Buck converter switching noise

[pree]

Hi I am facing an issue where the buck converter switching frequency is being coupled to the input of the buck.
It seems an issue with the layout, however the buck converter datasheet doesnot provide any layout guildelines and it is a chinese supplier.
So would be great if someone can look at it and help me.
I attached the snapshot of the layout and buck converter schematics.

[tszaboo]

Your layout and schematic looks fine. How much current is being drawn from this supply?
Maybe take a look at the ferrite bead, sometimes it makes things worse. Also if it's EMC testing it fails, then sometimes common mode filters and such is unavoidable.

[pree]

not more than 500mA. i tried adding an inductor at the input 2.2uh nd changing a 1uf to 4.7uf then the switching noise goes away.
but i want to know the buck switching is coupling at the input and

[electronx]

Can you post a photo of the switching noise?I am always afraid of resonance where LC is visible. You can try to make a jumper with 0 ohm resistor instead of ferrite bead. What is the switching frequency? Did you calculate the output resonance frequency? and ı want to know  that inductor self resonant frequency
I think the output characteristic resonance frequency is 9.93 kHz, in this case I guess your switching frequency is at least 100 kHz above it.

Did you only place it locally on the PCB?

[tszaboo]

not more than 500mA. i tried adding an inductor at the input 2.2uh nd changing a 1uf to 4.7uf then the switching noise goes away.
but i want to know the buck switching is coupling at the input and

You have to be careful with probing DC-DC converters. Noise measurements are tricky, because the the 1/10MOhm scope input can pick up a lot of noise next to an inductor, which is not conducted. Looking at the noise, you should differentiate ripple and noise pickup. Noise pickup often times several times larger than the actual ripple.
You can reduce that somewhat with proper probing. As small loop area for the probing as possible, far from the inductor.

[jwet]

Your layout isn't bad but could be improved.  The main object is to reduce loop areas.  By turing the large cap at the bottom of the layout 90 degrees, you could significantly reduce the loop area of that path.  In a buck, there are very fast high currents in that path with the filter cap.  That loop inductance/area , the ESR of the CAP make for trouble.  You didn't mention frequency but you can't get away with much in the modern >500 Khz switchers.  Also check the HF specs of your caps.  Organic Semi electrolytics (OSCONS) in parallel with a good sized ceramic, 10 uF make a nice very low ESR buck filter. 

I'm also a bit dubious about a ferrite bead in that position.  Ferrites create real resistance (rather than complex impedance) which can make a mess.  You have a lot of bypass but a ferrite shouldn't be in so close.

Good Luck

[pree]

i measured at the input of the buck, before the ferrite bead. in the schematics it is a ferrite but on the PCB it is 0ohm. so i measured it before that. the swtiching noise is 480KHz which the swtiching frequency of the buck or what it is set to usign the Rt resistor.

[pree]

the ferrite bead in the schematics is just a symbol but on the pCB it is 0ohm or short trace as of now. I measured the noise at the input before the ferrite bead/0hm resistor. and the noise is 480KHz which is the swtiching frequency of the buck.
however i put a 2.2uF inductor instead of the bead and the increase the 1uF capacitor to 4.7uF then the switching noise is gone

[electronx]

I did a few simulation studies. First of all, this definitely does not give correct results .It does not include parasitic values ​​but can give an idea.  If s parameters are assigned to capacitors and inductors in HFSs and examined, it will include better pcb traces itself. However, the general idea is that you lowered the cut-off frequency. How did you measure that the noise frequency is equal to the switching frequency? FFT analysis or oscilloscope?Is your noise on the sharp edges of the sawtooth ripple? What is its amplitude?
You can pull back the cut-off frequency by increasing the direct capacitor values. However, as the capacitance increases, the capacitor will grow. You added an inductor and pulled back the cut-off frequency again. This is good. But be careful about the resonance points, If you get harmonics at the resonance frequency, this will cause the amplitude of the harmonic noise to increase.



----------------------------------------------------------------------------------

[pree]

the noise was measured with Scope and I used the smallest possible ground path with a pogo pin ground of the scope.
Attached is the noise image from scope.
the buck gets input from the SMPS which is 18V and I isolated the buck and measured directly at the AC-DC output and that 18V is clean.
In the design the resistor which sets the switching frequency of the buck is set to 100k ohm and that makes the buck at 480KHz. When i tried different resistor values, then the noise pattern changes.
there is nothing else in the system switching at this frequency.

[tszaboo]

the noise was measured with Scope and I used the smallest possible ground path with a pogo pin ground of the scope.
Attached is the noise image from scope.
the buck gets input from the SMPS which is 18V and I isolated the buck and measured directly at the AC-DC output and that 18V is clean.
In the design the resistor which sets the switching frequency of the buck is set to 100k ohm and that makes the buck at 480KHz. When i tried different resistor values, then the noise pattern changes.
there is nothing else in the system switching at this frequency.

Those high frequency spikes are likely still picked up through the air. The noise without those, the ripple is ~20mV.
You should make some specifications, what's acceptable and what's not. It always will be there, maybe you will not be able to measure it with your equipment. A few questions. Where does this 18V go, is it connected to any other sensitive circuits? Is it connected to a wire? If so, how much radiated noise does this cause, is it below the required legal limits, and the functional limits for your circuit? Did you measure it under load, does this change depending on your load?

What I'm trying to tell is, that seeing noise does not necessarily mean the circuit need to change.