Ground Isolation for Daisy-Chains

[Baran Ismen]

Hi. I wonder how these below extensions work. Anyone used these (or similar) before?

These are supposedly to eliminate high pitch whines of digital pedals when interconnected with other analogues. For example, I got 2 in my board, M-Wave's Universe Reverb pedal, and Mooer X2 Preamp, these both need isolated powering for some reason. Whatever I connect as an extra, these two start to whine faintly like a baby in a high frequency.

Godlyke ground hound
JOYO JP-06
Virtual Battery, now this is a bit different & interesting
Quiet Box


My question is, can we somehow derive what's going on inside, its a small DC-DC buck converter, or a ground lift thing, or what would it be? I think, as pedal makers, we can make such small circuits to fix minor problems like mine, eh?  I honestly don't want to spend a fortune on an ISO Brick just for 2 among other 16 pedals. There must be a way to solve this "crosstalk"

[ElectricDruid]

I think those two things might be different things. The Godlyke one descibes itself as a "ground isolator", whereas the other one sounds more like a filter. Both of them claim to "break the ground loop" though, and I'm not sure I see how that's done, if true.

[antonis]

I think we've made a mixed-up ragbag of ground isolation, hum elimination and high pitch oscillations..
(tempting for a salad demonstration but for an universal power supply..?? dunno..)

[Baran Ismen]

I think we've made a mixed-up ragbag of ground isolation, hum elimination and high pitch oscillations..
(tempting for a salad demonstration but for an universal power supply..?? dunno..)

Might be Antonis    I'm not entirely sure what the problem I've is called; probably digital whining due to crosstalk and/or processor clock speed, or some sort of.. But as the terms are commonly used in all products, it might sound misguiding as you said. Main problem is high-pitched whine.

can be heard here also

[FiveseveN]

Start with an LC "pi filter" and go from there.

[R.G.]

The problem you're trying to solve is shared ground voltage noise caused by (1) the resistance of the shared ground wires and (2) high pulsed ground currents caused by switching-type power usage in some pedals. Digital pedals are sometimes power hogs, especially the earliest ones, and they don't play well with other pedals sharing a power supply. This is because they pull sudden pulses of current from the shared power lines and the resistance of the shared wires does the rest.

You can attack this two ways. One is to use a second (or third, or fourth) power supply for just the offending pedals that don't play nice with other pedals. The second is to buy an isolator/filter/whatever to go on a daisy chain to try to fix the problem.
Filters may not help. They have to make assumptions about what the noise is, so a one-size-fits-all solution is tough to make

Isolator supplies are also tricky. In concept, you just make a baby switching power supply that uses a tiny transformer to isolate, then rectify/filter/etc. the output. I've designed and prototyped several of these.

The problem with this approach is that you're isolating a doesn't-play-well-with-others switching power supply in a pedal from the analog stuff by adding a new switching power supply in series with the pedal. A quit look at the link suggests they act this way. Now you have transferred the problem of switching power supply input ground currents from the pedal to the new isolator. Maybe it can be quieter than the pedal. It takes some good tech work to design a quiet-for-pedals switching power supply, as we all know.

The costs on the isolators seem to be running nearly the cost of a well-known, good-reputation, quiet single 9V power supply.  It might make more sense to put another whole 9V power supply on only the bad-actor pedals. No ground noise there, and no isolator noise either.

[Baran Ismen]

I've just noticed that this whining fades away in time, or at least lessens quite much after keeping the pedal active for about 5-10 minutes. It's obvious on a cold start, then slowly fades but not entirely gone.

[Baran Ismen]

I gave this try yesterday upon a suggestion on YT;

I've made a patch cable without the GND connection, just a single lead for signal, and connected 2 pedals (one of being digital, other analog), and I saw that they both be working, their powers were daisy-chained, yet that high-pitch digital noise was still there..

Usually I merge all grounds on the board on the pedals I make, which I believe it's the correct way, but is it really right? Like, in such situations, I think it causes ground loop(s) ?

[FiveseveN]

Repeat after me: not every power issue is a ground loop.
Start with a filter. Pretty please?

[Baran Ismen]

Repeat after me: not every power issue is a ground loop.
Start with a filter. Pretty please?

Okay, then which frequency shall I be targeting to apply a low-pass?

https://3roam.com/pi-filter-design-calculator/

Also I know little about the terminology of the inductors, there are resistor and cap-type of them, and some has an Amperage indicator on them, what type shall I get?

[antonis]

Okay, then which frequency shall I be targeting to apply a low-pass?

The lower the better..

There is mains frequency (50-60 Hz) and FW rectified one (double the mains)
Better aim for an octave lower than mains..
(i.e. 25Hz for 50Hz mains frequency)

[FiveseveN]

The lower the cutoff freq, the more attenuation you'll have at the offending "whine" freq. You have a wide margin where it can be effective without starting to be a nuisance (large size, big inrush current that can trip up the PSU).
You'll want an inductor that looks like this:



The value is not important. Salvage one from anything (dead) with a switchmode PSU, along with some big electrolytics. In a pinch, use a ~10R resistor instead or make your own by wrapping ~10 turns of wire around some ferrite or other ferrous core.
Put the filter inline with the power input to one of the digital pedals (just the one first) and see if it helps. You might need one for each or you might not, depending on the specific source of the noise.

