A few of my pedals have terrible RF noise. I fixed it. Help me understand how?

[drdn0]

Ok, so I'm not a total beginner to electronics but I'm definitely not an expert. I've made a series of small effect PCB's. These are similar (but not direct copies of):

• DAM Meathead
• Crackle-Not-Okay SHO
• Devi Ever Hyperion
• COT50
• Acapulco Gold

All of these have a universal offboard bus which connects to a filtering/switching board which I've used across all the PCBs. This is nothing fancy except for jacks/3PDT/LED, along with some basic RC filtering (47r resistor, 100uf capacitor) on the power. I've done all the testing with these powered off of a Mosky ISO10 supply. This has worked for quite some time, with all the boards reasonably low noise (some minor RF noise, but nothing unmanageable) and sounding great.

I have created a few new boards, using the same switching/filtering board as before, and SOME have a crazy amount of additional RF noise that is coming from my ISO10 - not through the power rails, but through the guitar leads.

• NPN silicon Rangemaster - horrendous RF noise powered off the ISO10, and still horrendous noise if the ISO10 is on my desk plugged in but not being used
• Lizard-Queen-ish Fuzz - medium amount of noise at max volume if using the ISO10, tiny amount if it's not being used and on my desk -but ONLY if my guitar volume knob is turned all the way down-
• Woolly Mammoth - zero additional noise whatever
• MOSFET Boost>LM386>SWTC - only noise if my guitar volume is rolled all the way down and I'm using the ISO10 to power it

I started looking if there was any way to limit noise that was being induced on the effect input (probably on my leads). I put a 1nf capacitor in parallel with the RPD (1M) on each of the noisy effects, and I'd say 75% of the noise disappeared straight away with no real impact on sound. I was on the right track. Seeing if I could do better, I soldered a 1nf MLCC across the input jack tip/sleeve directly, and I'd say 85-90% of the RF noise was gone entirely. Both the Rangemaster and the 386-based distortion still had noise, but only if it was powering the effects and my guitar volume was all the way down.

I then looked into enclosure grounding. Everything inside the enclosure is insulated, but I have a small wire running from the ground plane on the switching board that is sandwiched between the 3PDT and enclosure. Out of interest, I removed this wire and it made minimal to no difference to the RF noise (either the original, or the additional noise from the PSU). While poking around, I connected a 100nf MLCC between the input jack sleeve and the enclosure, and boom - all noise was instantly and immediately gone, including a significant amount of the RF noise I had on the 'good' effects.

I tried it (removing the enclosure ground and having the capacitor between them) with all the original pedals which had no PSU noise and only some background RF, and the noise floor dropped -significantly-. As in, almost entirely gone.

I can understand how the 1nf capacitor could work, although I'm not sure why it was such a problem on a few pedals but not on others (none of the 'good' ones have a similar cap setup on the input). What I don't get how the MLCC to the enclosure ground could be -so- effective, and why? I asked a few EE friends (one in powerlines, one in automotive control systems) and as soon as I mentioned it was an RF problem, they clapped their hands over their ears and ran away screaming (well not really, but they independently suggested it was probably ghosts).

Thanks for feedback!

[antonis]

1nF cap forms a LPF with any resistance before it (signal source impedance included)..

For more effective RF elimination, place that cap across B-E junction of the first transistor..

[ElectricDruid]

1nF cap forms a LPF with any resistance before it

...and this is a key point. Have your circuits got a resistor in series with the input? They should. It serves two purposes:

1) If someone tries to whack the pedal with a super-high voltage, it limits the amount of current that can flow, hopefully to something safe.
2) In tandem with the little cap to ground, it filters RF or other high frequency noise.

1K is a pretty common value to see in this position. The 1K forms a voltage divider with the 1M, but since that's a 1:1000 relation, it's not like you're losing loads of signal!
You could probably go a bit higher with you 1n cap value, even after you add the 1K input resistor. Try 2n2, 3n3, 4n7 and see what you think.

