AnsweredAssumed Answered

Viper12A buck-converter. Inductor value selection for peak currents vs EMI

Question asked by dm-ml on Jul 17, 2012

I am developing a product that is very similar to the buck-converter circuit presented in AN2544, using a viper12AS setup for 12V, 100mA output (using the 12V 200mA circuit as similar reference). This product is awaiting EMI approvals (EN55022 class B) to allow production, however our EMI scans have shown the conducted emissions to be borderline on the limits around the 150KHz to 1.5MHz region; this is despite careful layout of the PCB, input X-capacitor (upto 100nF), and input pi-filter. Detailed scans show that by far, the greatest contributor to conducted emissions here are those frequencies that are multiples of the viper12A's switching frequency.

I've run scans to find improvements in changing capacitor and filter values, but these all offer very marginal changes, even at very light (ie 10mA) load. Changing the buck-converter's output inductor, however, from 1000uH to something lower has given dramatic improvements in the EMI scans, even with the circuit loaded to its maximum of 100mA. Going to a 330uH inductor for example, provides an approximate 6dB clearance from the EN55022 B limit line, and appears to still provide a stable 12V output at 100mA. As expected, the circuit becomes more discontinuous.

Unfortunately, the current waveforms for the buck-converter's output inductor suggest the original 1000uH inductor is already quite close to the viper12A's lower Idlim of 320mA (I record 240mA and 270mA peak currents with 10mA and 100mA of load respectively). Despite this, the 330uH inductor appears to charge with 670mA and 700mA peak currents (10mA and 100mA loads), considerably more than the datasheet's upper Idlim of 480mA - which is surprising - yet still provides a very stable 12V output (ripple less than 0.5Vp-p).

So I find myself quite confused at this stage. The lower inductor provides massive improvements in conducted EMI and appears to have no impact on the output provided by the circuit. Examining the inductor's current waveforms however shows the current 1000uH inductor to be close enough to the lower Idlim to suggest not using a lower inductor (and causing the higher peak current, as explained in AN2544). I have considered changing to a viper22AS for the higher Idlim, but given that the circuit output is only 1.2W, something does not quite seem right about doing this!

I'd greatly appreciate any input or assistance with this matter.

Kind Regards,

PS) Some example waveforms have been attached to show the inductor currents mentioned above. These were measured with a calibrated current-probe, accurate to +/-10mA.