[Rob Strand]

Okay, then which frequency shall I be targeting to apply a low-pass?

https://3roam.com/pi-filter-design-calculator/

Also I know little about the terminology of the inductors, there are resistor and cap-type of them, and some has an Amperage indicator on them, what type shall I get?

At the end of the day try 100uF or 220uF caps and about 330uH inductance.

The inductor current rating needs to greater than the sum of the digital pedal
currents. If you don't have a spec for the current rating there is a risk the filter won't
work because the inductance value could drop to below 50% of its value due to
over current.  In this case it's likely the DC resistance will be too high and
cause voltage drops and the the inductor may overheat.

The filter powers only the digital effects.

You might get away with one filter.  However if you have a problematic
pedal you might need to power it from its own filter.

That might not be enough.  Next would be to add a separate filter for the
analog pedals.   The same filter will work but for low power analog
you can use RC filters maybe 1 to 10 ohm ; need to check voltage drop.

You don't want to daisy chain the power wiring.  Wire the filter inputs
back to the PSU output.  Then wire each pedal to the filter output.

If the filters still don't work well enough you might need to add a resistor
(say 1 ohm), or an inductor, between the between the PSU ground and the analog
effects ground.  Place between the PSU ground and the analog filter cap.

There's so many way noise can get in.   It's not easy for a hobbyist to understand
enough to know what do try next if a test (ie. filter) doesn't work.

[Baran Ismen]

Okay, so the rating of the inductor should be bigger than the need of the pedal, is that correct? Say the inductor is rated for 1A and the pedal requires 200ma, it should be good to go, right?

A pair of 470uf and 1mH inductors gives a cutoff frequency of about 465hz, It is possible to decrease this more as Antonis suggested below to the mains frequency, but it seems a bit hard to find bigger inductors than 1mH, and also I'll need 10.000uf'S of cap's for around 25hz of cut. Isn't it a bit of overdoing?

Made this small schematic, how's it?


Same filter (not the same, its RC actually) can be done with a 1k + 10uf with a low pass filter begins from 15hz. But I think inductor plays a role there, right?

What about RLC filter?

[FiveseveN]

How are you going to pull 300mA through a 100R resistor @ 9V?

[Baran Ismen]

How are you going to pull 300mA through a 100R resistor @ 9V?

You're right, its left from another project, it should be around 22ohms for a 0.4A output, is that right? But it seems it needs to be rated for 3W ?

[Rob Strand]

Okay, so the rating of the inductor should be bigger than the need of the pedal, is that correct? Say the inductor is rated for 1A and the pedal requires 200ma, it should be good to go, right?

Yes it should work.  However if you spec the inductor current too high you will have either a big inductor, or it could limit the inductors you can find.

A pair of 470uf and 1mH inductors gives a cutoff frequency of about 465hz, It is possible to decrease this more as Antonis suggested below to the mains frequency, but it seems a bit hard to find bigger inductors than 1mH, and also I'll need 10.000uf'S of cap's for around 25hz of cut. Isn't it a bit of overdoing?

First thing, is making the inductors high in value will mean a lower current rating for a given size inductor. 

Second,  filtering 50Hz  with inductors is solving the wrong problem.  You want a high-frequency filter to get rid of the buzz.  All it does is make the inductors and caps silly values.  If you need to filter 50Hz you would be best with other solutions like regulators or capacitance multipliers.  For your current problem 50Hz isn't an issue.

Made this small schematic, how's it?


Same filter (not the same, its RC actually) can be done with a 1k + 10uf with a low pass filter begins from 15hz. But I think inductor plays a role there, right?

What about RLC filter

As FiveSeveN said resistors will cause voltage drops.

It's easy to get caught up on filtering.  The way you are viewing the filter is that it is removing noise from the input supply by acting as a lowpass filter.  It is highly unlikely input supply will be the source of noise.    What is more likely is there are pulsating currents at the power input of the digital effect.  Those current pulses are causing voltages fluctuations on the supply rail and onto the ground wires.

As shown the filter will help this scenario but the how it does that is much more complicated than the low-pass filter view.  What you really want is for C2 to be large enough to absorb the majority of the current pulses.  Then L1 and C1 will prevent what's left getting back onto the input DC supply and getting into the analog pedal.  It also helps stop the current pulses getting on the audio grounds via the audio cables (6.5mm jack + coax).

For this experiment the filter only goes to the digital pedals or the problematic digital pedals.  The analog pedals are powered from the input supply.  if you power them from DC supply at C2 then L1 and C1 aren't doing an awful lot to help stop the noise getting into the analog pedals.

If you need more filtering then you either need separate filters to the digital pedals and/or filters to the analog pedals.

All that might be hard to grasp because there's a lot of things going on and a lot of finer points which are hard so see/understand.

There definitely won't be a wide reaching magic pill solution you will have to play around a bit to narrow down what the actual cause is then plug it up.  Even I cannot work that out sitting here at my computer.  You will need try a few things and try to converge to a solution.

