UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Observer kalicutt
Observer
1,066 Views
Registered: ‎12-29-2015

Custom Hardware Zynq Ethernet Failure

Jump to solution

Hello,

I have a few boards I am prototyping.  Two of them have had to go through rework that required re-heating the board twice just next to the Zynq chip, as there were BGA components that needed changing.  The last board I have not done any re-heating besides some basic soldering.  All 3 of the boards have the same hardware components.  It is also important to note, the Ethernet PHY exists on a separate board, and so I am using the same one for every Zynq.  With the exact same boot image and software, the boards work identically except for Ethernet.

 

Board 1&3: Ethernet does not work properly while booted in Linux.  Pings to anything other than a local IP do not reach destination.  ARP responses are not forthcoming.  Peripheral test (Xilinx SDK project 2016.4) says there is an error when interrupting Ethernet for 1, and hangs on Ethernet interrupt on 3.

Board 2: Ethernet works properly.  With Linux booted, I can ping it from another machine.  Peripheral test works fine, Ethernet passes interrupt test.


Is this enough information to conclude that boards 1 and 3 were damaged due to the heating, and that the internals of the Zynq are irreversibly harmed?  Is there anything else I should do to debug this problem?  Thanks.

 

EDIT: Some extra information, all of the other functionality of the hardware has been well characterized.  Power supplies, clocks, resets, and all of the other interfaces are working properly, on all of the boards.  I am confident it is not a problem external to the Zynq.

0 Kudos
1 Solution

Accepted Solutions
Scholar austin
Scholar
1,598 Views
Registered: ‎02-27-2008

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

I would just remove, clean, re-ball, and replace both devices,

 

You already know it isn't a design problem, so that leaves an assembly problem.

 

If re-work doesn't resolve, then it may be a damaged (bad) component (least likely).  Then you are off to removing and replacing those.

 

There really isn't much you can do after checking voltages, observable signals, etc.  If the power supplies are all good, and signals look good (those you can see), it leaves those you can't see, which means an assembly issue.

 

 

Austin Lesea
Principal Engineer
Xilinx San Jose
0 Kudos
6 Replies
Scholar austin
Scholar
1,031 Views
Registered: ‎02-27-2008

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

Damage to the device itself is unlikely during rework.

 

If the correct heat vs. time profiles are used, and there has been no moisture uptake by the board and package.

 

The result of re-work near by is usually opens or shorts from the FPGA bumps to the board (use x-ray inspection to locate, or remove, re-ball, and remount the device --- good rework technique can remove/replace devices multiple times without damage).

 

High humidity storage (water uptake) may cause 'popcorning' which may damage the device package beyond repair/rework.  De-humidifying the board prior to rework is good practice (low temp bake for 24 hrs).  If you do not know the temp profiles, or pre-bake rules, contact a good manufacturing house, or your local Xilinx sales office to get information and referrals.

Austin Lesea
Principal Engineer
Xilinx San Jose
Observer kalicutt
Observer
1,026 Views
Registered: ‎12-29-2015

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

Hey Austin,

Due to time constraints, these were done roughshod using cheap heat guns over the course of ~30 minutes, not professional ovens or temperature controlled environments.  Or humidity controlled for that matter.

Does that change the likelihood of there being damage?  I don't think pins being disconnected underneath is a problem since even the internal self-test is failing for PS7 Ethernet.  Doesn't rule out shorting though, I suppose we can get the boards X-rayed again.  Should this type of rework be avoided for anything nearby the Zynq?

0 Kudos
Scholar austin
Scholar
1,019 Views
Registered: ‎02-27-2008

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

Xilinx can only guarantee that which follow our procedures,

 

You may have saved money on rework to lose more in ruined equipment.  After what you revealed, I have no way to judge if the devices are permanently damaged or not.

 

It has nothing to do with Xilinx devices:  any 'modern' devices are likely to act the same (minimum pitch ball spacing, organic substrates, 28nm technology node, etc.).

 

 

Austin Lesea
Principal Engineer
Xilinx San Jose
0 Kudos
Observer kalicutt
Observer
1,014 Views
Registered: ‎12-29-2015

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

I may have not presented this correctly, what I said about the heat gun was what was done on board #1, which had shorts under a very small pitch BGA component and so we had to remove it due to time constraints and debugging a new design.  For board 3, there was a short found after initial population, thanks to X-rays.  Since it was still with the assembly house it was removed then replaced with the proper methodology.  So that means I have a board I did with a sloppy re-work, and one with the proper procedure.  Both are showing similar symptoms (different in their reaction to the peripheral self test in Xilinx SDK).  I think my question is not so much is my hardware damaged, but is there any further testing I can do to get information on the problem...  If there isn't then I will just have to give up on these boards.  It seems unlikely both boards should fail in almost the same way in two different scenarios.  See below the self-test UART output for the boards (Board#2 finished all of these perfectly fine).

 

Board#1:

Hello World
---Entering main---

Running ScuGicSelfTestExample() for ps7_scugic_0...
ScuGicSelfTestExample PASSED
ScuGic Interrupt Setup PASSED

Running XDmaPs_Example_W_Intr() for ps7_dma_s...
Test round 0
XDmaPs_Example_W_Intr PASSED

Running Interrupt Test for ps7_ethernet_0...
Error detect phy

 

Board#3:

---Entering main---

Running ScuGicSelfTestExample() for ps7_scugic_0...
ScuGicSelfTestExample PASSED
ScuGic Interrupt Setup PASSED

Running XDmaPs_Example_W_Intr() for ps7_dma_s...
Test round 0
XDmaPs_Example_W_Intr PASSED

Running Interrupt Test for ps7_ethernet_0...

0 Kudos
Scholar austin
Scholar
1,599 Views
Registered: ‎02-27-2008

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

I would just remove, clean, re-ball, and replace both devices,

 

You already know it isn't a design problem, so that leaves an assembly problem.

 

If re-work doesn't resolve, then it may be a damaged (bad) component (least likely).  Then you are off to removing and replacing those.

 

There really isn't much you can do after checking voltages, observable signals, etc.  If the power supplies are all good, and signals look good (those you can see), it leaves those you can't see, which means an assembly issue.

 

 

Austin Lesea
Principal Engineer
Xilinx San Jose
0 Kudos
Observer kalicutt
Observer
1,006 Views
Registered: ‎12-29-2015

Re: Custom Hardware Zynq Ethernet Failure

Jump to solution

Thanks Austin,

I just talked to the assembly house and actually they told me they do much the same thing we did in house here, which is use a hot plate and a heat gun.  I think I will just call it then.  Either a re-population of the (probably entire) board with new parts or assembling new ones seems to be the only choice.

0 Kudos