UCD9090A: UCD9090A

Hi Hoangyen, 

We will check and reply to you. 

Thanks

Qian 

Hello

UCD9090A supports peaking log which is to store the maximum monitoring voltage/current/temperature into the non-volatile memory.

If you have those values, those rails have been to such high voltage in the past.

Regards

Yihe

Yihe, 

While the UCD9090A has the ability to store max monitoring voltage/current/temp into the non-volatile memory, I don't think I really see those high voltages on my board.  Those seemed to be erroneous readings in the peak logs.  We are seeing 25% higher than nominal on 12V, 3V3, 2V5 rails, and 66% on the 1V5_CPU rail.  With such high peaks, parts could get damaged. 

Are you aware of any issue that I can get erroneous readings?  

Yen

Hello

Do you decode the peak value yourselves?If so,  Do you try Fusion GUI to see what the value are reported?

We are not aware any similar issue.

Regards

Yihe

Jihe,

Our SW Engineer decodes the peak value for me.  Below is his formula for your reference:

 **********************

I read register LOGGED_PAGE_PEAKS 0xED page by page

Then I use the following algorithm to get the voltage and/or current values.

Voltage uses linear16

Current uses linear11

Linear16: register value is mantissa

    exp = ~(vout_mode | 0xE0) + 1; 

    f_exp = (float)exp;

    smnum = pow(2, -f_exp);

    result = (float)mantissa * smnum;

Linear11: register value is mantissa

result = (float)(mantissa & 0x7ff) * pow(2, -(float)( ~((mantissa >>11) | 0xE0) + 1));

 **********************

The Fusion GUI reports correct voltages for all rails.  Only the PMON logs reported high peaks.

OK, you are not aware of any similar issue.  Thank you for your help.

Yen

Yihe,

Let me consult with our SW Engineer on how we decode the peak readings along with your formula above.

I spent more time on the Fusion GUI.  As mentioned before, the Fusion GUI reports correct voltages for all rails while "polling".  I also confirmed the correct voltages for all rails with either scope or multimeter.  However, on Fusion GUI, under "Status" and "Peak Readings", Fusion logged all the high peaks, identical to how we retrieved from PMON logs in our SW.   I have attached 2 files for your reference.  I circled the fourth rail for simplicity.   We have high peaks on all except Rail #5 on the first PMON.  

We are on the same page that at one point, those rails have been recorded at such high voltage, and parts are still functioning.

Again, could we somehow getting erroneous readings from Fusion tool?

Thank you, Yen

Hello

Have you checked the logged fault under status? are there any OV(OVER_VOLTAGE) faults reported on those high peak rails?

Have you had the fault response set for OV event? if so, rails shall be off due to a OV fault.

Please share you project file so that we can review.

Regards

Yihe

Yihe,

Yes, I have checked the logged fault under status.  There is no fault.  

We have enabled OV event.  

Please find attached our project file.  

YenUCD9090A @ PMBus Address 115d Project_For TI.xml

Hello

Have you ever clear the fault log or peak log? How long has the device being run?

If you clear the peak log, will the similar peak value be back? I checked the project file and they seems ok except that you do not have fault response set. If there is any OV or UV faults, device will continue running.

Regards

Yihe

Yihe,

Yes, I have cleared the peak log, on a different board.  I have not cleared on this board yet.  The device on this board has been running since June of 2018.  

Once I clear the peak log, I will not see the peak again.   Regarding the OV or UV faults, we will get the logged faults if any fault occur now, but nothing on that very first record.  In the project file that I sent you, under Global Configuration,you will see that we have enabled all Faults.  This is how we configure it.  Would you check?  Below is cut/paste as well.

Hi

 Normally, if a voltage is high than the fault threshold, a OV fault shall be logged.

In your system, there is no any fault logged but with unexpected high voltage log. I am wondering whether the fault log had been cleared in the past somehow. otherwise I can not explain.  Have you clearer the fault log on the device which has high peak value?

I know you have the fault log enabled. but what I meant is the fault response which is to how the UCD9090A shall behave when a fault was detected.

in your system, this is not configured.

Regards

yihe

Yihe,

Yes, agreed, if a voltage is higher than the fault threshold, an OV fault shall be logged.  Somehow, we are not getting any fault in Logged Faults on these peak voltages. Therefore, we think these are erroneous readings because we have seen OV fault logged on real OV events.

As for whether the fault log had been cleared in the past somehow, I can’t tell.

I am working on 2 boards exhibiting high voltages logs.  I have cleared the high peak logs on board #1, and the Peak Readings in Status report normal voltages now. I have not cleared the high peaks on board #2 yet.

Regarding the Fault Responses, you are right that we did not configure the Fault Responses.  We are able to get OV faults, UV faults, etc with all Faults enabled in Global Configuration.  What is the difference between setting in Fault Responses and Global Configuration?  Why isn’t what we are currently configured efficient?  We will look into the Fault Responses but we’d need to have more information from you. We’d definitely interested in improving our PMON monitors.  Please advise. 

Yen

Hello

now you have cleared the peak log on board #1. could you let it run and not clear any fault and peak log to see whether the high peak will be back?

Fault response and Fault enable(at Global configuration) are different functions. The fault enable is to enable the logging for the fault event so that application can retrieve the fault information afterwards

Fault response is to set the action taken by the device when a fault event is detected. For example, if any rails have a OV/UV faults, you want to shutdown all/partial rails to prevent any damage, the fault response is the way to help you.

Hope this help.

Regards

Yihe

Yihe,

On board #1, after I have cleared the peak log, I did let it run and monitored the rails for a very long time, have not seen any high peak!!  These seem to be just a one time, erroneous readings, which stayed in the Peak Readings until I manually clear them.  I think we have exhausted the investigation and not sure what else you can help with.  Do you agree?

Thank you for the explanation on the Fault Response and Fault Enable functions.  I'll follow up with team internally. 

Yen

Hello

My suspicion is that the fault log was cleared somehow. but at this point, we have to wait the problem back to further investigate.

Please let's know.

Regards

Yihe