UCD90160A: Re-sequence Issue

Hello

Could you take a snapshot of the system monitor when the 3V3 is low. with the system monitor I can see all the rail status which may give some hints

There are couples things to improve in the configuration

1. For the rail#12(0.9V), it does not have any fault shutdown slave. So  the resequencing in the fault response of the rail#12 shall not be enabled. For a single rail, restart(retry) shall be used instead of resequencing

#2. for the rail#16, it does not have fault response set. so rail#16 shall not be fault shutdown slave of any rail. It does not make sense have it. 

#3. under Global Configuration->Misc Configs-.Re-sequencing options. Rail#1 and Rail#16 shall not selected under the re-sequence rails mask since those two rails are not part of resequencing operation.

Regards

Yihe

HI Yihe

The connector is not installed on the board, and also the card is not in our Lab, we could not access the UCD90160A through TI dongle. I'd also like to see the registers of UCD90160A.

To the questions about improvement:

1. 0V9 is not used on this board, and it is divided from 3V3_PM, this power rail is used by UCD90160A only, if this power supply(3V3_PM) is gone, UCD90160A can't re-start the power rail. I agree in logic, it should be re-start instead of re-sequence. But it should be no difference for this application.

2. Rail #16(BP_PWR), is backplane power, there are two cards in the chassis, both of them could drive this power rail, it doesn't have fault response configuration, but we like UCD90160A to record the fault. Its If the power fault shutdown slave is not checked, UCD90160A will record the power fault when other rails failed. BP_PWR if from 3V3 on board.

3. Because rail #16 power fault shutdown slave is checked under other rails for the reason above, it need be masked here, and 5V mask could not be removed, we think there is no side effect, so we just mask 5V too, does it have any side effect?

I still like to know what the possible reason for this case, we could not understand it

1. Since 3V3 is low, why UCD90160A hasn’t re-sequenced the power rails per configuration?

2. What’s the possible reasons that UCD90160A disable 3V3 converter? the 3V3 dependency, 0V95 and 0V9 are fine.

Is there any configuration not correctly?

Thanks

Frank

Hi

I do not see anything special which may cause what you observed.. 

System monitor can tell the status of all rails. from there, i may get some hints why 3V3 stays off?

Do you have a way to duplicate this in your lab? is this at low temp or high temp environment? Will be any soldering issue due to temperature?

Regards

Yihe

Hi Yihe

I can't reproduce this issue in our lab, if there is a 3V3 fault, UCD90160A will generate a re-sequence when I simulate the 3V3 fault in Lab.

The chassis is in the house, I think it should be under room temperature. I will double confirm it.

For the soldering issue, even it's not soldered well, for example 3V3 enable signal(3V3_PS_EN), then UCD90160A should noticed the 3V3 gone, and I think UCD90160A already knew 3V3 gone, because these signals behavior

• Signal PMON_PG_LED_N is high, that means there should be one POL is not good,S
• Signal PMON_PORST_N, PMON_SYS_FPGA_GRST_N are low, that also means there shoud be one POL is not good

Since UCD90160A noticed 3V3 fault, why it didn't re-sequence the others? and all the other power rails should be gone too.

PS: I've simulated 3V3 fault, through E-load over current; or short 3V3 enable signal(3V3_PS_EN) to GND, or create a fake 3V3 fault on UCD90160A pin 2(3V3 monitor pin), UCD90160A will re-sequence for the fault above.

Thanks

Frank 

Hi Yihe

It's in house, and temperature is about 25 to 30C, it's not cold or hot.

Frank

Yihe,

We are seeing similar issues on multiple cards in multiple customer locations.  I do not believe we are seeing soldering or environmental issues.  This appears to be the UCD90160A not performing as we expect.  Perhaps we have not configured the device properly or could there be noise or incorrect design which is causing these behaviors?

We are seeing what appears to be two types of faulty UCD90160A behavior

1) Not re-sequencing the power rails when there is a out of spec power rail (has happened on different rails)

2) The UCD90160A appears to randomly shut down a power rail (as described in this ticket).

Regards

Jose Enciso

Sr. Dir, hdw eng

Hi 

Thank you for the updating of the temperature.

Based on descriptions, it is hard to know where the source of the fault is from? is 3v3V down due to its own fault or be fault shutdown slave of other rails' fault?

Have you able to remove access the board? 

Regards

Yihe