LP8764-Q1: shutdown at high temp

Hi Ryan,

Could you clarify which OV is set? Are you referring to VCCA OV? 

Do they have scope shot about the event happening at the time the all devices shuts down. Also regarding interrupts i'm suspecting there should be interrupts than this one. 

What is LP8764-Q1 full part number? Is it LP876411B4-Q1?

Br, Jari

Jari,

Yes, it is VCCA OV.

It is LP876411B4-Q1.

Below is all register dump.  TPS6594 all registers return 0. LP8764 return valid data.

registers dump.docx

Below is scope capture. EN_3V3IO_LDSW，VCCA_3V3，VDD_CORE_0V8 are LP8764 output.  VDD_CPU_AVS is TPS6594 output. Both silicon shutdown at the same time. 

Below are other outputs of TPS6594. They are all shutdown at the same time.

Hi Ryan,

Having everything zeroes from TPS65941213-Q1 is not indicating good. Even when device shuts down there should be communication open to device to read registers. Also I had a little hard times to read register dump. Would you have it in register address = register data format. 

Also based on figures TPS65941213-Q1 starts shutdown earlier. Not sure but also looks like rails are coming down with pulldown enabled. Meaning no switching which means severe error has triggered shutdown. Is the time scale 10ms or what in the figures.

I'll check from TPS65941213-Q1 his opinion about this shutdown event. 

Br, Jari

Jari,

Sorry, I find last register dump has only 6 bits for every register shown. Pls see below register dump for LP8764.

The 1st pic is 10ms time scale. 

The 2nd pic is 4ms time scale.

The 3st pic is  8ms time scale.

If LP8764 OV disable, the problem won't happen. And ABA swap follows LP8764. Customer wants to send this pcs LP8764 for CQE process.

Now they report 5 more pcs LP8764 (among 15Ku) has the similar problem. But these 5 pcs LP8764 only show this problem once during high temp test and cannot be reproducible any more.

I want to send this pcs LP8764 for CQE process tomorrow. Need your comment on this. 

 LP8764 register dump.xlsx

Hi Ryan,

Thanks for the data. Now I can see all registers. From there still quite interesting that TPS6594-Q1 cannot be read but LP8764-Q1 can in the error case. 

Scope shots are ok but there should be one same signal in every shot to align timing accurately. 

With these 5 new pcs are those able to recover back to normal operation after event? And are all previous 3 pcs having same reproducible symptom every time. No restart happening at all. I mean system shuts down from normal operation and no restart?

I think CQE is needed but let's don't rush to that yet before we try to figure out issue.

Br, Jari

Hi Ryan,

Few clarifications:

1. After first cold boot (VCCA rising from 0 V to 3.3 V) have all interrupts been cleared before system goes to continuous operation state?
2. Regarding A-B-A swap, is the suspected LP8764-Q1 moved to expected working board? After discussing with design and systems, TPS6594-Q1 should be also accessible after event. Not only LP8764-Q1.
3. Could you request register content of both devices before error happens?

Br, Jari

Hi Ryan,

1. Also could they change VCCA UV/OV thresholds from current setting to max and check operation.
2. VCCA_GRP_SEL from MCU group to No group
3. Scope triggering to nINT falling and very good zoom to VCCA to see what happens in VCCA at that exact time. Voltage scale hundreds of millivolts or tens of millivolts. 
4. Picture from top mark.

Br, Jari

Jari,

“With these 5 new pcs are those able to recover back to normal operation after event?“

Ryan: The problem only happen once during high temp test. After power shutdown, it cannot back to normal. After they power cycle system, the problem cannot be duplicated any more.

"And are all previous 3 pcs having same reproducible symptom every time. No restart happening at all. I mean system shuts down from normal operation and no restart?”

Ryan: Only one silicon is reproducible every time. Another 2pcs are among those 5 new pcs. Correct, system shut down and no restart.

"After first cold boot (VCCA rising from 0 V to 3.3 V) have all interrupts been cleared before system goes to continuous operation state?"

Ryan: I will let them download register after cold boot and send to you.

"Regarding A-B-A swap, is the suspected LP8764-Q1 moved to expected working board? "

Ryan: yes. Phenomenon follows this pcs LP8764-Q1.

I will do other experiments you suggested soon.

Top mark: