TPS65217: PMIC and processor communication issue

Shwetha,

If you refer to TPS65217x Schematic Checklist, you will find that some of the recommended terminations are not followed:

• AC: if not used, connect to GND
• ISINK1/2: if not used, connect to GND
• BAT1/2, BAT_SENSE: if no battery is present, connect to GND through a pull-down resistor (1kΩ to 10kΩ resistance)

But the BEAGLEBK board has the same incorrect terminations (except for AC supply) so I do not think this is the cause of your problem.

The biggest difference is that the BEAGLEBK  is a single-board computer, (or system-on-module, SoM) while you have a CoM (computer-on-module) and a base-board with a connector between the 2 boards. The TPS65217 PMIC is on the base-board (bottom) and the AM335x processor is on the CoM board (top). Also, the TPS65217D version is for DDR3L at 1.35V but the BEAGLEBK  uses TPS65217C for DDR3 at 1.5V

For your system, I would start by checking the following:

• Bottom board (with TPS65217): is LED D4 on? This indicates LDO2 supply is enabled
• There are no LEDs showing on the Top board schematic you shared, so I would advise you to check the following
  • VRTC = ?
  • VIO = ?
  • VDDS = ? 
  • VDD_1V8 = ?
  • VDD_MPU = ?
  • VDD_CORE = ?
  • VDDS_DDR = ?
• If all the power supplied on the top board are at the expected voltage, then the PMIC has delivered the power to the processor correctly
• At this point, both LDO_PGOOD and PMIC_PGOOD signals should be high

The assembly method and issues you are describing:

Shwetha Narayan said:

Actually now, we have hand soldered the power board including the 60 pin connector and sometimes, the system is not booting.

Indicate to me that this is a soldering issue. If you have multiple boards and the same board fails to boot all the time, then this definitely points to a soldering issue and at least one connection has probably not been made.

If you have a single board that boots sometimes and fails to boot sometimes, it also could point to a soldering issue. If the connector is pressed down, there might be a good electrical connection. If there is no pressure on the connector, it is possible one of the pins is not connected.

The best way to determine if the power supply is involved in a bad solder connection would be to setup an oscilloscope and measure these signals during power-on:

• LDO_PGOOD
• PMIC_PWR_EN
• WAKEUP
• PMIC_PGOOD

Measure them on the TOP board to see if they are making it to the CoM board, but this doesn't necessarily mean they are making it to the AM3358 processor.

You can monitor LED D4 and VCC_3V3A with a multi-meter during this time. If VCC_3V3A is always 3.3V, then the PMIC is not related to your booting issue because it is the last power supply to turn on in the power-up sequence. If PGOOD goes high and stays high, then this means the same thing as VCC_3V3A = 3.3V continuously. If you capture these signals on the scope and the board fails to boot, you might find the root cause.

The schematic wiring looks correct for the PMIC and Processor. If I were to re-assign this to the Sitara team for further debug, they would probably want to see the full schematic (showing SD Card and eMMC). Of a 10-page schematic, only 3 pages are shared for the top board.

The only other thing I can learn from these schematic is that both eMMC and SD Card interface are on the bottom board. If these connections are not made properly, it would definitely result in a failure to boot the system.

• Does eMMC always work and SD boot sometimes fail? 
• Does SD Card always work and eMMC boot sometimes fail? 

This is also a good way to look for root cause related to connector, either due to soldering issues or poor high-speed routing.