Well, you can always hit the single Like.

There you go... it'll have to suffice with two likes on two different posts...

 

Now, let's try to put some numbers to this discussion.

My designs are normally powered by an ordinary 12VDC @ 2 amp wall wart which in turn is powered by a domestic 110VAC line. The wall wart powers a few inductive sensors that draw a maximum of 0.7 amps between them. And then is connected to digital circuitry through a diode and capacitor filter like the one shown in post #1:

​

Here's a picture of the "naked" wall wart. I like to disassemble them and then incorporate them into my pcb's, because things look more elegant that way ...

​

This filter's output is then connected to a 1 amp SMPS with a 5V output. The complete 5V circuitry consumes in average about 0.6 amp. And that's hard data because I've actually measured it.

Among other things, the digital part of the circuit is in charge of applying PWM (through opto isolators) to a couple of 0.1 HP 90VDC motors, which normally work at 50% of their capacity and are switched at a 14 kHz. Problem is that the motors and the wall wart are both connected to the same power supply, and spurious MCU resets were happening way too often before I incorporated the previous filter into the circuitry. It is worth noting that the electronic circuitry and the motors are both completely isolated from each other. They don't even share a common ground. Although the motors' armatures and the PCB's ground are both connected to the same earthing electrode.

For the large cap, I've been using two 4,700 uF electrolytic caps. I know that it sounds like overkill, but the circuit has been working fine since then, and I don't want to know what would happen if I were to use only one cap. Every single chip in the PCB has a 0.1 uF bypass cap between its Vcc and Gnd pins.

Question, considering the previous information, is there a "rule of thumb", or a ball-park figure, that could be used to estimate the value of the inductors to be added to the aforementioned filter to further improve it?

 

Question, considering the previous information, is there a "rule of thumb", or a ball-park figure, that could be used to estimate the value of the inductors to be added to the aforementioned filter to further improve it?

I'm not aware of any worthwhile rule of thumb for this stuff; there are simply too many factors involved, starting with the nature of the interfering noise source together with the susceptibility of the subsystem you're trying to protect. Knowing the amplitude and spectral characteristics of the interfering noise helps. Also helpful is knowing the exact mechanism responsible for the problems experienced by the affected subsystem (noise coupled into the uC's reset input? spikes on the Vdd line? ground bounce? etc?).

If there's no choice but to operate completely blind regarding the above, I'd say start with 10 μH and work upward from that until the problem goes away. Take care to use inductors that won't saturate given the DC current that'll be flowing through them. Consider ferrite beads for very high-frequency noise. Consider that the problem may be exacerbated, or even caused, by inappropriate ground topology: heavy load currents, especially switched currents with high dI/dT, absolutely MUST not be allowed to flow through the ground paths of sensitive analog or digital circuits. A single-point, "star" ground arrangement may be necessary.

That's about as good as I can do early on a Sunday morning, fortified with only a single cup of coffee...

EDIT: after second cup of coffee, I note that the above comment on ground topology applies equally to power connections for Vcc, Vdd, and so forth; provide separate paths for supply currents (especially switched or PWM'd) for loads, uC, analog, etc.

 

Some types of noise can be minimized by using a common-mode choke at the input and/or output.
Those connect to both the power and common, and reject noise that is common to both lines.

 

Some types of noise can be minimized by using a common-mode choke at the input and/or output.
Those connect to both the power and common, and reject noise that is common to both lines.

I believe said choke is shown in the upper part of the wall wart's circuit, in the picture on my previous post? ... so maybe I should add another one to my pcb?

 

I have several inductors like this one, and this one laying around. Would they be a good starting point?

The first one is 10 uH @ 330 mA
The second one is 100 uH @ 1.1 A

Would it be convenient to install both the power and the ground inductors side by side? (I've seen how they "couple" when at close distances) Or would it be counterproductive?

 

I have several inductors like this one, and this one laying around. Would they be a good starting point?

The first one is 10 uH @ 330 mA
The second one is 100 uH @ 1.1 A

Try them, and see what kind of difference they make.

Would it be convenient to install both the power and the ground inductors side by side? (I've seen how they "couple" when at close distances) Or would it be counterproductive?

How did you determine that you need both power and ground inductors? I question the need for both-- is this based on experience, or measurements, or... what?

As for placement, side by side or otherwise, I don't know.

 

How did you determine that you need both power and ground inductors? I question the need for both-- is this based on experience, or measurements, or... what?

I didn't determine that ... I'm just following Scott's recommendation, as shown in figure 2 of my very first post.

 

I only can say that the noise interfering and spike, how to deal with them is big issue, should be handle it case by case, there is no one circuit to handle all the problems.

As you can see the diagram below, actually there is one condition that I didn't drew it into the diagram, and that is the transformerless power supply, because our forum doesn't allowed to discuss it, so I ignored it, but I have to mention it here, and I don't want to discuss the theory or circuit, the circuit that you uploaded in #1, it just one part of the whole diagram, the problem that I met was from one of my friends, his had an spike interfering problem from the city power AC220V, and the problem was happened in the factory, so it is more complicated.

