Is there any other way to filter the power supply? Plus other questions

[marcelomd]

Hi,

I have a bunch of questions here regarding power supply filtering in pedals.

I'm trying to 1- make it more effective and 2- optimize my SMD layouts. I couldn't find anything with more detail in the forum. Also, internet has a bunch of info, but most of it is about digital circuits.

I'm also quite frustrated with noise issues, even after following as many EMC guidelines as I can. E.g: I have two 9V PSUs here. No name chinese PSU is quite silent, except it makes MOSFETs oscillate in some situations. I also have a 1Spot. Doesn't oscillate, but has super high noise floor.

---

So, this is the usual power supply filtering scheme I use. I believe something like this, with small variations, is used in 99% of all pedals.



- Low pass filter with a big bulk electrolytic capacitor, 47-100uF, possibly with an extra 100nF for high frequencies. 9V should pass through this cap before going to the circuit;
- One 100nF for every opamp, near the power pins;
- Voltage divider plus another big bulk capacitor, 10-100uF, for the bias voltage.

Questions:

1- How necessary and/or effective is all of it?
2- Is there any other means to reasonably filter the power?
3- How did we get to these values? 47-100uF for the bulk capacitors?
4- How low can we go?
5- If we disregard cost and inventory, can I use a bunch of parallel MLCC capacitors instead of a big electrolytic?
6- Can I distribute these smaller caps around the board? Usual guideline for digital is to have the bulk cap near the power connector;
7- When using an SMD electrolytic, how do I connect it to the ground plane? How many vias of which size?

Thanks!

[mozz]

Simple things, try bypass/jump the diode, some make noise.  I have seen 1 spot get noisy after they get old. 

[antonis]

https://www.analog.com/en/resources/app-notes/an-581.html

https://www.neodensmt.com/news/method-for-single-supply-amplifier-circuits-59316819.html

https://www.analog.com/en/resources/analog-dialogue/articles/avoiding-op-amp-instability-problems.html

[R.G.]

Power supply filtering is the eternal gift - it just keeps on giving. Here are some thoughts on your post.

No name chinese PSU is quite silent, except it makes MOSFETs oscillate in some situations. I also have a 1Spot. Doesn't oscillate, but has super high noise floor.

I designed switching power supplies for quite a few years. We (the power supplies design area) wound up resolving quite few system-wide problems when the systems people had odd, mysterious failures and wound up deciding that it was the fault of the power system. In a few of these instances, it was even the power supply. In many, though, it was a pervasive problem in the rest of the system.
In the example you quote, does the Chinese PSU make MOSFETs prone to oscillation in all pedals, or just a specific one? Does the 1Spot have high noise in all pedals or just a specific one?
As an aside, is the 1Spot power supply a Visual Sound/Truetone made 1Spot, or a similar clone, and how old is it? It could simply be defective. High noise floor is a very unusual thing in real 1Spot power supplies.
MOSFETs have very -very- high frequency responses. They are also prone to oscillation at frequencies that can be resonated with PCB stray capacitance and trace inductance. They  <want> to oscillate 8-) Any time you use a MOSFET in a linear amplification circuit, insert a 10R to 1K resistor in series with the gate lead as close to the MOSFET package as possible, and decouple the drain-source power supply locally with a ceramic cap as close to the drain-source power as you can; also, for MOSFET source followers, isolate capacitors from the source with a small resistor; followers, especially MOSFET followers, are prone to oscillating with capacitive loading.

So, this is the usual power supply filtering scheme I use. I believe something like this, with small variations, is used in 99% of all pedals.

For noise issues, a schematic with a single ground line is deceptive, and can lead to noisy layouts. This is the old star-ground-versus-random-ground issue. In placing decoupling and fighting oscillation, the placement of each ground path back to the bulk power decoupling capacitor on the PCB matters, and it matters even more with ultra-high-frequency capable MOSFETs. Ground lines (not wires) on schematics are not zero-resistance connections. Each section of ground trace is an R-L-C component capable of having an effect on (especially) MOSFETs. You can think of it this way: on the schematic, replace every section of "ground" line with a resistor in series with an inductor, and a capacitor across the R-L, and a capacitor to ground. The resistance, inductance, and capacitance may be small, but they are real, and can have funny effects. There are only two ways to have zero volts across a wire – (1) have the wire be a superconductor, or (2) have the current in the wire be really, truly zero.

