TMS320F28335: SYSCLK Change suddenly

Ali,

It is likely that you are causing a noise event in your system. Can you please address the following questions?

• Does the problem only occur during active debug with you touching/probing different parts of your circuit? .
• Does it occur during normal operation on it's own without you probing or touching circuits?
• Do any regions of your board seem to cause the issue to occur more often?
• Do you have XCLKOUT enabled and can you observe the clock for sure changing frequencies?
• Does your system continue to operate, but at this reduced frequency during the event

As a general protocol - be sure to follow safe ESD handling procedures while handling your PCBs. If the issue is in fact due to an ESD event, you may encounter permanent device damage in the future. 

https://www.esda.org/about-esd/esd-fundamentals/

Regards,

Mark

Dear Mark,

"Does the problem only occur during active debug with you touching/probing different parts of your circuit? " Yes, Actually while my system is under test, it has randomly occurred that my power switches damage. The latest experience was last week that after a month of constant work my switches suddenly exploded and i could not find a rational reason for that. So I got suspicious of the mentioned phenomena.

"Does it occur during normal operation on it's own without you probing or touching circuits?" No it does not. I have never experienced this.

"Do any regions of your board seem to cause the issue to occur more often?" Yes. the regions near oscillator are more sensitive.

"Do you have XCLKOUT enabled and can you observe the clock for sure changing frequencies?" Yes I do. I saw the change of frequency with oscilloscope.

"Does your system continue to operate, but at this reduced frequency during the event" Yes it does. even the frequency of of my PWMs change according to the change that occurs to the SYSCLK frequency

Regards

Ali,

Thanks for responding. This does sound like an ESD event, potentially causing an oscillator failure.

• It seems like the failure is catastrophic to your system, where your power switches are blowing up. Is that correct?
• How do you recover your system?  Just a reset? (and replacing parts)
• Are you able to reproduce this on demand? if you can do this without the power switches connected (preventing damage) that would be optimal.
• I would like to see a scope capture of the event in question. please share both the input clock, and XCLKOUT. I am curious if the external oscillator is in fact failing. 
• If possible, have the debugger connected while this event occurs so that you can save the status of your PLL control Registers, and share them.
• What frequency does the SYSCLK change to? What was it before?

It is possible that the event has caused the device to go into limp mode due to a missing clock event. Check the status of PLLSTS.MCLKSTS after the event and let me know. Please review the "Main Oscillator Fail Detection" section of the SysCtrl and Interrupts User Guide - SPRUFB0 for guidance on this type of issue.

-Mark

If you are able to reproduce this with CCS running, please examine the status of the PLLSTS.DIVSEL register and check if/how the values have changed. Also, is a change in system clock the only effect you see due to noise? I would expect a few other random phenomenon like corruption of other bits/registers, code control gone awry etc.

Mark,
Thanks for responding

"It seems like the failure is catastrophic to your system, where your power switches are blowing up. Is that correct?" Yes, when the frequency of PWMs change the behavior of the system totally changes.

"How do you recover your system? Just a reset? (and replacing parts)" Yes, i recover my system with replacing parts. I must add that I am not sure of the fact that this phenomena occurs while the system is under operation. This is because it is not a rational thing to touch the sensitive parts of the system while its under operation. But I have seen the occurance of this phenomena while I am testing the system in signal mode.

"Are you able to reproduce this on demand? if you can do this without the power switches connected (preventing damage) that would be optimal." Yes, I am. I can simply reproduce this phenomena by touching the sensitive parts ( section near oscillator )

"I would like to see a scope capture of the event in question. please share both the input clock, and XCLKOUT. I am curious if the external oscillator is in fact failing." Sure, I would send the capture soon

"If possible, have the debugger connected while this event occurs so that you can save the status of your PLL control Registers, and share them." Sure, I would do

"What frequency does the SYSCLK change to? What was it before?" My main frequency is 150MHz, and after that, it changed to 75MHz or 37.5MHz

Ali,

It really does sound like an ESD or noise event from being touched with probes. If you can replicate this without damaging parts that would be optimal. Hopefully you are able to do so by just removing the damaged switches without replacing them (Open the circuits).

I look forward to seeing the results of the debug.

Regards,
Mark

Have you been able to debug this issue further?

Thanks,
Mark