[drdn0]

1nF cap forms a LPF with any resistance before it (signal source impedance included)..

For more effective RF elimination, place that cap across B-E junction of the first transistor..

I guessed that would be the case, but didn't think it would end up being in the audio range. I could definitely be (and am probably) wrong on the math though. I'm really not sure why some pedals are terrible and some are great though.

That bit is fine and understandable if it's just bad math on my behalf, but I don't understand the capacitive coupling to ground being so, so much more effective than direct coupling the enclosure side of things.

1nF cap forms a LPF with any resistance before it

...and this is a key point. Have your circuits got a resistor in series with the input? They should. It serves two purposes:

1) If someone tries to whack the pedal with a super-high voltage, it limits the amount of current that can flow, hopefully to something safe.
2) In tandem with the little cap to ground, it filters RF or other high frequency noise.

1K is a pretty common value to see in this position. The 1K forms a voltage divider with the 1M, but since that's a 1:1000 relation, it's not like you're losing loads of signal!
You could probably go a bit higher with you 1n cap value, even after you add the 1K input resistor. Try 2n2, 3n3, 4n7 and see what you think.

So 1k in series + 1nf (or more) to ground could just be something I could have on all the off-board wiring as it would be pretty universal for most pedals? Oddly enough, none of my others have any series input resistance or a cap to ground and none have any of this specific PSU noise at all

[antonis]

It has nothing to do with PSU noise..



C3 could be wired to GND instead of BJT Emitter..
(it's there mainly for RF rectification prevention..)

[drdn0]

It has nothing to do with PSU noise..

In the sense that's it's RF noise coming from a PSU that plays along perfectly with 80% of previous pedals I've built, not that's it's coming from the power feed

[antonis]

I guessed that would be the case, but didn't think it would end up being in the audio range. I could definitely be (and am probably) wrong on the math though. I'm really not sure why some pedals are terrible and some are great though.

Don't know the maths you used but I believe you aren't wrong..
Simply, you don't consider that RF is demodulated into the audio band by semiconductor devices..
That kind of passive 1st orfer LPF should be placed as close to the input jack as physically possible to prevent RF from being radiated inside the enclosure before it is shunted to ground - or else, any further attempt at RF filtering should be pointless..
(I think you've already verified the above by placing 1nF cap directly on input jack..)

[GibsonGM]

I often will place something like a 1n cap to ground from my input jack in tube preamp builds.  This shunts RF and anything else that is very high and nasty to ground, taking it right out of the way at the front!   Some people us a slightly higher cap value, say 10n, to reduce fret squeal and then boost highs preferentially later in the preamp.    This is 'to taste' - a smaller value such as 470p might work very well for your problem and be transparent in terms of your tone.

[idy]

If you put those 1n to ground on the in jack of all five pedals and use them together, do those LPFs accumulate in a less-than-desirable manner? Curly cable territory? It's not really "true bypass" anymore does it matter?

[Rob Strand]

C3 could be wired to GND instead of BJT Emitter..
(it's there mainly for RF rectification prevention..)

The input stuff is the most sensitive but don't forget R9, 470R to 1k, even on opamp designs.

[antonis]

So you are not a "purely resistive load" and "zero stray capacitance" believer, Rob..

[GibsonGM]

If you put those 1n to ground on the in jack of all five pedals and use them together, do those LPFs accumulate in a less-than-desirable manner? Curly cable territory? It's not really "true bypass" anymore does it matter?

If the caps are chosen properly (small values), they shouldn't have undesirable effects. In fact, I'd encourage the OP to try the smallest cap possible (100p?) that will stop the RF.   The frequencies involved are above the audible range...

[antonis]

If the caps are chosen properly (small values), they shouldn't have undesirable effects. In fact, I'd encourage the OP to try the smallest cap possible (100p?) that will stop the RF.   The frequencies involved are above the audible range...

