01-27-2017 06:03 AM
I have a KCU105 that unboxed for the first time a few days ago (rev 1.1). I ran the BIST and everything on the board ran correctly. I then have been using in the PCIe application and ran test in a PC for about 48 hours which is performed well. However, at the end of the test the card was no longer responding correctly. I removed the card from the PC and noticed that on the bench that the Power Good LED will go red after about 30 sec and that the MGTVCC rail was the reason for this. As noted, on initial power up all power appears correctly, but after some period of time it fails. The longer the board is off the longer the duration that the power stays good. Which agrees with that I found when I further investigated and found that it appears that the temperature fault is tripping on U135. Also, the BIST application no longer runs successfully.
I have examined various post and have tried reprogramming the power controller using the PowerTool with the Rev 1.1 XML file, which did not solve my issue. I would appreciate any help and on what may have happened and how to resolve this issue.
Has anyone else experienced this issue or is there a problem with these boards after a period of time?
01-27-2017 10:51 AM
I would like to add that using the Maxim PowerTool the fault listed on the MAX20751 used for the +1.000 volt MGTVCC bus appears to the a OVP_FAULT. Any ideas on what this may be caused by?
01-27-2017 10:30 PM
Due to power spike or some other unexpected reson Maxim power controller configuration program might disturbed. Please try to reconfigure as per below Answer Records (AR) information
To do the above you need Maxim power controller Dongle. You get that as per https://www.xilinx.com/support/answers/62629.html
01-27-2017 10:32 PM
After doing the above steps if still problem exists please check as per https://www.xilinx.com/support/answers/63739.html to find out any board issue exists
01-28-2017 01:05 AM - edited 01-28-2017 01:14 AM
01-28-2017 06:10 AM
Thanks for the reply. I did try the other day to reprogram the power controller again as it was suggested in other ARs. However, to allow for the possibility I did go ahead and retry it again following the steps for my revision (1.1) in the provided PDF file. This did not solve the issue. The +1.0 MGTAVCC rail will go offline about 2 minutes into operation and appears to be a OVP related issue.
I will seek at this point to get the board replaced as this should not be happening to a new board.
03-09-2017 10:00 AM
I have the exact same problem with my KCU105 board. After power-up, the power good LED is GREEN for about 30 seconds and then goes RED. When I check the system controller during power-up, I can see that MGTAVCC is 1.0V at first, then it drops to 0.49V. All other power supplies appear normal. I have also been using the board for a PCIe application, however via the HPC FMC connector.
I have also found another person with a similar issue. He sees MGTAVCC at 0.34V on his KCU105 board.
If anyone has found a solution, please let us know.
05-05-2017 07:06 PM
Just to help anyone with the same problem and to provide an update - in the end, I applied for an RMA and it was approved. The replacement board has been working well and I have had no issues with the power supplies.
01-12-2019 07:32 AM
I have a very similar problem on my KCU105. MGTAVCC is no longer working on my board. I downloaded a design to test a fiber connection between the two SFP+ cages. The design worked well, but after a few hours, the design didn't work anymore, and I saw the LED DS3 glowing red. Further, the LED D11 (MGTAVCC_PGOOD) was dark, while all other power good LED's were glowing.
With an oscilloscope, I measured zero voltage at MGTAVCC. I found out that the voltage on the TS_FAULT pin at the slave IC VT1697 was zero, although there was not fault condition notable. The power supplies VDD, VCC, VDDH are correct, VX has no short, and there was no over temperature condition. As long as TS_FAULT is low, the Master IC MAX20751 will keep in shut down mode.
So it looks like the VT1697 is defect, but how can this happen? I didn't touched the board during operation!