non functional board

[uzmeed]

Hi

I have a customized Kintex 7 board

Its power rail of 2.5V that is providing biasing to bank 15 and 16 are showing almost zero impedance to ground. And off course board is not functioning properly

How much likely is it that FPGA is broken

Regards

Uzmeed

[drjohnsmith]

If its from an authorised distributor  

   as near zero as make no difference.

Is it one board or multiple boards ?

What's the PSU for that bank ? 

   is it providing any voltage ?

[uzmeed]

Hi @drjohnsmith 

Yes it is from a reliable source.

The PSU is managed by power sequencer and regulators.

2.5 V terminal is showing 0.25V output .

Regards

Uzmeed

[drjohnsmith]

Is the power sequencer one of these programable ones ?

      Has it re programmed itself  ?

We've had a few designs that have had this particular feature. We no longer use programable power supply sequencers.

[uzmeed]

yes the power sequencer is programmable

But I dont know whether it programmable by itself I dont have much details of it 

Regards

Uzmeed

[uzmeed]

Hi @drjhonsmith

How can we know that either fpga bank is faulty or its supply is not working

Best Regards

Uzmeed

[joancab]

You probably have a short, that could even be in the supply (so you measure zero ohms) (By the way, you probably didn't measure impedance, but resistance). If just the PS failed (hard or impossible to tell), it's relatively easy to replace. But the PS may have failed because of another short.... so you may end up with the new PS failing again. Better to take the PS out and check again for shorts. If the FPGA is damaged, it's probably better to bin the whole board as its cost is usually less than FPGA + time/cost of replacing.

[drjohnsmith]

Well, how to tell if its the fpga or the PSU,

Easiest is can you isolate the PSU form the rest of the board, 

     If you have a series inductance in the PSU that makes it easy, else you might have to lift pin or cut track.

Other thing to look for is a solder short under the package,

     the way of sorting that is to X-ray , but that's not cheap and not guaranteed to find a solder bridge.

A short on a psu tends to generate a fair amount of heat ,

    you could try powering up the offending rail with an external PSU, and turn up its current output slowly

        a thermal camera will show the short location, hopefully before the IC / board catches fire.

Of course it could be that you have the PSU shorted to ground some where, 

    have you probed around with a multi meter ?

[uzmeed]

Yes I probed around the PCB using multimeter

On thing is that PCB has started drawing more current than before about 0.7mA

The banks of FPGA that are receiving proper biasing are working properly

I am able to program the FPGA as well

As the PCB is using the sequencer so I also debugged the power supply using PMBus that showed only one voltage as short.

Is it possible that few banks are OK and few are faulty

Best Regards

Uzmeed

[drjohnsmith]

Any thing is possible,

   but the probability of a faulty FPGA from a reliable supplier is vey low.

much more likely is a manufacturing fault, 

    solder blobs, internal layers miss aligned, et all.

The only way you are going to find out if its the FPGA , which is the lowest probability, is to take it off and test it in a production test jig which IM guessing you do not have.

The way to solve this is to exclude each option one by one,

   so you resistance meter, 

       is the Power for the FPGA shorted to ground ?

          If so the resistance will be less the closer to the short you are.

what is the assembly houses expertise with placing these parts ?

Its just a long detective job you have to do, no short cuts, just turn the handle, keep a log of all the tests and results you do, 

   and if you still have problems, post the tests you have done and the results and we can try to guide you more.

[uzmeed]

Hi 

The board had been running fine and we have been working with all interfaces 

Suddenly we were unable to grab data from two of its bank

Its is due to fact that supply to this bank is reading 0.174 V instead of 2.5V and resistance is 1.2ohms contrary to earlier where the resistance of this point was 800Ohms when it was functioning fine

All else reading fine in terms of resistance and voltages

I hope you can guide me further

Best Regards

Uzmeed

[bruce_karaffa]

 Is anything on the board getting hot?  As @drjohnsmith mentioned, a thermal camera is useful to find hot spots, but if you don't have one, just feel around carefully.  Sometimes an electrolytic capacitor will just short out.

[uzmeed]

I will get it checked and then let you know 

But my basic question needs your attention is it possible that FPGA is working while a few banks are blown up

Regards

Uzmeed

[joancab]

"is it possible that FPGA is working while a few banks are blown up?"

The problem is that you don't know the extent of the damage. It could only be the IO drivers, it could be more than that.

[avrumw]

But my basic question needs your attention is it possible that FPGA is working while a few banks are blown up

Yes - it is possible. (I agree with others - the FPGA is not the most likely failure, but) If it is a failure in the FPGA and the failure is confined to the VCCO and the I/O drivers of one bank or a couple of banks, then, assuming the failed bank is not the configuration bank, then it is possible that the FPGA would program normally and the other I/O banks would work. It's not necessarily guaranteed, but it is possible.

Avrum

[uzmeed]

Thank you all 

I hope I will reach to some result with your help 

Please advice if there is any way in software by which we can see that if it is the FPGA or power supply

Best Regards

Uzmeed

[drjohnsmith]

@uzmeed

You do seem insistent that its the FPGA

  where as we are all saying the same thing, its very unlikely to be the FPGA if it came from an authorised supplier.

Like all chips , FPGAs are susceptible to damage during assembly or testing. shorting pins to ground is a good one.

But a shiped FPGA is verywell tested during assembly, so the chances of you havign received a dead one are very very remote.

  The biggest likely source is your board or the way it has been tested.

i once had a board that was dead similar to this,

    turned out that someone had decided to do a mega voltage test on the internal PSU's on the board.

As I mentioned also, we have had quiet a few of these programable PSU systems decide to re program themselves 

We also have had boards that have lovely shorts under the package, when we chased turned out the assembler had had a problem on that board, and tried to fix it ..

These are much more likely scenarios than the FPGA being purchased dead,

 Please go and do the tests we have all been telling you to do, and come back with the results,

[bruce_karaffa]

You have a hardware fault on the board.  Until you find and fix it, software will do you no good.