Removing Noise From Power Line

This is the gerber output

Looks better than a feared.

The most problematic point is probably the power supply crossing the isolation boundary. It should have massive common mode filtering and overvoltage suppression. You may want to make tests where the connection is cut and replaced by an external supply.

In addition all external cables connected to the upper circuit part should be checked. Ferrite rings as common mode filter should be checked for each cable. Additional filtering options with respect to the interface function.

Make a test software where all external cables except for the powersupply can be disconnected and connected but correct processor operation can be checked, e.g. by a LED.

PLC is Programmable Logic Controller?

i suggest:
1 Y-cap C3 5n-200n
2 C1,C2,C5 capacitors NP0 (because of better freq char) 20n-100n connect directly to mc pins
3 (i have doubt about it) C4 - Y cap 5-200n
try it (1,2,3) separately and together

Velkarn said:

PLC is Programmable Logic Controller?

i suggest:
1 Y-cap C3 5n-200n
2 C1,C2,C5 capacitors NP0 (because of better freq char) 20n-100n connect directly to mc pins
3 (i have doubt about it) C4 - Y cap 5-200n
try it (1,2,3) separately and together

Click to expand...

Thak you velkarn, i will try them.

Before trying i have some questions:
1_What is the purpose of c4.
2_I have a frequency generator(Protek G5100), and it has VCF IN input, what is the purpose of that input. Can i modulate the frequency which is generated by this function generator, over 10V DC with this input =).

1 C4 together with C3 make low impedance way for noise. or shunts noise voltage. a kind of this.
2 i do not know can or can not

i wrote you to measure a lot of points. it need to understand with what kind of noise you have a deal: differential or common-mode. if at output of you EMI filter there is no noise, but at power of mcu - there is, that means that it is common-mode noise (i guess .
if you problem is common-mode noise you need to use Y capacitors.
if you problem is differential noise you need to increase values of you EMI filter and C1,C2,C5 caps, improve ground tracing of pcb

I will do everthing on monday and post results. I think my problem is differential noise. Because while problem occurs, every power line around card there is a noise. I really cant understand how can that noise passes all that filters and reset my MCU. Is it possible?

Do you have any idea of active filters? Should i try them? Texas instruments have some active low pass filters which are uses OP-Amps.

Here is a pogram which helps you to design a low pass filter.

https://webench.ti.com/webench5/power/webench5.cgi?app=filterarchitect&filterType=Lowpass

What is this active filter issue? Is it bad or good.

Interferences that reset a microcontroller will also exceed the linear voltage range of an active filter. I don't see how active filters can solve your problems.

Regarding common mode or differntial noise, it's difficult to determine likely interference pathes without a complete wiring diagram showing the external board connections. From similar cases, I guess that common mode is more plausible. But common mode noise will be converted into differential nosie in combination with asymmetrical signal termination, either on the board side or externally. And vice versa.

It is hard to believe but cables whichs shields are not connected anywhere, works as antenna When shorten the length of the cable reset disseapeard.

The questions are:
1_Even if cable bring noise how can it penetrate over optocouplers?
2_As you know i dont have earth ground. Where should i connect cables shield?
3_I am thinking about make a faraday cage to my card, where should i connect this cage?

Thank you =)

There are other interference pathes in your circuit than through the optocouplers.

rhnrgn said:

I think my problem is differential noise. Because while problem occurs, every power line around card there is a noise. I really cant understand how can that noise passes all that filters and reset my MCU. Is it possible?

Click to expand...

what you mean "every power line around card"? every Vcc? or Vcc and GND lines?
i think you problem is common mode noise. this kind of noise go thru parasitic capacinances. thats why it possible. to simplify understanding, imagine (or draw) line with changing with high dv/dt potential (created by valve) and all parts of you system, having parasitic capacitance to that line. all this capacitors forms loops. and interference currents flow thru that loops.
active filter just increase suppression ratio. you measured noise after EMI filter. now, using suppression ratio of that filter (from datasheet), calculate value of noise, which have to be at the input of filter. than ask yourself "is it really so big"?

- - - Updated - - -

FvM said:

Interferences that reset a microcontroller will also exceed the linear voltage range of an active filter. I don't see how active filters can solve your problems.

Regarding common mode or differntial noise, it's difficult to determine likely interference pathes without a complete wiring diagram showing the external board connections. From similar cases, I guess that common mode is more plausible. But common mode noise will be converted into differential nosie in combination with asymmetrical signal termination, either on the board side or externally. And vice versa.

Click to expand...

I agree about active filters.
Yes, a complete wiring diagram showing the external board connections, including all metal cases and connected or not connected shields is very useful.

Velkarn said:

what you mean "every power line around card"? every Vcc? or Vcc and GND lines?
i think you problem is common mode noise. this kind of noise go thru parasitic capacinances. thats why it possible. to simplify understanding, imagine (or draw) line with changing with high dv/dt potential (created by valve) and all parts of you system, having parasitic capacitance to that line. all this capacitors forms loops. and interference currents flow thru that loops.
active filter just increase suppression ratio. you measured noise after EMI filter. now, using suppression ratio of that filter (from datasheet), calculate value of noise, which have to be at the input of filter. than ask yourself "is it really so big"?

- - - Updated - - -


I agree about active filters.
Yes, a complete wiring diagram showing the external board connections, including all metal cases and connected or not connected shields is very useful.

Click to expand...

All situation is changed, when shorten the length of the parallel port cable reset disseapeard. When i activate solneoid valve next to card reset occurs, when tkae it away there is no reset. It is something about radiated noise :/ Parallel port cable is shielded but shield is not connected anywhere. I think it is working as an antenna and sum whole noise at air to board. Now i am trying to find how can noise passes through optocoupler.

There is an enterance pcb card between parallel port and main electronik pcb card, i will make a drowing tomorrow and show to you the enterance part of the parallel part to board.
I think i am focused and my investigation area is getting smaller.

As previously discussed, several terminals of the "parallel port" connector are bypassing the optocouplers.

Hello friends,

I found what is the problem. I am using external interrupt 3 in order to watch motor drive signal input. If there is a noise at this input problem occurs.

When i disabled interrupt resetting dissapears. There is a situation like because of noise , interrupt triggered many times over and over and that is why PIC become crazy and reset itself. It is like, enabling interrupt decreases noise immunity of that pin. :/

Now, my system works ok. My pic is just checking the state of that pin but while checking the state of that pin there is still noise that is why i am taking wrong values. I am trying to write a software filter like listen port 100 times and "if 80 times high 20 times low than it is high".

But it is not enough, because of the frequency of noise if very high this code cannot consider if it is high or not, i mean when i changed values to "if 60 times high 40 times low than it is high" it can consider better =). probably frequency of noise is that high half of them high half of them low. Because what i am doing is sampling.

Can i explain my situation? Do you have any ideas or sugesstions?

It's been said. https://www.edaboard.com/threads/311568/#post1332886

A trivial explanation would be that the reset is caused by your software design.

Click to expand...

Anyway, good you found it.

I have been exposed to such situation. it was solved through using a separate mains transformer to feed only my circuit and the transformer must have short wires to your power supply

i'm are new in mcu. i have experien in same case . my cicuit use PIC mcu drive magnetic coil. after drive magnetic mcu will reset.
i try use optocouple drive transistor power my circuit can work good in my idear. you try use this for solve.
i hope this you help.