Step-up converter causing disturbances in radio

[Mercury]

Step-up converter noise

Hi guys!
In my car I have a 12V Step-up regulator (LM3421) at 400kHz, driving 1A 25V of LEDs. It's causing disturbances (noise) in my radio. I've tried putting 100uF - 22uH 100uF filter on the power supply of the driver, but It doesn't help.

I've attached a picture of the voltage on a 0.1 Ohm shunt.

(200mV/div , 1us/div)

The filter doesn't help at all with the spikes. Any idea what I should do?

Thanks!

 

[dick_freebird]

An open-bobbin inductor will spray magnetic field all over.

If your LEDs are being driven pulse-mode then the strings
are all antennae. If the output is well filtered "DC" then
that ought to be better. Still there is ripple and switching
spikes that can blow through the inductor winding capacitance.
A ferrite bead on the output wire might help that.

Decoupling the input supply to "ground" may, or may not
help - depending on which ground. The real key is to make
current loops short, close-in to the converter.

You need to be thinking about the switching spikes as impulses,
full of high frequency content, and forget about the baseband.
You probably don't need hundreds of uF - you need single uF
or less, with a self-resonant frequency at or above your radio
IF frequencies if not the carrier.

 

[Mercury]

Thanks!
I will try to filter the output of the converter and see if that helps. However I would still like to know, why the PI LC filter on the input did nothing to reduce the spikes?

Best regards!

 

[stephen-7486]

try earthing the case of the step-up converter to ground
i know it works with dc/ac converters

steve

 

[dick_freebird]

Thanks!
I will try to filter the output of the converter and see if that helps. However I would still like to know, why the PI LC filter on the input did nothing to reduce the spikes?

Best regards!

The input filter is only one of many paths for RF energy to
escape. It is a first defense against conducted emissions
through the input path. However it does nothing for radiated
emissions other than make the input power feed a poorer
antenna.

Automotive grounding schemes are often lousy, cost-driven.
Your "signal" path might really be in the chassis sheet metal,
which can be full of eddy currents including brand new high-
frequency ones. Try a "star" ground with everything run
back to the prime ground stud (where battery ties to block and
chassis) in separate conductors (at least, the audio and switcher)
and see whether that fixes things up any.

 

[Mercury]

Thanks for all your help guys. I've tried with star grounding and with a 2mm thick aluminum casing for the switcher, but bit no success.

So, I'm redesigning the switcher. In the LT3755 datasheet it says

"To reduce electromagnetic interference (EMI), it is important to minimize the area of the high dV/dt switching node between the inductor, switch drain and anode of the
Schottky rectifier."

So I will try to minimize that area, but there is a problem. PowerPak, D2PAK, TO-220... packages have the drain terminal as a thermal pad, which is very large by itself. So, how can I minimize that area?

Thanks again!