AM3358: SYS_RESETn pin always low despite good power supply to the MCU

Hello Taringana Guranu

Thank you for the query.

Checking if you have reached out to the beagle team to understand if they have seen similar issues?

Regards,

Sreenivasa

I had not reached out to them. I assumed they will probably direct me to TI since the issue is not happening exactly on a Beagle Bone board but let me do that and see if this is a known issue.

Hello Taringana Guranu

Thank you for the reply.

I would suggest reaching out to beagle and see what you hear from them.

The schematics page you attached is also from beagle, so it may be a good idea to seek their answer.

Regards,

Sreenivasa

Hi Sreenivasa,

I have been unlucky in getting a response from the Beagle team. I also decided to have the CM do an X-ray in case the AM3358 warmrstn pin is shorted to ground. The X-ray did not show any shorts. 

Hello Taringana Guranu

Thank you.

Do you see this issue on a specific board or on multiple boards.

The information for the custom board is partial.

Are you able to share additional information.

Regards,

Sreenivasa

I am seeing this issue on all the three boards I am working with. I have provided the whole schematic as it related to the AM3358 and the TPS65217 which is powering the system. Is there more information you need me to provide?

MCU and PWR.pdf

Hello Taringana Guranu

Thank you.

Could you please share a searchable schematic.

Regards,

Sreenivasa

Hi Sreenivasa,

The document below should be searchable.

MCU and Power.pdf

Hello Taringana Guranu

Thank you.

Let me review the schematics and comeback.

Regards,

Sreenivasa

Hello Taringana Guranu

I cannot locate the VIO/2.4C supply source. Could you please verify the source.

Regards,

Sreenivasa

Hi Sreenivasa,

The source is pin 3 of the TPS65217CRSLR. 

Hello Taringana Guranu

Thank you.

Could you please verify the supply level on the buffer supply since the net names seems different.

Regards,

Sreenivasa

Hi Sreenivasa,

The voltage is 1.8V. The numbers that appear after "/" on the net name are for referencing the sheet in the schematic where the signal also appears if it is not on the same schematic sheet. The actual signal name is just VIO.

Best,

Taringana

Hello Taringana Guranu

Thank you.

Do you see the PMIC_Pgood toggling during power-up.

Regards,

Sreenivasa

Yes the PMIC_PGOOD goes up during power up. On one of the boards, I depopulated the buffer (SN74LVC1G07) to isolate any behavior from the PMIC from the SYS_RESETn line, that did not change the results on the SYS_RESETn line.

Hello Taringana Guranu

Thank you.

Could you please confirm the clock is functional.

I assume all the supply rails have been verified to be normal.

Regards,

Sreenivasa

Hi Sreenivasa,

I just confirmed the 24MHz master clock is not working. I will be doing more debugging to figure out the issue and see if that is something I can fix on these boards. 

Thanks,

Taringana

Hello Taringana Guranu

Thank you.

I assume you have verified availability of all the supply rails for the processor. If not please verify the supplies as you debug the oscillator.

Regards,

Sreenivasa

Hello Taringana Guranu

Please verify the crystal load and the load caps used.

You may have to use a lower load crystal.

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/388785/am335x-osc0-crystal-circuit-load-capacitance/1372124?tisearch=e2e-sitesearch&keymatch=am335x%25252525252525252520xtalin#1372124

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/249172/am3352-stuck-in-reset/872030?tisearch=e2e-sitesearch&keymatch=am335x%252525252525252520xtalin#872030

Regards,

Sreenivasa

Hello Taringana Guranu

Plese follow the below procedure to measure the clock on the clockout1 pin.

If possible, set SYBOOT[5]=1 to enable CLKOUT1 to output after power up and check its frequency.  This should reflect a square wave clock at the frequency of the crystal if everything is working correctly.

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/980121/am3358-custom-board-bring-up-issue/3621869?tisearch=e2e-sitesearch&keymatch=am3358%20clkout1#3621869

regards,

Sreenivasa

Thank you for the help Sreenivasa.

I used a 4 pad 24MHz oscillator that had a wrong library footprint. The output and the oscillator ground pins were swapped therefore the input clock was not starting. All the power rails are good so the oscillator is probably responsible for the reset staying low. I am working on a fix that should hopefully fix the reset pin issue.

