Radio interference fixed by 1 cable. Why?

[LaloFP]

Hi! I was building a Keeley C4 compressor and was getting a LOT of radio (even with sustain at min).

I "fixed" it but I don't know exactly what's happening (that's why I say "fixed" and not just fixed).

The symptoms:

• The radio only happens when the Clipping POT is touching the enclosure (full assembled pedal or just a light touch of the enclosure with the pot-chassis)(The other pots doesn't interact with the issue)
• The enclosure is aluminum and Grounded by the Jacks. I tried soldering a cable to it to GND but it didn't change the issue (although the cable doesn't stick strongly to the aluminum)
• If I take the 151p cap from IN to GND out, the radio gets quieter.
• If I solder a little cable from the Clipping pot's chassis to a near GND in the PCB (and the 151p cap is present), the Radio disappears 
• The Clipping pot chassis has not continuity to any of the 3 legs.
• With that Clipping-pot-chassis-to-gnd cable in, If I take out the 151p cap from IN to GND, some Radio gets back in again.

So, The 151p cap helps when the Clipping chassis is grounded with the extra cable, but it makes the problem worst if the Clipping chassis is not grounded by the cable but touching the enclosure (which should ground it...)

Whats happening? 

Is my enclosure picking interference and not being able to Ground it completely? so then it gets transferred to the Clipping chassis and (if not grounded by another cable) then transfered by air to the internal parts of the pot? But why is the 151p cap making it worse in that case...?

And how can I be sure in future builds that that kind of stuff is not going to happen? Im left with fear 

[antonis]

A 1nF (or so) cap between input buffer Base - Emitter junction should solve your issue..

[LaloFP]

Hi antonis. Why would that work? I already "fixed" the issue with the mentioned cable. Im looking to understand the scenario

[antonis]

That cap should by-pass Base-Emitter junction, which acts as RF rectifier..

see C3 here: https://www.diystompboxes.com/smfforum/index.php?topic=124166.0
posts #30 & #33..

[PRR]

Whats happening? 

Schematic. Pictures!!!

[LaloFP]

That cap should by-pass Base-Emitter junction, which acts as RF rectifier..

see C3 here: https://www.diystompboxes.com/smfforum/index.php?topic=124166.0
posts #30 & #33..

Great, I now know about RF rectification  haha. But that prevents the rectification of the interference, its a palliative. I'm more interested in understanding the nature of the interference and why it only happens when the clipping pot chassis touches the enclosure, while the enclosure is grounded by the 2 jacks (interior of the enclosure is not painted). And the fact that the 150p cap makes it worse when the pot is not grounded by the added cable but better when it is...

Whats happening? 

Schematic. Pictures!!!

Its already in my friends pedalboard and I forgot to take photos. I was with him and frustrated by the radio...
I made this one (the perfboard) http://tagboardeffects.blogspot.com/2012/07/keeley-4-knob-compressor.html
The cables were short n neat, and didnt change the issue when moved or touched.

[Rob Strand]

It's probably oscillating.

You can try one of these mods, especially the first one.   Ideally the resistor is located at the PCB end - put in series with the cap C7.  Large resistor works best but might affect the ckt that's why I suggest using a small value ( I would have to analyse how much to work out the largest resistor).



If you want to understand the problem a little more you could try the second and thirds mods.  If only the first mod works well then it's probably the buffer oscillating:

You can see how buffers oscillate with capacitive loads in this thread, near the end around post #30.
I've posted about this a few times.  (Yes, it's complicated.)

https://www.diystompboxes.com/smfforum/index.php?topic=119501.20

The capacitive load comes from the pot wiring to ground, and from the pot internals to the pot chassis (when the pot chassis is grounded the capacitance goes to ground).  The capacitor is like a textbook plate capacitor where the chassis is one of the plates.  When you lift the pot chassis from the main chassis that capacitance is reduced.

If the second and third mods work then the cause is due to a feedback oscillator.  Perhaps caused by the wiring inductance and stray capacitances - OTA's have a high bandwidth so they can expose things you can usually ignore in audio.