The general way to get around the resonant RLC ground (and power!) traces issue is to put local decoupling caps near where the current is being used (like, one ceramic by each opamp, etc.) and distinguish between power current return traces, returning "used" power to the bulk power filter, and signal ground reference traces that establish voltages that are quiet for the signal lines to follow. The reference voltage generator, for instance, really ought to have its own ground line going back to the bulk power supply so it's not waved around by ground current caused voltages in other traces.

- Low pass filter with a big bulk electrolytic capacitor, 47-100uF, possibly with an extra 100nF for high frequencies. 9V should pass through this cap before going to the circuit;

9V doesn't go through this cap. The external 9V feeds this cap up to a certain amount of charge. The cap then feeds out charge (current) to the circuit. It's a mini-sized local bucket of charge.

- Voltage divider plus another big bulk capacitor, 10-100uF, for the bias voltage.

Where does the "ground" wire for this capacitor connect? If it connects to a nearby section of the ground wire for a MOSFET amplifier or an amplifier, this then feeds a voltage right back into your bias voltage through that capacitor. Maybe OK, maybe not, depends on what currents flow in the nearby ground RLC-wire, and what >phase< the currents in the ground wire happen to be.

1- How necessary and/or effective is all of it?

Sadly – that depends on the circuit it's feeding. I wish I could give you an always-good recipe. I can't. My approach has been to put the spaces for decoupling caps on the PCB and leave them off later if they're not needed.

2- Is there any other means to reasonably filter the power?

Yes. Unfortunately, thousands of them. This kind of filtering is what has built up over time as answers to common decoupling issues arose and were solved. Designers just started including the latest answers into their "standard model". It grew like a coral reef.  All parts may not be necessary on some circuits, and it may not be enough for others.

3- How did we get to these values? 47-100uF for the bulk capacitors?

By a path that's not as much electronics-math related as economics related. 47 to 100uF is the smallest and cheapest value that generally works fine for many circuits. It's a start-here-and-modify-as-needed value.

4- How low can we go?

Depends on the circuit. Could be as low as zero for circuits that are not very sensitive to power supply and ground noise. Could be totally inadequate for big circuits with lots of CMOS, MOSFET, JFET opamp, high gain...
I know that's a non-answer, but it's the truth.
The standard model is something that evolved over time to solve most power filtering for small circuits in pedals. It's a starting point suggested by history. I use my own variant of the standard model for new designs I do.

5- If we disregard cost and inventory, can I use a bunch of parallel MLCC capacitors instead of a big electrolytic?

Yes, in most cases. This approach will usually work better if the circuit noise producers are distributed around the board too. If there is one section that uses big lumps of current, a bigger cap near that part will work better. You want the capacitor (bucket of charge) near where the current is being used to keep the return current off the ground net. A bunch of MLCCs totaling the same as one electro will have as much low frequency filtering as an equivalent electro, and less ESR and ESL.

6- Can I distribute these smaller caps around the board? Usual guideline for digital is to have the bulk cap near the power connector;

Yes, you can. I think that the sources you're reading are taking into account that electrolytic bulk caps are only good for filtering low (e.g. audio) frequency noise, and their higher ESR and ESL make them just about useless for filtering the sharp, fast edges of digital signals. So if you have to have, say, 100uF of bulk capacitor, it's cheaper and easier to route if it's one cheap electrolytic where the power comes in. This has the desirable effect of keeping the power wires feeding into the board from sending in noise at low frequencies too, and using the distributed capacitance and inductance of the wires bringing in power as a power filter on the input.

7- When using an SMD electrolytic, how do I connect it to the ground plane? How many vias of which size?

Best practice is usually to have it connected to a local ground plane area bigger than the cap itself and run vias from that down to the ground plane. Second choice for me is to have the via be just off the end of the SMD cap to the ground plane, observing the PCB manufacturer's guidelines/rules for how close a via can be to solderable pads. Some PCB makers allow vias inside pads, but there may be special rules for this in manufacturing, parts placement and wave soldering.

