C2C King Nothing - great oscillator, terrible pedal

[drdn0]

Howdy,

I built a King Nothing. Everything works fine, voltages are spot on (254v on the plate resistors, and 29v heading into the J201), but there's significant amounts of oscillation that decreases as the gain is turned up. Low switch has no impact on the noise, but the left position of the mode decreases the range where it oscillates from about 1-9 to 1-3 (assuming each tick is about 30 degrees of rotation).

Weirdly enough, it oscillates MORE when it's in bypass - when it's in bypass, grounding the input stops the noise (even though the switch PCB is meant to ground the input when it's off), when the pedal is on grounding the input changes the frequency of noise by a few hundred hz but doesn't stop it. It oscillates at 2500-3200hz, with increases in tone increasing the frequency, and increases in gain decreasing the frequency/volume.

I've tried multiple power supplies, from isolated DC bricks, to cheap SMPS to an old unregulated power supply I've got around (all capable of +500ma), and no difference. 9/12v are identical; no difference in frequency of volume of oscillation. Tried multiple guitars, multiple leads, and multiple amps with no difference.

The pedal is built as per the schematic (except for using a IRFB17N50 instead of IRF740; I have used these on other builds with no problems), and I have verified all components. I have reflowed the entire board, scoped all joints and also cleaned the PCB's off with an ultrasonic IPA bath in case there was any schmoo on there. Photos of the PCB below are pre-bath.

Tube was a brand new Shuguang 7021 (confirmed to be good in other pedals), J201 is a onSemi MMBFJ201 on a SOT23>T092 adapter board.

Since originally building it, I have:
*verified all components on the board and polarities
*swapped the J201 and adaptor for a TO92 2N5458 (no difference)
*swapped to other tubes, both other 7021s and EHX 12AX7 (no difference)
*reduced the plate resistor from 220k to 100k (no difference)
*paralleled RF suppression caps at the input (no difference)
*added Miller cap on the JFET (no difference)
*moved input/output wires around, tapped components, poked stuff (zero difference with anything)
*used a disconnected lead on the input to see if I could isolate the oscillation (the frequency increases from ~2600hz to 3100hz if I'm poking the tube, but that's it - no increase/decrease in volume).

There are probably a dozen successful builds I have read of, but I've also read of a few people having the same oscillation issues that were never solved.

Thoughts?

[drdn0]

Ok, so just for shits and giggles I wrapped the entire tube/socket in aluminium foil and earthed it directly on the power jack.

It probably lead to a 40% reduction in oscillation - weirdly enough now it oscillates from somewhere between 2 and 8 gain - at either max or min it is quiet. Still oscillates more on bypass than when the pedal is on, and still utterly unusable; the oscillation is significantly louder than my guitar tone coming through.

I also remembered aluminium is a phenomenal conductor for B+ after a little slip 

I also saw R9 (triode 1's grid stopper) is a mile away from the grid, so I removed it, ran a shielded jumper from there to the tube directly (with R9 right against the tube) - no change either.

[Chillums]

Did you use a UF4007?  I'm pretty sure it is a must have in the power supply.

[GibsonGM]

Did you use a UF4007?  I'm pretty sure it is a must have in the power supply.

It is, or an equivalent high speed diode.

[drdn0]

Did you use a UF4007?  I'm pretty sure it is a must have in the power supply.

Yes. As I wrote, every other component except for the MOSFET/FET is as per the schematic.

I have built probably half a dozen power supplies based off a 555, none had any switching noise whatsoever.

[duck_arse]

scope the output pin of the 555. what frequency? can you scope the noise in the audio section as well?

[R.G.]

DA's right - first find out if the oscillator is emitting/leaking into the audio path.

More generally, the board seems to have the (very) high voltage oscillator section close to the J201. The J201 would be more than happy to pick up high frequency spikes.

More generally still, grounding and routing will be critical on a layout like this, especially on the exact manner of grounding the 555 (which pulls sub-microsecond amp-size spikes by itself) and its MOSFET and inductor/rectifier setup so that ground spikes don't leak back into the grounding of the J201.

[PRR]

at either max or min it is quiet.

The receiving antenna is probably the wire from the pot wiper.

[drdn0]

scope the output pin of the 555. what frequency? can you scope the noise in the audio section as well?

About ~40khz. Ran out of time to scope the rest of the audio path, but it doesn't change when I can manipulate the oscillation frequency with knobs so I'm not sure its directly the issue (although that's not saying that it isn't getting modulated anywhere else in the circuit).

DA's right - first find out if the oscillator is emitting/leaking into the audio path.

More generally, the board seems to have the (very) high voltage oscillator section close to the J201. The J201 would be more than happy to pick up high frequency spikes.

Are you looking at the switching transistor for the PSU, not the FET? FET and the HV section are on opposite sides of the board, and there's not many traces from anywhere pre-tube that goes near HV from a quick squiz.

at either max or min it is quiet.

The receiving antenna is probably the wire from the pot wiper.