[LaloFP]

Thank you Rob! Not only to this answer. Thanks to this https://www.diystompboxes.com/smfforum/index.php?topic=49928.0 (great old post! More about Humm than about Radio, but It helped to picture the capacitive thing)

So, If I understand correctly: the ClippingPot resistor band and wipper is one plate, and the chassis is the other. When the chassis touches the enclosure, it becomes a bigger plate, and now catches more Radio, just because it's bigger.

So now there is Radio current in the Chassis plate, and I assume that the resistance (or impedance?) is less in the path _PotChassis=>air=>PotInternals(wipper and resistive track)=>Buffer_ than in the path  _Enclosure=>Jacks=>cables=>PCB=>Buffer_.  The buffer (the "part" that amplifies(and rectifies) the Radio) has its GND closer to the Pot internals(and air between pot-internals and chassis) and its more distant to the relative GND of the enclosure (in terms of resistance?impedance?, not in terms of physical distance centimeters, obviously).

So I conclude that I my GNDs are "poorly" connected. The resistance(impedance?) from Enclosure=>Jacks=>cables=>PCB=>Buffer is greater than the resistance(impedance?) from PotChassis=>air=>PotInternals=>Buffer. And that's why the radio choses to take that route and piss me off   

Is that right? (at least in theory, without having the pedal to test it now)

EDIT: I remember checking the GNDs (Enclosure, jacks, supply-jack-gnd, PCB) and all of them had the same almost Cero ohms in the multimeter. I'm assuming that there is a difference that can't be picked by the multimeter. Maybe because its not resistance and its impedance?

[R.G.]

For higher frequencies, like RF, the resistance of things doesn't matter nearly as much as the inductance and/or capacitance, since both of these have effects on signal that get more intense with rising frequency. Cables will have a "characteristic impedance" which determines how well they transfer high frequencies and how much loss per length they have. The impedance characteristics of a cable are a function mostly of the internal mechanical construction and materials. Audio is so very low in frequency that resistance dominates how it acts. A cable may act as an antenna, an attenuator or >yikes!< a resonant tuner for some RF frequencies.
To keep RF out of your circuits, you need to set the metal shell/enclosure to be a grounded overall shield and introduce RF attenuation into the circuit pathways into the works. One trick that is very useful is to connect a small (0.001uF to 0.01uF) ceramic cap from the signal lug of the input jack to the chassis, a small (100 ohm or so) resistor in series from the input jack to its next connection, and another small ceramic cap back to the shell/RF ground after the resistor. The resistor is important, as it adds damping and attenuation that keeps the ceramic caps from tuning in some random RF frequency by resonance of the caps and the wire inductances.

As an aside, electronics gets seriously complicated with RF and fast, sharp edged logic signals. One textbook on dealing with fast logic signals in circuits was entitled " A Handbook of Black Magic". 8-)

[PRR]

... ... ... ... One textbook on dealing with fast logic signals in circuits was entitled "A Handbook of Black Magic".

https://www.amazon.com/High-Speed-Digital-Design-Handbook/dp/0133957241/
https://www.amazon.com/High-Speed-Signal-Propagation-Advanced/dp/013084408X/
-

[LaloFP]

Awesome, thank you guys.

So, aside the fact that inductance and capacitance in high freqs is something I need to clearly learn more of to understand it (Im looking at it now), we can say (at least as a hypothesis) that in my scenario the problem is that the enclosure is NOT well grounded, or that it is a more "distant" GND compared to the Buffer´s GND, so it doesn't act as a GND (in relation to the buffer) and in fact helps the ClippingPot to become a larger capacitor.

I drill on that because it helps me to detect which "simpler"/more-basic part was done poorly in this build, to learn. I don't really care about the pedal now (Its working without radio). Now I want to learn the root cause (or at least learn another possible root cause  ), not how to treat the symptom. And changing the schematic and adding components that the original pedal doesn't need, feels like obscuring the error being made in the mechanical design.

If that's the case, it will fuel more learning on the grounding and signal return topic.

Thank you X1000000 <3

[Rob Strand]

Please forgive me for my late response.  I feel asleep early last night - a lack of sleep finally caught up with me.

Thank you Rob! Not only to this answer. Thanks to this https://www.diystompboxes.com/smfforum/index.php?topic=49928.0 (great old post! More about Humm than about Radio, but It helped to picture the capacitive thing)

Time has sure gone by.