[ElectricDruid]

1- How necessary and/or effective is all of it?

C9 I don't usually include.
D8 is vital unless you want idiots and drunk people to be able to blow up your circuit at will.
R10/C10 is generally a good idea, though you'll often get away without it.
C11 is best practice, but often left off. If C10 is effective at high frequencies, it wouldn't matter.
C12 is best practice, and also often left off. I'd do it for CMOS, but I rarely bother on op-amps. Mostly more out of habit than reason.
C17 with R17 and R18 is essential. It won't work without it.

2- Is there any other means to reasonably filter the power?

Yes. The history of pedals is mostly about doing things as cheaply as possible though, which is why we have the solutions we have.

3- How did we get to these values? 47-100uF for the bulk capacitors?

It's a simple RC filter with R10/C10, so the cutoff is 1/2PiRC. You need R10 to be small enough that the voltage drop at whatever current your circuit draws (V=IR) is tolerably small, but you need R10 to be large enough that you don't need to use a massive capacitor to get the filter frequency down low enough to remove noise.
Generally you'd want the filter cutoff lower than the mains hum, and a lot lower if you can do it since a basic filter like this only gives -6dB/Oct of hum reduction.
If you assume R10 is from 10R to 100R, that implies C is 330uF to 33uF for a 50Hz cutoff. So you see where the 47-100u ballpark figure comes from.

4- How low can we go?

How much money and space have you got?!?
By all means stick a 4700uF cap in there if you want. But like I said, historically, most pedal designers have been *total cheapskates*!!

5- If we disregard cost and inventory, can I use a bunch of parallel MLCC capacitors instead of a big electrolytic?

Yeah, good idea. Give it a try and tell us how well it works!

[mozz]

D8 is vital unless you want idiots and drunk people to be able to blow up your circuit at will.

Yes it is, but if he is having noise problems, easy enough to jumper and see if it helps.  Inductors, ferrite beads also help at times.  Try . 01 instead of 0.1.

[drdn0]

I had perpetual noise issues using cheaper power supplies, and no matter how much filtering I added in I could never get them super quiet even after shielding the hell out of my guitars, using star grounding, etc. What I eventually established that the radiated noise wasn't coming in through the DC feeds - it was getting picked up by my bloody leads/input jack wiring!

I added in a series 10k resistor/100pf cap to ground on every single pedal I've built since, and the noise floor almost disappeared.

[Rob Strand]

I'm also quite frustrated with noise issues, even after following as many EMC guidelines as I can. E.g: I have two 9V PSUs here. No name chinese PSU is quite silent, except it makes MOSFETs oscillate in some situations. I also have a 1Spot. Doesn't oscillate, but has super high noise floor.

If the power supply oscillates you are wasting your time trying to filter it.   That's a common design issue with SMPS power supplies with light loads -  like pedals, and especially those with MOSFETs and BJTs which are both light loads and have poor supply rejection.

I recommend adding a dummy load to the PSU.    Unfortunately it's not really practical to add the dummy load to each pedal.  It's best done with a "box" between the PSU and the pedal.

https://www.diystompboxes.com/smfforum/index.php?topic=130841.msg1271394#msg1271394

The reasons are discusses in this post.   The rest of the thread has other info.   The interesting thing about this particular thread is the problem turned out to be a ground loop problem.  ie. filtering doesn't remove all the problems.

https://www.diystompboxes.com/smfforum/index.php?topic=131537.msg1278429#msg1278429

Beyond that: If you follow the Boss designs (see later pedals) they use an active filter at the input.  That will reduce small levels of noise to very low levels.

[R.G.]

[...] I also have a 1Spot. Doesn't oscillate, but has super high noise floor.

I recommend adding a dummy load to the PSU.    Unfortunately it's not really practical to add the dummy load to each pedal.  It's best done with a "box" between the PSU and the pedal.

This is part of the line of thinking that led me to speculate that the 1Spot might either be a knockoff or defective. I have some insight into the 1Spot and, as manufactured, they are stable at zero external load. They used to be 100% listen-tested with [ a commercial pedal that is particularly prone to power noise ] but that may have been discontinued - they always passed. I don't know if it still happens, but at one time there were some clone-copy-knockoffs of the real 1Spot that had issues.

