DSP suddenly freezes while the program is running

zqx123

DSP suddenly freezes while the program is running [Copy link]

 

I am currently developing a power supply project using TI TMS320F28035. When the power supply is close to full load, the DSP sometimes dies, the operating indicator light no longer flashes, the PWM port does not send waves, and no fault code is sent to the host computer. When it is unloaded or lightly loaded, it runs normally. It seems that when it is heavily loaded, the DSP is harassed. Has anyone encountered this problem? Please advise. Or point out where to start to find the problem. Thank you!

maychang

"When the system is close to full load, the DSP sometimes dies, the operating indicator light stops flashing, and the PWM port stops emitting waves."

This is because your DSP chip is interfered with by the power supply and the program is running away.

maychang

Finding the interference path is a very troublesome task.

Generally speaking, the power supply of your DSP chip is a suspicious path. Other paths include: the various input terminals of the DSP chip, especially the interrupt input terminal.

zqx123

Hi, first of all, thank you very much for your reply!

How bad does the 3.3V power supply have to be to cause the DSP to crash? When the DSP crashes, there is no obvious change in the 3.3V power supply on the oscilloscope? If we use an external DC source to power the DSP, will it still crash?

maychang

zqx123 posted on 2021-1-4 20:22 Hello, first of all, thank you very much for your reply! How bad does the 3.3V power supply have to be to cause the DSP to run away? I didn't see the DSP freeze on the oscilloscope...

A single large spike in the power supply to the DSP may cause the DSP to malfunction. The power supply may be normal for a long time after the error occurs. Your oscilloscope may not capture the spike on the power supply.

maychang

zqx123 posted on 2021-1-4 20:22 Hello, first of all, thank you very much for your reply! How bad does the 3.3V power supply have to be to cause the DSP to run away? I didn't see the DSP freeze on the oscilloscope...

"We use an external DC source to power the DSP, but it will still die?"

It is entirely possible. Even if an external DC power supply is used to power your DSP, the DSP may still be affected by your power supply (probably a switching power supply) through spatial coupling (there is distributed capacitance between your power supply and the DSP pins) and magnetic field coupling.

maychang

zqx123 posted on 2021-1-4 20:22 Hello, first of all, thank you very much for your reply! How bad does the 3.3V power supply have to be to cause the DSP to run away? I didn't see the DSP freeze on the oscilloscope...

It is common for the switching power supply to interfere with the control section (your DSP chip), especially if your switching power supply and the control section are mechanically located in the same metal box.

freebsder

If it runs away, we have to resort to trickery. This kind of interference is difficult to find.

chunyang

Pay attention to two parts, one is the power supply circuit, and the other is the clock part. The power supply circuit mainly relies on enhanced decoupling, including the use of properly arranged decoupling capacitors and magnetic beads to suppress the interference of the power supply circuit. The design of the PCB is also particular, especially the layout of the ground loop should be highly paid attention to. The clock part mainly relies on PCB design to avoid interference. If there is a long IO line in the circuit, you should also pay attention to the interference coupled by the long IO line. The solution mainly relies on capacitors and inductors and clamp bypass.

zqx123

Thank you for the guidance from the seniors. I have just entered the power supply industry and I will learn a lot from you in the future.

everlastingzero

Have you solved this problem? Do you have any experience to share?

394865392

I have the same problem, and I am also 28035. Has the OP found out the reason?