So, If I understand correctly: the ClippingPot resistor band and wipper is one plate, and the chassis is the other. When the chassis touches the enclosure, it becomes a bigger plate, and now catches more Radio, just because it's bigger.

So now there is Radio current in the Chassis plate, and I assume that the resistance (or impedance?) is less in the path _PotChassis=>air=>PotInternals(wipper and resistive track)=>Buffer_ than in the path  _Enclosure=>Jacks=>cables=>PCB=>Buffer_.  The buffer (the "part" that amplifies(and rectifies) the Radio) has its GND closer to the Pot internals(and air between pot-internals and chassis) and its more distant to the relative GND of the enclosure (in terms of resistance?impedance?, not in terms of physical distance centimeters, obviously).

So I conclude that I my GNDs are "poorly" connected. The resistance(impedance?) from Enclosure=>Jacks=>cables=>PCB=>Buffer is greater than the resistance(impedance?) from PotChassis=>air=>PotInternals=>Buffer. And that's why the radio choses to take that route and piss me off   

Is that right? (at least in theory, without having the pedal to test it now)

EDIT: I remember checking the GNDs (Enclosure, jacks, supply-jack-gnd, PCB) and all of them had the same almost Cero ohms in the multimeter. I'm assuming that there is a difference that can't be picked by the multimeter. Maybe because its not resistance and its impedance?

What you have said here is fine.  It's more correct to use impedance than resistance.  Anywhere there is a capacitance think of that as a lower impedance to RF, somewhere for the RF to hop between capacitor plates.    In general ungrounded metal is a real problem for this type of situation.

However, when I read first your post, this one was a key piece of information.

The radio only happens when the Clipping POT is touching the enclosure (full assembled pedal or just a light touch of the enclosure with the pot-chassis)(The other pots doesn't interact with the issue)

I assumed the ground connection between the circuit ground and the enclosure was fine. Your "EDIT" seems to confirm that's the case.

From that result the RF problem you are seeing isn't the ungrounded metal type of problem.   It seems more complicated.

There capacitance between the innards of the pot and the pot chassis.  Think of that as a capacitor between the pot pins and the pot chassis.  When the pot is mounted on the enclosure the pot chassis is ground  that connect one side of the capacitor to ground.   The pins of the pot are the other terminal of the capacitor.    In the circuit the pot pins connect to the emitter of the transistor buffer.   So the overall effect of that is pot looks like a capacitor is connected from the emitter of the transistor to ground.     That can cause the buffer to oscillate and high freqencies and could cause RF to get into the circuit.

When the pot is removed from the chassis you still have the capacitance between the pot innards and the pot chassis.  However, now it is not connected to ground.   What you have is a second capacitor cause by the outside of the pot chassis being one plate and the enclosure being the other plate.      As you move the pot body away from the enclosure the capacitance of the second capacitor gets smaller.  The key here is this second capacitor appears in series with the first capacitor.  So in this scenario the capacitance seen from the emitter to ground is smaller.   That can help reduce oscillations (The smaller capacitance implies a higher oscillation frequency and the losses and stray inductances prevent it oscillating).

That's how I saw the symptoms playing out.  I could be wrong.  It's more of a starting point.  If it is the case then adding the resistor on C7  should stop the oscillations and the RF.

[Rob Strand]

Here's an old thread which had a similar issue:

https://www.diystompboxes.com/smfforum/index.php?topic=119654

The difference here is the capacitance caused an opamp to oscillate instead of the buffer.   Some cause but different details.

[LaloFP]

Awesome, Rob! I'm so happy that I really understood your whole last post  (Im doing homework to get this right ).

BUT () that doesn't explain why my fix was to jumper the pot chassis to the nearest PCB GND pad. That implies that there is a difference between GND at the enclosure and GND at the PCB (even with 0 ohms between those) and that the problem is not "just" the grounding of the pot-chassis/cap-plate. There seems to be an extra mechanical aspect here.

I asked my friend the pedal to further test. I'll post what I find !!!

