Hi,
Two months ago, I posted a topic related to EMI's performance of The LM5085 DC/DC.
We use it to convert Dc voltage from 28V to 3,3V. The electronic diagram is the following :
We have to respect class 5 of the CISPR25 EMC standard. With this configuration We meet some problems with radiated emission levels. (between 150kHz and 30MHz).
First the requirements were :
- Fsw = 400kHz;
- Output ripple : 100mV (not fixed);
- output max current = 1,4A;
After several researches, we tested different options :
1) Put a bigger gate resistance R4003 in order to slow commutation steps down :
We managed to go over 27 ohms but we have bad thermal and cunsomption results and the spectrum level is the same
2) Put a Gate-Drain capacitor :
We don't know how choosing its value
3) Add a RC snubber :
We think it is not necessary because we don't observe overshoot (switch ringing)
4) Changing switching frequency Fsw :
Thanks to the Rt resistance, We can choose different switching frequencies.
After several tests We noticed that when we made Fsw higher ( > 500kHz), the radiated emission level seemed to be less important.
But We noticed the switching frequency was not steady. So we wondered, if the LM5085 has stability criteria above a known-frequency.
5) Changing the ripple network (R4010, C4011 and C4016) :
We think when reducing output ripple a little we could improve EMC performance.
We found different tutorials answering calculate ripple adjustement but we are not sure, especially for 600 kHz switching ...
6) Put a shielded inductance (L4000) :
We noticed that a great amount of magnetic field came from the buck inductance L4000, so we would like to try a shielded inductance like HILP series.
We can't route our circuit again, so we are trying to find a great deal of hardware solutions (changing components).
Do you know some tips or advices that treats this kind of question ?
Thanks a lot for your answers,
Kind regards,
Olivier D.