It depends on preceding effect out pot value/setting, Sir..

[drdn0]

I guessed that would be the case, but didn't think it would end up being in the audio range. I could definitely be (and am probably) wrong on the math though. I'm really not sure why some pedals are terrible and some are great though.

Don't know the maths you used but I believe you aren't wrong..
Simply, you don't consider that RF is demodulated into the audio band by semiconductor devices..
That kind of passive 1st orfer LPF should be placed as close to the input jack as physically possible to prevent RF from being radiated inside the enclosure before it is shunted to ground - or else, any further attempt at RF filtering should be pointless..
(I think you've already verified the above by placing 1nF cap directly on input jack..)

Ahh right, so it is black magic. Perfect 

If you put those 1n to ground on the in jack of all five pedals and use them together, do those LPFs accumulate in a less-than-desirable manner? Curly cable territory? It's not really "true bypass" anymore does it matter?

Funnily enough, all these boards aren't really for me or going to be used together - I've been teaching beginners how identify and read schematics, what basic components are/do and how to solder. I did half of an electrical engineering degree 15 years ago, and most of the stuff we covered (up to basic filter topology) came back to me pretty quick. Halfway through we swapped from electronics (interesting) to power engineering (not interesting), and didn't cover any of the juicy stuff that would actually be nice to remember now.

I've had about 20 attendees so far (16 on my own PCBs), and we've had a 100% success rate on boards I have designed so far. I thought I had fluked my way into being a half-competent basic PCB designer until all this nonsense happened 

[Rob Strand]

So you are not a "purely resistive load" and "zero stray capacitance" believer, Rob..

RF can get in through the output wires as well.   You would expect this happens when the external RF is fairly strong.

Only a few weeks back,
https://www.diystompboxes.com/smfforum/index.php?topic=130316.0

Other sources of RF can be the whole device oscillating due bad layout of wires inside the enclosure, especially around the switch.

An input cap *to ground* would be first on the list, perhaps in combination with an input resistor (R3 on your schematic).

[ElectricDruid]

So 1k in series + 1nf (or more) to ground could just be something I could have on all the off-board wiring as it would be pretty universal for most pedals?

Yes, I'd say so. It's certainly "best practice", even if that means it's not something I've always practised myself!

Oddly enough, none of my others have any series input resistance or a cap to ground and none have any of this specific PSU noise at all

That's not a surprise. Some circuits are going to be much more affected by high frequency noise than others. Some have "built-in" lowpass filtering that would render them largely immune. Others (digital circuits particularly) would really hate it.

[drdn0]

FWIW, this is one of the boards that's dead quiet - Woolly Mammoth-ish. All the others have a pretty identical layout and setup.

Yes I have bottom jacks.

No I have no idea why I decided to go with it.

Please ignore the flux residue on the boards - lighting makes it look way, WAY worse than it actually is.

[MrStab]

As others have alluded to, you should make that cap as small as possible and stick some fixed series resistance before it, because a 1M log guitar volume pot turned halfway down (assume 150k in series with signal) will interact with a 1nF cap to roll off 1,061Hz and upwards, which you probably don't want unless you're into the darker genres of metal!

33k/100p isn't the end of the world in terms of impedance change or nyquist noise. If such a high-value resistor causes unwanted attenuation of the whole signal in certain circuits, then certain circuits have always been a trade-off of signal processing imperfection.

[drdn0]

Ok, so this is what I've got. Decided to keep it true true bypass and have C3 + R3 after the switch - can't have any snobs complaining it's not the TRVEST of bypasses.

Those specific values might change, but I think this will probably do the trick. C1 is absolutely overkill, but it doesn't take up much more board room than the 100uf I'd be using otherwise (with less voltage drop as I can lower R1)

[drdn0]

Ahh small update, a few of the pedals don't like the 1nf to ground - in particular, the SHO loses -all- top end.

Back to experimenting!