But yeah - adding a minimum load is not a bad idea for a single or few pedals. A real 1Spot won't care - one of these will happily power a couple of dozen or more pedals, so a minimum load won't hurt things, and could make other power supplies stable.

[Rob Strand]

This is part of the line of thinking that led me to speculate that the 1Spot might either be a knockoff or defective. I have some insight into the 1Spot and, as manufactured, they are stable at zero external load. They used to be 100% listen-tested with [ a commercial pedal that is particularly prone to power noise ] but that may have been discontinued - they always passed. I don't know if it still happens, but at one time there were some clone-copy-knockoffs of the real 1Spot that had issues.

That's the problem with fakes.   They are usually junk and they make the real products look bad because the users don't know it's a fake.

I have seen switch-modes go noisy when the input or output caps get old and the ESR goes up.

I'm sure you have seen crazy behaviours of SMPS's with light loads.  There's quite a few examples on the forum where the dummy load helps.  (Just to be clear: the dummy load thing helps particular *working* supplies.  It's not intended as a workaround for supplies with faults.)

[marcelomd]

Wow. Thank you guys for all the answers!


TL;DR: Designed preamp. Silent but horrible oscillation in some use cases. Designed improved version. Oscillation, but quieter, even with a benchmark power supply. I'm thinking that my layout skills must be really shitty, so I'm trying to learn a bit.


Longer version:

The particular pedal that started this was created in early 2023. Check it here. The pedal is mostly noise free. No hum or hiss or etc.

The problem is that it had real bad oscillation when using it with my DIY amp and any external PSUs I had back then. The amp had a cheap buck converter module as an integrated 9V for pedals. Can't hear oscillation with it. I didn't have the 1spot back then.

I don't have easy access to that setup anymore.


Later that year I created a second version with one 100nF per MOSFET, larger bulk electrolytic, plus some minor changes. I also bought the 1spot in a visit to the US.

Now I'm using the preamp with either a Focusrite interface or a Hotone IR loader with headphone outputs. The preamp oscillates veeery quietly. Also has some unexpected noise with the volume at zero. This is with the 1spot and the other chinese PSUs.


Yesterday I've tried adding a few pedals in a daisy chain. Oscillation is GONE (Yay!).

Beyond that: If you follow the Boss designs (see later pedals) they use an active filter at the input.  That will reduce small levels of noise to very low levels.

Do you have a schematic?

I added in a series 10k resistor/100pf cap to ground on every single pedal I've built since, and the noise floor almost disappeared.

The pedals I had problems with had a 1k/100pF filter at the input. They are very quiet, but prone to oscillation. MOSFET issues, apparently.

How much money and space have you got?!?
By all means stick a 4700uF cap in there if you want. But like I said, historically, most pedal designers have been *total cheapskates*!!

I meant "How low can I go with the bulk capacitance. Sorry. I'm working with SMD and want to use the smallest possible component size.

In the example you quote, does the Chinese PSU make MOSFETs prone to oscillation in all pedals, or just a specific one? Does the 1Spot have high noise in all pedals or just a specific one?
As an aside, is the 1Spot power supply a Visual Sound/Truetone made 1Spot, or a similar clone, and how old is it? It could simply be defective. High noise floor is a very unusual thing in real 1Spot power supplies.

The issue is with this preamp only. Oddly I can't get the higher noise floor today. It was very clear a month ago. The oscillation is there.

I bought my 1spot in the US late last year, from a real store. Date code 2212. How can I tell if it's genuine?

MOSFETs have very -very- high frequency responses. They are also prone to oscillation at frequencies that can be resonated with PCB stray capacitance and trace inductance. They  <want> to oscillate 8-) Any time you use a MOSFET in a linear amplification circuit, insert a 10R to 1K resistor in series with the gate lead as close to the MOSFET package as possible, and decouple the drain-source power supply locally with a ceramic cap as close to the drain-source power as you can; also, for MOSFET source followers, isolate capacitors from the source with a small resistor; followers, especially MOSFET followers, are prone to oscillating with capacitive loading.

Like this?


Re: grounding. I'm using a full board, uninterrupted, ground plane for everything. Should I use traces only?

