Problem with radiated emissions from TPS51116PWP

transfer to non-isolated

Bump

You may want to consider an RC snubber from LL to PGND.  Alternately you could look at a series boot resistor connected in series with the cap from LL to VBST.  Pay careful attention to layout.  make sure all the circulating current loops are small.  Make sure the LL trace is not too long or narrow.  Make sure the VIN caps are tightly coupled to the High side FET and PGND.

Thanks for the reply John. I've finally gotten back to investigating this problem. Per the layout guide section of the device datasheet, I checked the LL pin for the presence of any high frequency surges and observed switching and some ringing per the attached capture. Is this expected? Should I be concerned about the ringing? The timing of the pulses varies quite a bit, but the duration of the highs and lows are quite constant.

The waveform shows DCM operation, so you must be operating at light load.  The TPS54116 operates in pulse skipping forced DCM at light load, s the waveform you show looks normal.  If the load current is raised above one half the peak to peak inductor current, the device will operate in CCM and LL waveform will be a continuous stream of rectangular pulses.

We're only sourcing power to 2 DDR2 chips and a processor. Does this DCM mode of operation explain the radiated emissions at 3* the DDR2 clock rate?
Would you recommend adding loading resistors to increase draw? If so, on all rails?

When the inductor current runs dry there is ringing and exponential decay that you can see in your waveform.  Measure that ringing frequency and see how it compares to your radiated frequency.  Usually that ringing froquency is the problem for EMI.  If it is really a harmonic of your DDR cloack, then you may have to look at that circuit rather than the dc/dc converter.

John

It looks like the ringing frequency is about 2.7 MHz, well below the emissions at 800 MHz.We need to head to layout, so we'll just implement the snubber and the series boot resistor on the LL to VBST path that you suggested previously.

Do you have a recommended value for the series boot resistor connected in series with the cap from LL to VBST?

I have not worked directly with TPS51116.  But for other ICs, I have used 10 or 24 ohm.  You might start with one of those values.

Thanks for your help John