[Rob Strand]

BUT () that doesn't explain why my fix was to jumper the pot chassis to the nearest PCB GND pad. That implies that there is a difference between GND at the enclosure and GND at the PCB (even with 0 ohms between those) and that the problem is not "just" the grounding of the pot-chassis/cap-plate. There seems to be an extra mechanical aspect here.

Agreed.   There's actually quite a few things that don't add up.

I guess one possibility is the front mount of the pot is grounded but the rear cap of the pots isn't connecting to the front very well.  Like no connection or a bad connection.    Then the back of the pot becomes more like an unconnected metal problem.

However if your enclosure is metal and connected to circuit grounded then you would thing that would already be doing a reasonable job shielding the RF.    The RF would have to be quite strong to get through that.

The fact the input cap has an effect seems like the RF is actually getting into the input side of the circuit.

When the RF is very high frequency the ground wiring can look like inductances.  What happens the ground disconnects for RF signals making it ineffective.   When you use the short piece of wire of the pot chassis to circuit ground that reduces the inductance and allows the shielding and grounding to function at RF.   In you case perhaps enough to shunt the RF to ground.   (FYI the resistor on C7 can actually help this case.)

When you do RF work having wire pigtails on the ground/shield are bad because it adds inductance into the grounds,
https://www.focusondrives.com/more-about-grounding-and-shielding/


When the grounds are connected with wires, like in a pedal, pretty much all the grounds are long.  Totally the wrong way to do it for RF but usually OK.   The only places where the shields are done without pigtails/wires is where the input and output jack touch the chassis.

A pic of your wiring might help spark some ideas.

[LaloFP]

Awesome Rob! I love these situations, to learn from/for a SPECIFIC practical case.

Next week I'm going to have the pedal and test everything, plus photos. Thanksss!!!

[LaloFP]

Hi guys! I've got good and bad news.

The good: I got the pedal again, took photos, tested your suggestions and collected new info.

The bad: nothing (except one new thing) worked! (remember that I already had the workaround with the jumper from Pot's body to a near GND)

New info 1: All the suggestions/tests did not remove the radio (resistor between C7 and ClippingPot/C8, resistors at both ends of C8, and capacitor between input buffer Base - Emitter junction (tested 1n to 4n7))

New info 2: Another way to remove the radio completely that puzzles me: Adding a connection between the PowerJack's GND cable to any other Jack's (In/OUT) GND. That too removed the Radio interference. Without disconnecting the PowerJack's GND from the PCB, only making both set of cables touch each other (In&Out Jacks GND and PowerJack's GND). It happens if I power the pedal with power supply AND with 9v Battery. Why would that be the case?     

The photo showing that:


New info 3: The pedal only gets radio (without the jumper from Pot's body to GND) in my parents house (where I built it and suspects it lacks earth connection). In my new place (that has a great electrical network) the pedal is death silent. Near my parents home there is a big antenna (I don't know which type, I guess it's for cellphones).

Before taking the photos, I removed the 3PDT and  added new cables, to be sure there was nothing weird in the wiring (it didn't change anything). (I forgot to tell you that I added a Tone pot, replacing 15K (R8) with 7K5 and a A100K Pot)

The rest of the photos: (I LOVE how the perfboards look from behind with all the manual wiring (L)(L)(L)

[Phend]

See you are using 2 mono jacks.
This jack will connect the power supply neg to the jacks ground.
Use as the input jack. Switchcraft 12B



Here is an example of how  it is wired.



Ps, great stuff posted here, I was going to post a question, which has been posted many times, that is, can you use the metal jacks inner sleeve as a ground to the box. I was thinking of eliminating the output jacks ground to the board, because it is connected via the box to the input jacks inner sleeve, gnd.

[LaloFP]

See you are using 2 mono jacks.
This jack will connect the power supply neg to the jacks ground.
Use as the input jack. Switchcraft 12B

Thank you Phend! That could work as a workaround, but I'm super curious about WHY could that phenomena happen.

Its all in the same lines of differential GNDs

[Phend]

Technically I am not qualified as to say why, but the members making comments and advice in the posts above most certainly are. Very qualified. So, your picture shows that when the power jack and mono jacks leads are connected the "problem" goes away. Basically that is what the Switchcraft 12B does. How that eliminates radio interference ? The Effect's God's Mojo...EGM for short...