That's what I was thinking re: the grid stopper routing being right up in the guts of the PSU, but moving it offboard didn't make a difference either. I didn't remove the R9 trace from being connected to the rest of the circuit so could still be causing grief I guess.

[Rob Strand]

That's what I was thinking re: the grid stopper routing being right up in the guts of the PSU, but moving it offboard didn't make a difference either. I didn't remove the R9 trace from being connected to the rest of the circuit so could still be causing grief I guess.

I tried to understand your problem earlier but the precise wiring of the switches wasn't clear from the schematic - so I couldn't decipher what "but the left position of the mode" meant in terms of the schematic.

Given the low frequency nature of the oscillations, the effect of the filters and the reversed pot behaviour,  I was suspicious you have a motorboating problem.    This might be promoted by the ground trace issues on PCB layout.   Perhaps when you use certain 10uF PSU caps they don't work as well as others and that edges the units into motorboating.   If it was motorboating you might expect different tubes to at least behave a little differently. (And yes I am aware that wrapping the circuit in foil probably shouldn't affect motorboating, not directly anyway.)

Another way junk can get in is via the tube heaters.   There's been a case or two on this forum along those lines.   At this point I wouldn't go down that path yet.

[drdn0]

I did the final R9 relocation - with it now located on the back of the tube socket daughterboard. Again, noticeable reduction in oscillation (and a big reduction in noise floor), but pedal is still unusable as it is.

I'm gonna call it a wash I think.

I tried to understand your problem earlier but the precise wiring of the switches wasn't clear from the schematic - so I couldn't decipher what "but the left position of the mode" meant in terms of the schematic.

Left position corresponds to Mode 2 being grounded

[drdn0]

Further fun fact - somebody else noticed that the inductor is labeled 104, whereas 101 should correspond to 100uh.

I ordered 100uh inductors.

They are listed as 100uh inductors.

The datasheet says they're 100uh inductors.

Measuring them, they're 100uh inductors.

Coilcraft apparently just use whatever code they feel like on the day for marking 

I found my scope:
*Power supply has ~200mv of ripple @ 42khz after the filter cap.
*Zero noise to the grid stopper of the first triode
*~10v p2p at 2600hz at the plate resistor.
*~22v p2p at the next triode plate resistor.
*~10v p2p at the volume pot
*changing the tone controls changes the frequency of the oscillation all the way back at the first triode.

[Rob Strand]

Further fun fact - somebody else noticed that the inductor is labeled 104, whereas 101 should correspond to 100uh.

I ordered 100uh inductors.

They are listed as 100uh inductors.

The datasheet says they're 100uh inductors.

Measuring them, they're 100uh inductors.

Coilcraft apparently just use whatever code they feel like on the day for marking

It came up once before on the forum.  The different marking code is in the documents.   Nonetheless it's a weird decision to mark the parts like that.

I found my scope:
*Power supply has ~200mv of ripple @ 42khz after the filter cap.
*Zero noise to the grid stopper of the first triode
*~10v p2p at 2600hz at the plate resistor.
*~22v p2p at the next triode plate resistor.
*~10v p2p at the volume pot
*changing the tone controls changes the frequency of the oscillation all the way back at the first triode.

Try putting a dummy load across the supply.

[drdn0]

Try putting a dummy load across the supply.

Did you mean the 555 supply?

I superimposed both sides of the PCB, and drew the input/output jack traces (green), the first grid stopper (white) and the output to the 3PDT (orange).

The input jack trace goes right next to both plate resistors and the coupling caps.
The output jack trace goes right underneath the inductor, MOSFET AND switching diode.
The trace from the first grid stopper hits most of the high points of the onboard supply including the snubber cap/resistor.
The output trace from the volume pot is bang in line with the output jack trace, going right through the guts of the switching supply.

Honestly, I'm not really sure this PCB ever stood a chance of not oscillating

[Rob Strand]

Did you mean the 555 supply?

Yes.  A of grasping at straws to see if it does anything.

I superimposed both sides of the PCB, and drew the input/output jack traces (green), the first grid stopper (white) and the output to the 3PDT (orange).

The input jack trace goes right next to both plate resistors and the coupling caps.
The output jack trace goes right underneath the inductor, MOSFET AND switching diode.
The trace from the first grid stopper hits most of the high points of the onboard supply including the snubber cap/resistor.
The output trace from the volume pot is bang in line with the output jack trace, going right through the guts of the switching supply.

Honestly, I'm not really sure this PCB ever stood a chance of not oscillating

[/quote]
Yes sure sounds like there's a lot of problems with the PCB.

In the past I have judiciously cut tracks and replaced them with point to point wires.  Fairly brutal but at least it has more hope of working.   I remember on one board I cut the ground tracks quite a bit and replace them with wires.  It's not really that messy when you compare it against vero layouts.

[drdn0]

Did you mean the 555 supply?

Yes.  A of grasping at straws to see if it does anything.

