Hello community,
we have designed and built hardware with a TPS563219A PMIC voltage regulator.
For the last production batch (1000 boards) the TPS563219A was not available through our distributors, so based on the datasheet and
https://e2e.ti.com/support/power_management/non-isolated_dcdc/f/196/t/561563?TPS563219A-TPS563219A-vs-TPS563219
we used TPS563219(non-A) as a direct replacement. We are running into some serious problems with this part, see below.
After replacing the non-A version with the A version on 2 faulty boards, everything worked as intended.
Circuit
Our circuit is basically the reference design:
The main difference is a bigger inductor, in our production batch 4.7uH instead of 3.3uH were used.
Possible/recommended values in the datasheet Table 2 "recommended component values" list the values used, and the inductor is within
the specifications for current/peak current.so I dont see a problem here.
Layout
The board is a 2-layer single-side PCB.
Behaviour TPS563219(non-A)
When the TPS563219(non-A) is used, the output of the TPS seems to remain disabled.
On the following screenshots, the blue trace shows the input voltage.
The yellow traces show whatis named above the screenshot.
Output Voltage - faulty
device internal power supply
laboratory power supply, zoomed in on first peak.
Output Voltage - stable
Sometimes, on some boards, the output voltage reaches a stable state.
device internal power supply
Soft Start Voltage - faulty
time constant for Soft Start charge as expected, but discharging repeatedly.
laboratory power supply.
Switching output voltage - faulty
laboratory power supply.
Different resolutions on time axis.
Behaviour TPS563219A
Replacing the TPS563219 with TPS563219A fixed the problem.
Laboratory Power Supply.
Questions
- What are the differences between the A and non-A version (including minor changes)?
- Are there any values in the system that are on the edge of allowed values (with the A version being okay with these and the non-A version reacting less forgiving?)
- Do you see potential problems in the routing/layout?
- Is the behaviour for the non-A version expected for some known faults? What is the reason for this behaviour?
- How can we know we will not run into similar problems with the A version?
Thanks a lot,
Lars Heinrichs