Reducing Conduction EMC?

[leoren_tm]

emc conduction

I have these universal power supply, and connected to wall socket with 2 wire around 3Inch long. how to reduce the conductive noise on power supply? can i add some ferrite beads on it? or how will i plot the PCB with 2 layer to reduce it?
frequency =.15 to 80 Mhz


just found out that the noise is on 150Khz to 500Khz at around 75dBuV(hope this would help solve these)
thank you

 

[dipnirvana]

electromagnetic compatibility due to conduction

give circuit details for further input from us ,

rest i can say that you have problem due to differential mode noise

 

[leoren_tm]

what means emc noise

i could not provide a circuit diagram, but ill try my best to explain it.
i have wall socket then a wire around 4inch connected to the PCB( i read that wires connected from PCB could be a source of noise), then a .47uF capacitor connected from neutral to live, then pass to a diode rectifier, then some regulation to a transformer, then inductor and then a voltage regulator.

my question would be
1. reducing noise around 10dBuV, is it possible by just shortening the length or wire?
2. Considering poor PCB layout, if it will be improve how much would my noise be reduce?
3. i read about differential noise, with a capacitor in neutral line and live, then a inductor, will the noise be reduce the noise to around 10dBuV? can i ask for some value to be tested?
4. what are possible source of differential noise on a microcontrller circuit?


thank you.

 

[dipnirvana]

emc conduction tests

i will suggest to do some trial-error to satisfy your doubt on length of cable. My guess it shouldn't help much.

The differential noise is coming from SMPS i guess, what is the switching frequency btw. I would suggest 1mH to be put in Line/neutral. You may also try with 100uH. While selecting inductor , please look for it's current rating and SRF value.

 

[leoren_tm]

reduce conduction

im not good on SMPS, so i use scope to determine the freq. its around 190-200khz where also the noise on conduction occur.
Is it possible that the cause of noise is because of the semiconductor that was use? if changing it, is it possible to reduce the noise to 10dBuV?


and also, ill add the inductor(on testing), current consumption of the circuit is around 50mA, so i guess it is safe.

and about testing, may i ask what are the possible cheap component will be use on pre-compliance? on EMC conduction?

ive search some instrument from agilent, would these be ok?
**broken link removed**

Agilent
E7400A Series EMC Analyzers,
Precompliance Systems, and EMI
Measurement Software

thank you


about SRF? what should be use?
"SRF simply describes the frequency at which an inductor quits working as an inductor. For RF designs, an SRF should be chosen with a minimum value that exceeds the operating frequency of one's circuit."
so maybe close to that value?

 

[dipnirvana]

solution for conduction emc

By changing semicoductor , u r changing the whole design. try to search internet you will find lotz of information regarding CE.

Now you should give a try with inductor i said earlier. You may also want to put Y-cap between primary and isolated ground if you don't have a third "earth" wire.

SRF here also important , let's say u want to filter out noise up to 1MHz so u should not select inductor with SRF less than 1MHz.

 

[leoren_tm]

emc conduction solution

i mean changing= same specs with different manufacturer.
anyway..how much does 10dBuV would be? if the noise is around 80dBuV, and reducing it to 10dBuV.:?::?::?::?::?: dont know how to say it
ill try to figure it out.


cause if 10dBuV is not that much, then a change on component would likely to be a solution, than adding something on it.

 

[srinu9009]

y-cap emc

Dear dipnirvana

on your 17th april reply

Let me now how can you say the noise is on 150Khz to 500Khz at around 75dBuV is differential mode noise? why not it is commonmode noise?please kindly explain in brieaf.
let me know,How can we tell this is differenetial mode noise or this is commonmode noise based on test data?

Added after 14 minutes:

Attached application note is good one to understand EMC noise and method to reuduce noise.

 

[dipnirvana]

reducing conduction

That was a simple guess on the basis that generally DM noise dominate below 2Mhz and above it's CM noise which plays the bigger roles. But since i gather that current is just 50mA , so i no more think that it's DM (as it's around 80dBuV).

For starting to analysis u can proceed like this ,

1) check if the noise you are getting is the overtone of switching frequency
2) Use inductor i suggested
3) put 1nF Y-cap across the ground.


After this trail , u can decide on the future course of the action.

Any way u can differentiate DM and CM noise by using phase splitter.

 

[leoren_tm]

noise reduce pcb

overtone? does it mean that i operated the PSU freq in its peak? also if the switching is at 190kHz, can i conclude that the possible noise is because of that? And the solution would be reducing the the speed?
here is the link of similar transistor(cant find the exact one),
https://pdf1.alldatasheet.com/datasheet-pdf/view/236192/STMICROELECTRONICS/ST13003-K.html

 

[dipnirvana]

cm inductor with differential noise

Overtones means harmonics of your fundamental switching freuency . As you have said it's 190kHz , next would be at 380kHz and next 570kHz. Check if the noise you are getting at this frequency.

Changing semiconductor may be solution , but best bet would be to understand where is the problem . Look for the transformer design and layout.

 

[leoren_tm]

reducing emc

with FFT view on scope(dont know if its enough) i saw some spike near that value. and its increasing on every harmonics...

possible general source of problem is the switching element? lowering the freq of switching could solve the problem? any suggestion?

thank you

 

[FvM]

improving smps emc conducted spike

possible general source of problem is the switching element?

Surely true somehow, but it's actually a tautology. The problem is to identify the pathes, that emit the switched voltage and currents, obviously present in any SMPS, to the enviroment.

More precisely, consider a square wave as switcher waveform. It has a lot of harmonics. If you see a broadband series of harmonics, at best of decreasing magnitude over frequency, it's normal operation. Increased magnitudes at some higher frequency bands or peaks indicate resonances and should be interrogated more thorougly. Also some switching elements, e.g. schottky diodes can be expected to generate more high frequency harmonics, that need specific filtering or can be possibly reduced by using a different device. Non harmonic peaks may indicate parasitic oscillations of a switching transistor.

In you circuit, you apparently have nearly no effective EMC filtering, so it's quite normal, that part of the switching energy is spreaded all over.

A moderate reduction of switching frequency won't help much, I guess. A large reduction requires redesign of the magnetics, possibly causing additional interference problems. Please consider also, that in principle, higher frequency interferences can be filtered more easily.