The standard model is something that evolved over time to solve most power filtering for small circuits in pedals. It's a starting point suggested by history. I use my own variant of the standard model for new designs I do.

Can you share your version?

[Rob Strand]

Do you have a schematic?

The active filter loses 0.7V or so.   The loss is roughly equal to how much ripple can be removed before the circuit drops out and saturates.

Notice the HM-3 doesn't bother with an input cap.

Going the other way, a 10 ohm to 100 resistor + 100uF before the active filter has a lot of HF roll-off.

[bluebunny]

Slight deviation: @R.G. - is there anything obvious to a punter that may give away a fake 1Spot?  (Would telling us give the game away to the counterfeiters?)

[marcelomd]

The active filter loses 0.7V or so.   The loss is roughly equal to how much ripple can be removed before the circuit drops out and saturates.

More or less like a linear regulator, then?

What about inductors/chokes? Can I get decente filtering with a smallish one?

[Rob Strand]

More or less like a linear regulator, then?

Yes similar, kind of does a similar job but without actually regulating.

What about inductors/chokes? Can I get decente filtering with a smallish one?

The LCR filters work better at higher frequencies.   It all comes down to specifics: what frequencies you want to reject.  With switch-mode power supplies you could argue removing 50/100Hz ripple is unnecessary and that changes what frequencies need to be filtered.

Here's a mini survey of some typical filters:



You can see at these low frequencies the LC filter isn't *that* great.   Note also the inductors aren't really small but are still reasonable in size.

This post shows what happens when you have a lot of 100Hz ripple with an active filter.   So the take home message is they work great provided the input ripple isn't high.   Very high ripple would be filtered with an RC filter at the expense of more DC voltage drop.

https://www.diystompboxes.com/smfforum/index.php?topic=127644.msg1225116#msg1225116

There's no fix-all solution for a pedal.   You have to draw a line what you want to cope with.  IMHO trying to add a filter which will clean-up an oscillating switch-mode is going to far - time to push back on the PSU itself.

[R.G.]

I'm thinking that my layout skills must be really shitty, so I'm trying to learn a bit.

Layout skills are developed over a long time, just like guitar playing skills. Don't get discouraged, and just keep working at it. I've been involved in layout since it was done on mylar sheets with tape. I found a couple of new tricks this morning. The really big issue lurking under PCB layout is the need to understand what the electrons are doing on that trace you're laying down, for every trace and every section of trace. Keep at it and shittyness decreases; I look at my earlier stuff and slap my forehead, then mutter in disgust.

Yesterday I've tried adding a few pedals in a daisy chain. Oscillation is GONE (Yay!).

Problem solved. Rob's insight about minimum load is strongly, strongly supported by this.

The pedals I had problems with had a 1k/100pF filter at the input. They are very quiet, but prone to oscillation. MOSFET issues, apparently.

As a good rule of thumb, any time a MOSFET is being used as an amplifier and not a switch, put a small resistor in series with the gate as close as you can get the connected end to the actual gate lead.
Relating this to your schematic, it goes RIGHT AT THE GATE PIN.

The reason is that a MOSFET gate looks like a high-quality capacitor to the source. Any inductance, no matter how small, can form a resonant circuit with a lot of phase shift right at the gate. Using a resistor there destroys the Q-factor of the resonant circuit. Making the gate-to-resistor trace as short as possible, in fact zero if possible, reduces the inductance and might push the resonance above even the MOSFET's frequency range. For SMD, making the resistor pad nearly touch the MOSFET gate pad and running a few mills of trace between them is close to ideal. It's "close to" because most PCB software requires spacing between solder mask and other traces. Laying out everything as small plane areas with few-mil spaces between them and a ground plane converts the resistor-to-gate trace into a low inductance, high capacitance transmission line damped by the resistance.

I bought my 1spot in the US late last year, from a real store. Date code 2212. How can I tell if it's genuine?

US store, from 2212, it's probably genuine. The clonefakes were mostly online orders on ebay. Very few stores had them.

Re: grounding. I'm using a full board, uninterrupted, ground plane for everything. Should I use traces only?

