TMS320F28386S: F28386S did not boot up when power up

There could be some marginality in the design of your reset circuit. Would you be willing to share the reset circuit privately with me? You can also capture the power-up reset waveform for VDDIO, VDD & -XRS pins for both good and suspect boards and see if there is something amiss in the way the boards are powered up.

Thanks for the response. How do I upload the files? Thank you.

You can simply drag and drop image files in the reply window. You can share the schematics through a private message. For this, please accept the friendship request I have sent you.

I have accepted the friendship request. How do I privately message you?

The waveforms are not helpful. Unless each image is clearly annotated with the test conditions, I don't know what I am looking at. Also, the waveforms should not overlap each other. Instead of taking a photo, please save the waeform to a USB stick and then attach that file to the post. This way, the image resolution will be much better.

Chun-Te Li said:

I have accepted the friendship request. How do I privately message you?

I have sent you a private message. You can reply to that.

You mention you had no issues with your previous batch and that you only saw an issue in 3 out of 5 boards in the current batch. This tells me that nothing is drastically wrong with the design but there may be some marginality in either the HW or SW. I didn’t spot anything very wrong with the power-up waveforms or the schematics. Without access to the hardware, I can only offer suggestions for debug:

1. Do the 3.3v & 1.2v rails have the required current-sourcing capacity? “Starving” the device during power-up can lead to unpredictable behavior.
2. Does the power-stage have adequate margin to meet the in-rush current demands of the system?
3. Is the required power-sequencing being followed?
4. Is the -XRS pin driven by an open-drain (OD) OUTPUT?
5. Are the boot-mode-select (BMS) pins at the expected level when the device is taken out of reset? Incorrect pin levels would force the device into some other boot-mode, not Flash. This may be construed as a device failure while the device is just looping in some other boot-mode.
6. Does the device boot-up correctly if you flash a simple C2000ware example like “Blinky”?
7. Does the example code work fine off RAM? If it does and only has a problem running off Flash, it may point to some initialized constants in RAM.

Thanks for the suggestions. One thing we have noticed is that, two of the three boards with issue, we could boot up if, after initial power up (but not boot up), we turn power off and back on quickly, the boards would boot up correctly. Would this information helpful in debugging? Thank you.

Your schematics does not show how the boot-mode-select (BMS) pins (GPIO72 & GPIO84) are handled from a boot-mode select perspective. Their connection to the SDRAM and the RS485 chips are shown but at what state are they in during power-up? Please capture the following during power-up: 1.2v, -XRS, GPIO72 & GPIO84. Capture this for both good and bad power-up. 

From yesterday’s post: 

Are the boot-mode-select (BMS) pins at the expected level when the device is taken out of reset? Incorrect pin levels would force the device into some other boot-mode, not Flash. This may be construed as a device failure while the device is just looping in some other boot-mode.

Thaks for the information on BMS. I think our firmware have changed the BMS to different GPIO pins, but I would need to check with the firmware engineer. The BMS could be changed to different GPIO pins, right? Thank you.

Yes, the BMS pins can be reassigned on this device.