Voltage Spikes from Buck Converter Module

Hello,

i have a Audio Project where have to step down 44V to 5V ( Buck Module ) for an Arduino and some Sk9822 LEDs to let the LEDs blink to Music.
But as soon as i connect the Buck Converter with an 120ma Load ( 144 LEDs in Idle ) i get massive noise in my Audio system. When i look on the Power supply voltage with my oscilloscope i see many voltage spikes on it. Im pretty sure the Audio noise comes from these Spikes on the 44V section.
These spikes are 9uS apart from each other which is atleast close to the switching frequency of 120khz.

I tried putting an 1Khz LC Lowpassfilter ( 35mH and 900nF ) infront of the Buck to reduce that noise, but i didnt help.
Has anyone experienced similar problems with Voltage Spikes from Buck converters and knows how to reduce them?

I'm going to just take a wild guess here but if those are led strings then there is probably a 150% chance they are controlled with PWM so maybe the noise in your audio is the LED pwm switching frequency.
Try driving a load that doesn't use pwm to see if the noise goes away.

Jeah, the LEDs are Conntrolled by PWM just like the well known Neopixels.
But this problem also accours when all LEDs are off, so there should not be any PWM involved.

I'll measure again tomorrow with a Ohmic Load at the Buck converters output, and see if that changes anything. If so, i should filter the Supply voltage of the LEDs i guess.

Why is the low-pass filter before the buck converter?

Shouldn't it be between the output of the buck converter and the audio stuff? Probably with the LEDs' power connected before the low-pass filter....

I would also suggest an additional largeish cap on the power supply after the low-pass filter.

KidCe:
Jeah, the LEDs are Conntrolled by PWM just like the well known Neopixels.
But this problem also accours when all LEDs are off, so there should not be any PWM involved.

I'll measure again tomorrow with a Ohmic Load at the Buck converters output, and see if that changes anything. If so, i should filter the Supply voltage of the LEDs i guess.

Your filtering will have to go to the source of the noise. Could be radiating from the input wires of the buck converter. Use your scope to debug the problem!

Paul

DrAzzy:
Why is the low-pass filter before the buck converter?

Shouldn't it be between the output of the buck converter and the audio stuff? Probably with the LEDs' power connected before the low-pass filter....

I would also suggest an additional largeish cap on the power supply after the low-pass filter.

Analog Devices said it on their website that Buck converters have high Noise on the input and Boost Converters have high noise on the output, so the input of a buck should be filtered.

Okay, instead of going to bed i just tried it with an ohmic Load of 10 Ohms to put an 0.5A Load on the Buck.... And i still get terrible loud noise ! :confused: LEDs were fully disconnected.
So im Pretty sure ill have to do Filtering of the Buck converter.

It's the buck switching

Yes you will. If audio is involved, you may need to mount it in a metal enclosure (you can make it out of PCB) with the filter components mounted where the wires pass through the walls and all grounds terminated on the enclosure itself.

KidCe:
I tried putting an 1Khz LC Lowpassfilter ( 35mH and 900nF ) infront of the Buck to reduce that noise, but i didnt help.

Why only one such a small capacitor? For power supply decoupling it's common to use bigger capacitors, usually with a smaller ceramic in parallel for the higher frequencies. You don't filter a single frequency as it's a block wave, the buck converter basically switches on and off very quickly.

For your case, I'd consider a Pi filter - with capacitors on both sides of the inductor, e.g. 100µF electrolytic, 1µF ceramic and 100 nF ceramic in parallel.

KidCe:
144 LEDs

Here I see another major issue. That buck converter you link to is rated 3A, so a maximum continuous load of 1.5-2A output would be a safe limit (always derate those made-in-China modules), maybe even less as it has to step down 44V all the way down to 5V, which even at a very good 90% efficiency means about 1W of heat dissipation at 2A. Your 144 LEDs require 7.2A on maximum brightness, no way that buck converter is going to supply that much current.

wvmarle:
Why only one such a small capacitor? For power supply decoupling it's common to use bigger capacitors, usually with a smaller ceramic in parallel for the higher frequencies. You don't filter a single frequency as it's a block wave, the buck converter basically switches on and off very quickly.

For your case, I'd consider a Pi filter - with capacitors on both sides of the inductor, e.g. 100µF electrolytic, 1µF ceramic and 100 nF ceramic in parallel.

Here I see another major issue. That buck converter you link to is rated 3A, so a maximum continuous load of 1.5-2A output would be a safe limit (always derate those made-in-China modules), maybe even less as it has to step down 44V all the way down to 5V, which even at a very good 90% efficiency means about 1W of heat dissipation at 2A. Your 144 LEDs require 7.2A on maximum brightness, no way that buck converter is going to supply that much current.

The 900nF ist just for the LC Filter, so it is very close to the inductor. I have 500uf Electrolytic cap + some lower value ceramics close to the load.

Ill order some parts to try out different versions for a Pi filter soon. Hopefully that helps.

And yes i know this buck cant light up all LEDs at one at Fullbrightness, but thats not a problem becaus they just blink and do some Animations which i measured for worst case peaks of 2.2A, but 90% of the time less than 2A

If you are trying to attach a high current switch-mode converter to the supply rail of an audio amplifier, forget it,
you need an entirely separate supply. 44V -> 5V buck conversion without a transformer isn't very efficient
anyway - better to get a decent separate mains powered 5V switch-mode supply.