Not necessarily.  Ground planes are neither a panacea nor a problem. It's like I mentioned above, you have to keep in mind what the electrons (and possibly fields) are doing on that trace, compared to the ground plane. Ground planes are a good way to get as much low-resistance copper between two "grounded" points as possible. That works great, even at DC, excepting when you have large currents sharing the ground plane. Even a ground plane has some resistance, so there is some ground offset voltage from current flow.
Realistically, it's not an issue for the kinds of currents in your preamp, but the point is to know what's going on. With a few hundred db of gain, even sub-microvolts of V = I*R get audible. If a ground >>reference<< point has a zero current ground trace back to the power source coming onto the board, it has zero ground offset from the power source. If a reference shares part of a ground plane path with a higher current device, its reference voltage is wobbled around by the higher current by V = I*R. So the issue becomes exactly where on the ground plane your reference voltage for an amplifier stage is connected. At DC (and effectively, audio), current follows the lowest resistance path back to the power source. Usually, this is the shortest straight-line path, with a declining-square current density as distance increases from the shortest straight line. A big opening in your ground plane for other-side traces can pervert the path. Maybe fine, maybe not; I always look at the holes in a ground plane. At higher frequencies, the fields force return current flow to follow the path under the current carrying signal conductor. Holes and dropouts in the plane cause the ground current to detour from it's trace-following preferred path and you get small radiating antennae in the bend around plane interruptions. High speed logic and RF really needs an uninterrupted plane, or a plane with interruptions less than a quarter wavelength of the highest frequency you're worried about. A hatched plane is nearly as good as a solid plane if the wavelengths are longer than four times the hatching interval.
But I'm blathering. Where on the plane your zero-volt reference for an amplifying stage connects can matter.

The standard model is [...] starting point]...] 

Can you share your version?

Sure, but it's going to be underwhelming.
> Power input wires/pads connect to polarity protector. Series diode is easiest and cheapest, but loses you a diode drop of DC voltage. Shunt diode protector is OK, but must be itself protected from a long-term reverse connection by a PTR resettable fuse. This drops a little voltage in the PTR resistance, but makes the thing just about bomb proof. Series MOSFET "ideal diode" works fine and reliably, but costs more in parts and PCB space.
> Polarity protector connects to bulk cap. I used to go bigger and smaller, but usually now I use 100uF 25. Why 25V? Guitarist have taken to hooking up higher voltages to pedal "to see if it sounds better". Sigh. Bless their hearts... I don't use anything less than 25V for electros any more unless there is no other way.
> One space for a ceramic cap per IC, close to the IC. I just do it, then delete them off the parts list if they can be dispensed with. That way the PCB/product can be cost reduced without a new PCB layout.
> Depending on needs, I may use a PNP bipolar as a power off/on switch driven by a resistor from base to the ring contact on an input jack. It's of questionable practicality in a world of AC power adapters, but for only-batteries pedal setups may be needed.
> There is a theorem/practice in design automation that says to wire the nets with the fewest number of pads/nodes in them [edit..] first. Two pin nets first, and power/ground nets last, in other words.
In practice, I wire up power next to last and ground last, having duly been aware of where the ground pins are. At various stages I will highlight all the ground pin nodes that have been laid out to ensure that it's still at least theoretically possible to reach them. With everything else where it needs to be, wiring up ground can be based on ground currents and types, and parts can be locally reworked to make a section of the board fit better with ground's needs. It makes grounding tradeoffs easier to focus on.

[R.G.]

Slight deviation: @R.G. - is there anything obvious to a punter that may give away a fake 1Spot?  (Would telling us give the game away to the counterfeiters?)

No, sadly. The market may have flushed them out anyway. It was an issue a few years ago. We got quite a few calls about non-performing "1Spots". We started trading out a brand new 1Spot for the wonky ones to capture some of the fakes, so I got to tear a few of the fakes apart. There were at least three varieties, but externally they were pretty good copies. But circuits don't lie - the insides were different.

Ah. I remembered one consistent different thing. The DC output wires on real 1Spots are much thicker than you'd expect from the rest of the wall adapter market. The fakes I found had much skinnier DC wires and chintzier plugs. Those are more expensive to fake than writing and surface finish on a plastic cover.

