0
\$\begingroup\$

I'm trying to troubleshoot a tps65261rhbr triple synchronous buck converter from 12V producing 1.8V, 3.3V and 5V.

The 3.3V and 1.8V are stable and well-functioning but the 5V is not. An oscilloscope here shows the 5V (blue) next to the 1.8V (red) for reference: enter image description here

The 5V hovers around 620mV with occasional spikes to 2.9V, but the calculations from the datasheet to select component values, and the simulation I've run from PSpice suggest that these components should work correctly. The voltage divider uses the exact resistance values specified in the datasheet for 5V.

Does anybody know what might cause this behavior in a SMPS?

My current best guess is that I'm tripping the over-current protection circuitry as the 5V comes online, causes a delay for several switching cycles, then tries again. I think the oscilloscope behavior seems to match this description from the datasheet:

The TPS65261, TPS65261-1 is protected from overload and over temperature fault conditions. The converter minimizes excessive output overvoltage transients by taking advantage of the power good comparator. When the output is overvoltage, the high-side MOSFET is turned off until the internal feedback voltage is lower than 105% of the 0.6V reference voltage. The TPS65261, TPS65261-1 implements both high-side MOSFET overload protection and bidirectional low-side MOSFET overload protection to avoid inductor current runaway. If the over current condition has lasted for more than the OC wait time (256 clock cycles), the converter will shut down and re-start after the hiccup time (8192 clock cycles)

However, I don't think the loads I have on the 5V should exceed the 2A that the SMPS supports on this power line. Currently the 5V has several loads: a raspberry pi CM3 VBat (700mA), a LDO driving a 4G cellular radio (maximum 500mA during transmission), a ZED-F9P GNSS circuit (130 mA), 4 relays, and some other assorted circuitry that should consume negligible current (e.g. USB switch). In total if everything was consuming the maximum current I believe it should be around 1.7A. So unless I've miscalculated somewhere, the only reason I could see enough current draw to trip the overcurrent would be a short circuit.

What I'm really looking for are suggestions for finding and proving root cause. I'm currently trying to rule out a short circuit, after that my next idea is to cut the 5v power trace and measure current over the gap and see if if I can power it with a bench power supply.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ As a style recommendation, I would avoid creating "footprint equivalent" schematic symbols because the physical construction of the chip is generally not amenable to the signal flow and readability of a schematic. Having grounds on the top and input voltage on the bottom, for example, is opposite to what one would expect. When a schematic is hard to read, it conceals otherwise obvious issues. Have you tried unloading the regulator and observing what happens? While your steady state power budget might be ample, it could be tripping on some inrush. \$\endgroup\$ – BB ON Jul 18 at 15:03
  • \$\begingroup\$ @BB-ON Thanks for the feedback on the footprint equivalent schematic, will keep that in mind. As for the inrush - I'll try to measure it with the current limiting mode on a bench supply as noted below. If it is an inrush issue, maybe a slower soft-start by using a bigger SS cap would help? \$\endgroup\$ – James Suffolk Jul 19 at 1:40
1
\$\begingroup\$

What’s the FB pin doing?

Anyway, looks like it’s current limiting. Try disconnecting it from the load. Another option is to use a bench supply to add current to the 5V (use current-limit mode). This would be supplementing the DC-DC output, so you could see what the load looks like.

\$\endgroup\$
  • \$\begingroup\$ This is exactly the sort of feedback I was looking for! I'll take a look at the FB pin and upload a pic. I only outfitted some of the loads with jumpers, unfortunately, making it difficult to disconnect all the loads. Will definitely add more jumpers on the next iteration. Great idea with the bench supply. \$\endgroup\$ – James Suffolk Jul 19 at 1:18
  • \$\begingroup\$ It was an overcurrent related to inrush apparently. The bench supply added enough current to power the it on and then I was able to disconnect the bench supply and it works as expected. However, one real mystery is that ever since using the bench supply, subsequent power-ons do not trip the overcurrent protection - as though only the first power-on drew enough current to cause a problem. Once the circuit had one successful start, now it powers up without the assistance of the bench supply \$\endgroup\$ – James Suffolk Jul 24 at 12:10
  • \$\begingroup\$ Maybe you had a whisker short on the board and the extra power burned it away. I’ve had that happen. \$\endgroup\$ – hacktastical Jul 24 at 17:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.