Hello,
I recently helped debug a circuit board where this DC-DC boost switching supply would be taking a USB input voltage of 5V (standard USB 2.0 spec), then make Vo = 6V @ max 100mA under full load to electronics downstream. The layout is in very good agreement with pages 21 & 22 of the datasheet with the overall design taken from the page 1 simplified schematic with a ~62K resistor in the top of the feedback path (R1) to adjust for the 6V output from what would be 15V out using the 200K resistor. All other components including the recommended 100nF SS cap were implemented as stated in the applications section to operate at 1.2MHz.
Upon operation of the custom board, the chip was observed to draw ~190 & 250mA (2 separate tests), when conservative calculations showed it should've been close to ~40mA at the DC-DC's output. At that point it could still have been in our design, but the chip quickly self-destructed concluding the second "load" test and the input and switch nodes were observed as shorts to ground. On the same board a ferrite bead was removed in order to isolate the rest of the board's electronics then they were subsequently verified to be perfectly functional. It's suspected that the chip succumbed to these currents, which immediately pointed to either the in-rush being too great given the recommended capacitance being too large AND/OR the soft start not being sufficient. But page 6 Figure 2 of the datasheet indicate that the unit should be able to handle much more current than the suspected 250mA seen to cause the chip to burn out, although perhaps not that close to the rated 3.2A switching current. Additional testing was done that didn't really vindicate these two assumptions and as luck would have it, I happen to have an eval board for this chip for 5V -> 15V again operating @ 1.2MHz (with an on-board 3.3uH inductor).
From here, a 100ohm potentiometer was used to present a variable load at the output of the eval unit, which was tested as delivered to begin for a point of comparison. In short, outputting 15V and using the high power rated 100ohm pot, when the load got to ~300mA I heard a loud audible whine and watched temperature climb rapidly (as current would also do if I turned the pot any further to reduce resistance) indicating yielding so I disconnected power. This was repeatable and again, this ~300mA mark doesn't come close to the indicated performance of the chip, now being operated on an eval board manufactured by Ti, of the ~1.1A max from page 6 Figure 2 of the datasheet for 15V out produced from 5V input.
To take it further, we went ahead and replaced R2 of 200K in the feedback loop with the necessary ~62K resistor (now designated R2) to tune the circuit to output 6V. Again using the high power pot, the same audible pitch occurred at a load of ~800mA with similar temperature concerns. The conclusion was that operating beyond this limit for a matter of seconds or less would result in a catastrophic failure of the chip. Yet again. However this time extrapolating Figure 2, the chip was nowhere near reaching the potential output current from the datasheet. The operating conditions in the electrical characteristics in the datasheet are pretty lean too, they do specify Vin and Vout though.
So, considering layout sensitivity, in-rush/soft start, and a bad run of chips (virtually impossible given the eval board results), I need help understanding why I’m seeing the output current I’m seeing as well as the self-destructive nature of this component as it has been designed to the application example on the first page of the datasheet. I need to wrap my head around this then work on to why excessive current was seen when connected to the intended circuit/load. Save any of this, from the fact that the eval board right out of the package wasn’t making the spec’ed numbers either... Any thoughts?
Any help, advice, or observations would be appreciated and thanks in advance,
b