Best Regards,

Taringana

Hello Taringana Guranu

Thank you.

The crystal load based on the schematics part number is 18 pF.

FW2400063Q – 24 MHz ±15ppm Crystal 18pF

Any crystal with a 18pF load requirement is going to require more capacitance than we allow in the data sheet.

If possible, please, try to source a crystal with load of 12 pF to match the load caps used.

When you are done with the changes, please measure the clockout1 before continuing the tests.

Regards,

Sreenivasa

Hi Sreenivasa,

I don't understand why a crystal with an 18pF load requirement being out of the required range when the datasheet maximum is 24pF.

I managed to rewire the same crystal and I now have a 24MHz clock on both the OSC0_OUT (Yellow) and CLKOUT1 (Blue) as shown in the image below. The CLKOUT1 is not a square wave, should that be a concern? The datasheet does mention that CLKOUT1 should be an exact replica of OSC0_OUT although my signals do have different magnitudes. I am now having issues with booting the board which I need to investigate more as I am not observing any clock on the MMC0_CLK line.

Hello Taringana Guranu

Thank you.

Taringana Guranungo said:

I don't understand why a crystal with an 18pF load requirement being out of the required range when the datasheet maximum is 24pF.

The 24 pF is the load cap and not crystal load.

I am not in a position to comment of the crystal behavior. Recommend following the data sheet recommendations.

Regards,

Sreenivasa

Hi Sreenivasa,

I got a 12pF and replaced it into the design and it is behaving similar to the 18pF. I am having issues booting from the uSD and not observing any clock activity on the MMC0 line. I will be working on debugging the new problem. Thank you very much for helping me debug the reset line issue. 

I will replace the 24MHz crystal with one that requires a 12pF going forward.

Best,

Taringana

Hello Taringana Guranu

Thank you for the update.

I tried looking at the schematics but looks like the SD card section was not included.

Please verify if you have the required pulls on the MMC interface for SD card.

Regards,

Sreenivasa

Hello Taringana Guranu

Checking if you made some progress.

Regards,

Sreenivasa

Hi Sreenivasa,

I will only be able to do more debugging with the boards next week so I haven't made any progress. I have the pull ups needed and I have also attached the whole schematic to include the MMC and wireless interfaces.

Custom BBGW.pdf

Best,

Taringana

Hello Taringana Guranu

Thank you for the updates.

Regards,

Sreenivasa

Hi Sreenivasa,

I discovered that my schematic has the wrong part number for U3. It contains SN74LVC1G07DCK, a buffer,  instead of SN74LVC1G06DCK, an inverter. The output of the inverter controls eMMC_RSTn therefore my eMMC has been in reset mode.

I did remove a resistor, R162, to make sure eMMC_RTSn is always high but that did not fix the issue.

I also notices that MMC1_CLK (eMMC_CLK) is also tied low like MMC0_CLK when SYSBOOT2=high but it is high when SYSBOOT2=low. Seems like I still have another bug causing both my clocks to remain low even though they both have their pull up resistors.

I am waiting to get the correct inverter and replace it on the boards in case that might make a difference which I am doubting.

Best,

Taringana

Hello Taringana Guranu

Thank you for the updates.

Regards,

Sreenivasa

Hi Sreenivasa,

I have been doing more debugging but have not managed to solve the clock issue on MMC0. I connected a cable on UART 0 and I was able to only get 8 "ccc" before they stopped when the uSD card was connected although they were not coming out on every power cycle. When I remove the uSD, the "ccc" are always transmitted continuously without any issue. 

Any idea of where the board might be getting stuck? I double checked and VDD_CORE and VDD_MPU are both 1.1V while VDDS_DDR is 1.5V.

Best,

Taringana

Hello Taringana Guranu

Thank you.

Let me look at the schematics.

Regards,

Sreenivasa

Hi Sreenivasa,

I just wanted to check if you have managed to find any bug in the schematic that might be responsible for the issue I am experiencing on the MMC0 clock line.

Best,

Taringana

Hello Taringana Guranu

Thank you.

Let me review the schematics one more time.

Regards,

Sreenivasa