I superimposed both sides of the PCB, and drew the input/output jack traces (green), the first grid stopper (white) and the output to the 3PDT (orange).

The input jack trace goes right next to both plate resistors and the coupling caps.
The output jack trace goes right underneath the inductor, MOSFET AND switching diode.
The trace from the first grid stopper hits most of the high points of the onboard supply including the snubber cap/resistor.
The output trace from the volume pot is bang in line with the output jack trace, going right through the guts of the switching supply.

Honestly, I'm not really sure this PCB ever stood a chance of not oscillating

Yes sure sounds like there's a lot of problems with the PCB.

In the past I have judiciously cut tracks and replaced them with point to point wires.  Fairly brutal but at least it has more hope of working.   I remember on one board I cut the ground tracks quite a bit and replace them with wires.  It's not really that messy when you compare it against vero layouts.
[/quote]

Given how noisy it is even in bypass, do you think that disconnecting the jacks and running shielded wires down to the  switch could be a worthy of trying? Honestly I was just going to bin it after wasting dozens of hours trying to get it to work, but I'm infuriated by how much money PCB's/parts cost me that can't really be used for anything else 

[duck_arse]

one more scoping question - what's happens [and how often] at pin 5 of the 555?

Further fun fact - somebody else noticed that the inductor is labeled 104, whereas 101 should correspond to 100uh.

a quick quiz - what is the three number marking for a 100nF capacitor? and what would be the three number marking for, say, a 470nH inductor?

[Rob Strand]

Given how noisy it is even in bypass, do you think that disconnecting the jacks and running shielded wires down to the  switch could be a worthy of trying? Honestly I was just going to bin it after wasting dozens of hours trying to get it to work, but I'm infuriated by how much money PCB's/parts cost me that can't really be used for anything else

I would be try to separate the PSU and audio circuits.    That means having a separate ground for the PSU and the audio circuit which join at one point only.   A problem with switchers like this one is if you switch the power off with the input jacks there is a risk the current pulses from the switcher pass through the jack and get into the audio.   You could do a test where the input power is not switched by the jack.  The input power goes to the switcher ckt, the audio jacks go to the audio ckt, then bridge the grounds at one point.   You might also have to separate the heater supply.

If you do want to switch the power then there is a trick where you can use a MOSFET to switch the power of the noisy circuit and only pass light and quiet currents through the jack "switch" contacts.   I think RG had something on his website.  There was also a project or two which used the trick.    I remember a long thread with forum member Marcos-Munky regarding a switcher and battery pack, which had all the links - it might be in the
Lounge.

Here's RG's link,
http://www.geofex.com/Article_Folders/mosswitch/mosswitch.htm

It's always annoying trying to fix bad PCB layouts.   The way I look at it is you have any intent on doing another layout, the best platform for getting it right next time is to get the old PCB working.  No matter what it takes.  No matter how messy it looks.   The reason is what you think might be fixed on the new layout might not fix it.   Maybe there's many problems and the new layout only fixes 90% of them.   Getting the old PCB working forces you to address everything and it forces you to prove what you *think* will fix it actually does.  For a one off you just need it to work, so don't hold back on butchering the PCB into shape!

[drdn0]

Given how noisy it is even in bypass, do you think that disconnecting the jacks and running shielded wires down to the  switch could be a worthy of trying? Honestly I was just going to bin it after wasting dozens of hours trying to get it to work, but I'm infuriated by how much money PCB's/parts cost me that can't really be used for anything else

I would be try to separate the PSU and audio circuits.    That means having a separate ground for the PSU and the audio circuit which join at one point only.   A problem with switchers like this one is if you switch the power off with the input jacks there is a risk the current pulses from the switcher pass through the jack and get into the audio.   You could do a test where the input power is not switched by the jack.  The input power goes to the switcher ckt, the audio jacks go to the audio ckt, then bridge the grounds at one point.   You might also have to separate the heater supply.

If you do want to switch the power then there is a trick where you can use a MOSFET to switch the power of the noisy circuit and only pass light and quiet currents through the jack "switch" contacts.   I think RG had something on his website.  There was also a project or two which used the trick.    I remember a long thread with forum member Marcos-Munky regarding a switcher and battery pack, which had all the links - it might be in the
Lounge.

Here's RG's link,
http://www.geofex.com/Article_Folders/mosswitch/mosswitch.htm

It's always annoying trying to fix bad PCB layouts.   The way I look at it is you have any intent on doing another layout, the best platform for getting it right next time is to get the old PCB working.  No matter what it takes.  No matter how messy it looks.   The reason is what you think might be fixed on the new layout might not fix it.   Maybe there's many problems and the new layout only fixes 90% of them.   Getting the old PCB working forces you to address everything and it forces you to prove what you *think* will fix it actually does.  For a one off you just need it to work, so don't hold back on butchering the PCB into shape!

It's a PCB I bought - not made.

Definitely not how I would make anything with a switching supply (and I'd probably use subminiatures because it simplifies everything so much)