If you use the transformerless power supply to replace "110V/220V AC/DC 5V Adaptor", the third line (device) of the diagram below, and use the circuit that you uploaded to replace the "LC filter", the problem was that when the other machines (heavy power) turn on then it will occurred a spike to interfering to the device that designed by my friend, so he asked me to help him when he was in the factory (we communicated through the LINE App), I given the circuit with only one Toroidal coil (only the positive side) to try, but the interfering still existed, so I asked him to used the O'scope to measured the spike of two output lines of the transformerless power supply, and he found out that they all have had the spikes, so I decided to used two Toroidal coil, the inductor that you linked in #26 can't be used in that device, because the inductor can't provides too much current for it, the whole situation was more complicated, so I only make it shorter to understand it more easier.

 

Among other things, the digital part of the circuit is in charge of applying PWM (through opto isolators) to a couple of 0.1 HP 90VDC motors, which normally work at 50% of their capacity and are switched at a 14 kHz.

"It's the law! FCC regulations prohibit operators of any device generating radio noise from causing harmful interference. It's always better to correct problems as they occur rather than wait for a formal complaint through the FCC." Power Line Noise - ARRL.
In your case source of noise definitely is PWM circuit.
You can easily monitor this noise with your USB oscilloscope, connecting X10 probe in series with capacitor 10...20 pF to hot wire of mains.
Ground clip should be not connected to anything.
Using in PWM circuit (only!) power line filter like this and capacitor 470...560uF, 300...400V after diode bridge, will help.

 

As noise filters rather than supply filters, capacitors provide a shunt path to the opposite side of the power circuit, but they are only part of a voltage divider, the other part being the effective source impedance of the noise generation device. So there will usually be some noise left. Adding the series inductors puts their AC impedance in series with the noise generator impedance, and that works along with the capacitors to reduce the noise a lot more. That is the simple explanation of why series inductors are so useful 9in reducing electrical noise.

 

Well, I installed the filter on one of the power supplies used for energizing the controller circuits for a small coil winder machine I have. And all of the spurious resets and communication glitches that it had simply disappeared ... it's been working beautifully for days now without a hitch. The components I used were a PMEG4050EP schottky diode, a 20 uH, 3A inductor, and a 4,700 uF @ 16V electrolytic cap.

Problem solved

 

Well, I installed the filter on one of the power supplies used for energizing the controller circuits for a small coil winder machine I have. And all of the spurious resets and communication glitches that it had simply disappeared ... it's been working beautifully for days now without a hitch. The components I used were a TMS320F28069 schottky diode, a 20 uH, 3A inductor, and a 4,700 uF @ 16V electrolytic cap.

Problem solved

Good work! And now you know a bit more about noise filters, which are different from power filters.

 

Just simulation:

 

Just simulation:
View attachment 184942

Thanks, Danko. In this case, the circuit could afford the voltage drop caused by the diode, which turned out to be about 0.35V. But in the future, I plan to use a pFet-based circuit designed by crutschow that significantly reduces said drop.

 

Thanks, Danko.

You are welcome. See diagram below.
In real filters Rser of L1 used as R2.

But in the future, I plan to use a pFet-based circuit designed by crutschow that significantly reduces said drop.

Diode always is in conductive state, because of R1 current, so super diode will work as piece of wire with zero resistance and zero effect.

 

You are welcome. See diagram below.
In real filters Rser of L1 used as R2.

Diode always is in conductive state, because of R1 current, so super diode will work as piece of wire with zero resistance and zero effect.
View attachment 184949

That's interesting ... L1 and C1 are there for obvious reasons. But the diode is (supposedly) there in case the power supply falters for a moment, so that C1 feeds the load (in your diagram, shown as R1) and does not discharge itself back into the power supply.

 

That's interesting ... L1 and C1 are there for obvious reasons. But the diode is (supposedly) there in case the power supply falters for a moment, so that C1 feeds the load (in your diagram, shown as R1) and does not discharge itself back into the power supply.

I am thinking that adding a series diode mostly helps in simulated systems rather than real world ones, or in systems that are able to sink as well as source current. Most real power supplies do not go to a low impedance when they fail to develop voltage.

 

I am thinking that adding a series diode mostly helps in simulated systems rather than real world ones, or in systems that are able to sink as well as source current. Most real power supplies do not go to a low impedance when they fail to develop voltage.

A few years ago I installed an electronic device in a semi truck. The device would lose power and reset itself whenever the engine was started. The start motor was drawing so much current from the battery that its voltage went way, way down. The problem was solved with a diode in series, and a big fat cap.

 

A few years ago I installed an electronic device in a semi truck. The device would lose power and reset itself whenever the engine was started. The start motor was drawing so much current from the battery that its voltage went way, way down. The problem was solved with a diode in series, and a big fat cap.

OK, in the truck case it was a very low impedance power source that could sink as well as source current. Most supplies are not intended to do that. A direct battery connection is the one common exception.