[Baran Ismen]

Okay, so the rating of the inductor should be bigger than the need of the pedal, is that correct? Say the inductor is rated for 1A and the pedal requires 200ma, it should be good to go, right?

Yes it should work.  However if you spec the inductor current too high you will have either a big inductor, or it could limit the inductors you can find.

A pair of 470uf and 1mH inductors gives a cutoff frequency of about 465hz, It is possible to decrease this more as Antonis suggested below to the mains frequency, but it seems a bit hard to find bigger inductors than 1mH, and also I'll need 10.000uf'S of cap's for around 25hz of cut. Isn't it a bit of overdoing?

First thing, is making the inductors high in value will mean a lower current rating for a given size inductor. 

Second,  filtering 50Hz  with inductors is solving the wrong problem.  You want a high-frequency filter to get rid of the buzz.  All it does is make the inductors and caps silly values.  If you need to filter 50Hz you would be best with other solutions like regulators or capacitance multipliers.  For you current problem 50Hz isn't an issue.

Made this small schematic, how's it?


Same filter (not the same, its RC actually) can be done with a 1k + 10uf with a low pass filter begins from 15hz. But I think inductor plays a role there, right?

What about RLC filter

As FiveSeveN said resistors will cause voltage drops.

It's easy to get caught up on filtering.  The way you are viewing the filter is that it is removing noise from the input supply by acting as a lowpass filter.  It is highly unlikely input supply will be the source of noise.    What is more likely is there are pulsating current at the input of the digital effect.  Those current pulse are causing voltages to be induced into the supply rail and onto the ground wires.

As shown the filter will help this scenario but the how it does that is much more complicated than the low-pass filter view.  What you really want is for the C2 to be large enough to absorb the majority of the current pulses.  Then L1 and C1 will prevent what's left getting back onto the input DC supply and getting into the analog pedal.  It also helps stop the current pulses getting on the audio grounds via the audio cables (6.5mm jack + coax).

For this experiment the filter only goes to the digital pedals or the problematic digital pedals.  The analog pedals are powered from the input supply.  if you power them from DC supply at C2 then L1 and C1 aren't doing an awful lot to help stop the noise getting into the analog pedals.

If you need more filtering then you either need separate filters to the digital pedal and/or filters to the analog pedals.

All that might be hard to grasp because there's a lot of things going on and a lot of finer points which are hard so see/understand.

There definitely won't be a wide reaching magic pill solution you will have to play around a bit to narrow down what is the actual cause then plug it up.  Even I cannot work out that is sitting here at my computer.  You will need try a few thing and try to converge to a solution.

Thanks for the detailed explanations Rob, I'll get some different types of inductors and make some experiments over the same layout with various caps as well. The whining noise I hear is a high-pitched one, probably around 10khz, so theoretically, if I apply a low pass filter that would go below that frequency, say 5khz, it "should" be good to go, right?

[amptramp]

Beware!

A power converter with negative feedback that establishes the output voltage is actually a negative resistance element.  Look at it this way: suppose you have a 9 VDC supply rated at 2 amps powered from a 120 VAC supply.  If you draw 2 amps from the 9VDC supply, the AC input current for a system with no losses would be 0.15 A or 150 mA.  If the input voltage goes down to 110 VAC, the system still has to produce the 18 watts of output (9V x 2A) but with the lower input voltage, the input AC current increases to 0.1636363 A.  In other words, as the voltage falls, the current increases due to the negative feedback, which turns the power converter into a negative resistance device.  The product of voltage and current input is constant and the V-I curve is a hyperbola with the slope of the operating point giving the value of the negative resistance.

In this case, the 120 VAC supply is considered to be solid but the voltage seen at the input to the power converter varies due to the impedance of the filter.  If the magnitude of the positive impedance of the filter when placed in parallel with the negative impedance of the converter still ends up negative, you have an oscillator.  If it is close to negative, you have a power converter with bad transient response.

If you have a filter which is one inductor in series with the supply and a capacitor at the input to the converter, the filter impedance at the resonant frequency is SQRT (L/C), so it is apparent that you want high C and low L to get the filtering you need with a low impedance.  In many cases, parasitic resistances and operation away from the LC resonance will permit the use of a filter that would be unsuitable with ideal components, but would also degrade the suppression of interference.

Your best bet would be to simulate the response to a change in output load, say going from 2 amps to 1 amp and check the voltage at the power converter input.  It should be a damped oscillation but in pathological cases of too much inductance and not enough capacitance, you may get an oscillation that increases over time.  That is unusable.

[R.G.]

If you don't like the amount of filtering in the circuit shown, try adding another inductor in the ground lead between C1 and C2. This inductor works on smoothing the ground current. Pedal circuits often have zero rejection of both power supply noise and ground noise. You may have to smooth both of them.

This has to be done with some care. The DC resistance of the inductor will cause a DC offset so the resistance time the pedal current has to be low. Things get funny when you add impedances to "ground", so tinker carefully.

I still have the opinion that adding a power supply just for the noisy offending pedal may be the simplest and easiest solution.