Pretty much it comes down to buying from a reliable seller. One good thing about small companies is that they can do things that might not make sense for big ones. There is a preference for selling through local shops and dealers and not through internet sellers, with only a few exceptions. There really isn't a gray market, so if it's a funny seller, and a too-good price, that's reason to be cautious.

[bluebunny]

Good to know.  Not so much worried about my own brace: both bought from established music kit vendors and both with the seriously chunky wire.  (And they both work absolutely fine, of course!)

[marcelomd]

TL;DR: I'm designing an improved version with the insigths here. Might take a while, though.

If this one doesn't stop oscillating, I'm going back to opamps...

Thanks a lot!

Yes similar, kind of does a similar job but without actually regulating.

I read it's a capacitance multiplier, right?

Here's a mini survey of some typical filters:

Hmmm, looks like the simple RC is good enough.

Relating this to your schematic, it goes RIGHT AT THE GATE PIN.

Got it. I'm designing an updated version with this extra resistor. This broke my beautiful layout... oh well...

Sure, but it's going to be underwhelming...

By PTR you mean PTC resettable fuses, right? Any recommendations here?

Ah. I remembered one consistent different thing. The DC output wires on real 1Spots are much thicker than you'd expect from the rest of the wall adapter market. The fakes I found had much skinnier DC wires and chintzier plugs. Those are more expensive to fake than writing and surface finish on a plastic cover.

Yeah. The wires and connector on my unit are CHUNKY.

> There is a theorem/practice in design automation that says to wire the nets with the fewest number of pads/nodes in them [edit..] first. Two pin nets first, and power/ground nets last, in other words...

I used to route power and ground first. Started doing it later, and using ground traces even if I'm using a ground plane, just to guarantee the path I think is best.

[R.G.]

By PTR you mean PTC resettable fuses, right? Any recommendations here?

Ooops. Yes, PTC fuses.
I think they're especially critical to use if your pedal uses the reversed-diode polarity protection circuit. The adapters I know best ( 1Spot ) will supply a little less than 2A of 9Vdc, but will shut down and restart if the current is higher. Other kinds/brands of power supplies may or may not do a shutdown, and can simply current limit. If that happens, the typical 1N4000 series reverse protection diode can be forced to run at more than its 1A average current rating and can either die shorted, or melt the solder off its leads. A PTC will overheat and open, letting things cool down before reconnecting, so the diode and the circuits (and the power adapter !) not cook to death.
Picking a PTC is an exercise in picking one with a guaranteed current rating higher than your pedal needs for normal operation, but less than the disaster-current value of the (unknown) power adapter and any polarity protection circuits. So pick one with a constant current carrying rating of maybe 100ma if your circuit needs less than that, and with a trip current of under an amp. Not too hard to find. PTCs are rated for their ability to interrupt voltage too. In AC mains situations that can matter, but for low voltage pedals, nearly any of them are OK.

I used to route power and ground first. Started doing it later, and using ground traces even if I'm using a ground plane, just to guarantee the path I think is best.

That works fine. I tend to think in terms of isolated subsections, with a given circuit section - for instance, an input buffer circuit, or maybe a clipping stage - as its own little lump of circuit, with signal in and out, and then power in and out (ground), and a few wires going between the lumps.
I had a course in electronic design automation where they introduced us to the "minimum-connections cut set". Imagine a schematic of your circuit. Where can you draw a red line through it that divides the circuit into two parts so that the line cuts the fewest number of wires?
The minimum number of wires will be two: one signal and one ground. This is the case where a signal wire goes between two active/powered sections. The number is three for signal, power, and ground; this is for two signal processing stages separated by a signal line between them. And about here is where you realize that you might as well not count ground at all, as it has to go everywhere. Power lines can be ignored as they go nearly everywhere.
So if you do a number of cut lines, each with a cut number of one (ignoring power and ground), you have split your circuit into sections that interact only with a signal wire, and the sections can be laid out separately, with only signal wires between them. Other considerations then come into play, like the impedances of power and ground traces, capacitive/emi interference, etc., but for just signal function, the sections are independent.
This is really just a professor's way of formalizing the idea of keeping tightly connected parts of the circuit close together, but it's stated in a way that you can program a computer to run through the process of doing cut sets and figuring out what sections to keep together and which an be separated. But I find it a useful way to think about the circuits.