Buck-boost converter (LTC3442) not producing output until excited

Question

EDIT: Added datasheet for the inductor and some oscilloscope measurements.

I'm using a LTC3442 buck-boost converter from Linear Technology / Analog Devices (datasheet) in my design to produce a 3.3V output from a 2.5-4.5V input (Li-Ion cell, bat+ in schematic), but I'm facing some problems.

When the input is applied, nothing happens and the output is steady at 0.0V. However, if the output is excited by connecting it momentarily to the input (shorting +3V3 to bat+), everything works as expected and 3.3V is maintained at the output - but only until the power is disconnected. After that, the process must be repeated to "power up" the device.

Thus, something is definitely working, but the power up sequence is somehow not working properly. What can be the cause of this, and what can I do to fix it?

I hope you are able to help!

Other observations:

• The output can only be triggered by the shorting "trick", if the input is above 3.3V
• After powering up, the 3.3V output works across the entire input range (2.5 to 4.5V)
• When the output current is increased beyond 300mA, it turns off and cannot be powered up until the load is removed.

Inductor specs:

• Murata DFE201612E-4R7M=P2 (datasheet)
• Type: Wirewound and shielded 4.7uH inductor
• Maximum DC current: 1.2A
• Saturation current: 1.8A.

DSO plots Below is some DSO plots as the lines look just after power-up and after the output has been excited as described above:

Answer 1

Okay I found a solution now!

Inspired by @StainlessSteelRat's answer I tried switching out the inductor to some others ones I had lying around with a lower DCR, but to no avail.

Following the same line of thought, I reflected on what else could induce too high resistances in the power-carrying paths. In my PCB layout, the large PGND pad on the bottom of the chip is directly coupled to a quite large ground plane through multiple vias and thus provides very good thermal transfer. However, this could also cause some issues with the reflow of solder if not properly heated and thus induce a poor / high resistance connection from PGND to supply GND.

After reworking the solder with more heat and flux, everything works as expected (validated on 3 boards now) with the original inductor. I tested it up to 1A of output.

So it was a poor solder job on my side causing the problems. Sorry for the inconvenience and thanks for the inputs!

Answer 2

From linked LTC3442 datasheet:

For high efficiency, choose a ferrite inductor with a high frequency core material to reduce core loses. The inductor should have low ESR (equivalent series resistance) to reduce the I2R losses, and must be able to handle the peak inductor current without saturating. Molded chokes or chip inductors usually do not have enough core to support the peak inductor currents in the 1A to 2A region. To minimize radiated noise, use a shielded inductor. See Table 1 for a suggested list of inductor suppliers.

Sadly, Table 1 does not list recommended inductors.

Their DC799 Reference Design using part LTC3442 recommends the 5μF Coilcraft MSS7341.

From Mouser:

5μF Coilcraft MSS7341 has 23mΩ DCR. You have a fixed chip inductor at 252mΩ DCR. Thicker wire with less resistance and larger mass.

You do not have enough inductor for the LTC3442 to work. The fact you can trick it by charging up the output capacitors is meaningless because the inductor can not deliver expected current.