triangular current drain and can stand all semiconductor devices turning on at the same time due to nuclear photocurrents.  The resonant converter is designed for situations where noise is the prime concern because it uses a resonant circuit to limit much of the noise to one frequency.  Push-pull converters have a number of harmonics that vary with the current output.  The 7660 capacitive switching devices have plenty of noise and I am surprised we don't get more complaints about them.

Pushing the fundamental frequency of the convertor well beyond acoustic response is one way to help cut noise.  Even if it is not audible, it can sometimes limit the dynamic range of a circuit by lowering the supply voltage during parts of its waveform.  If you use a power converter, it pays to investigate these things.

In high gain circuits with SMPS and poor PSRR (i.e. tube amps) you often seen the spectra of the switching demodulated during clipping into the audio band which can make the amps sound gritty and harsh. Better to control the current loops of the high frequency switching circuits and only present clean power and ground signals to the rest of the circuit, especially the audio frequency parts where the PSRR is very low. Even with 7660 type converters I try to keep the traces short and fat, and use 10uF MLCC caps on the input and output, with the ground returns controlled locally, to avoid charge switching injection into the rest of the circuit. Basically make sure the ground traces seeing those 7kHz or 45kHz current spikes don't share any copper with your audio circuit.