EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: uer166 on September 19, 2019, 02:13:30 am
-
I've never had issues with TI Webench online tool before but lately it's been either misleading, or maybe I'm reading it wrong. Seems like the more time I spend trying to design a power supply the more it sucks..
Goal is a 48V->3.3V supply at 300mA, nothing hard. Attached circuit. According to the TI tool I remember the output voltage ripple being <10mV. But, what seems is the controller works in bursts: bursts a couple pulses, inductor goes to 0 (DCM), then rinse and repeat. While the voltage ripple during the pulses is well under 10mV (I mean it operates at 2.1MHz), the second-order ripple due to this bursty behavior at around 100KHz is something like 200mV, completely unusable for my application.
I don't see how the above could be an issue with ECO mode since this persists even at 600mA full output current. Only thing I can think of is there is a minimum ON time on the switch which happens to generate a voltage above 3.3V, enforcing burst mode?
I even double checked the output ripple voltage and other parameters using the datasheet provided equations.. So what gives? The layout is plenty tight, almost no visible ringing during the switch transitions (except low energy ringing whenever inductor current reaches 0, as expected).
-
The only odd thing I saw is that Cboot seems mighty large at 1µF, typically that's in the 100nF range. But I'm not sure if that would cause such behavior.
-
I think the problem is that you are dropping a lot of voltage, which requires a low switch duty cycle (3.3/48=6.9%). If I'm interpreting it correctly the datasheet shows the minimum on time at 2.1MHz is 80ns, giving a minimum duty cycle of 16.8%. Using the 700kHz part (with appropriate value inductor) would help.