1. Home
  2. Field-Programmable Gate Arrays
  3. Cyclone II short-circuit failure mode

Cyclone II short-circuit failure mode

A EP2C5Q208C8 in one of our boards has gone short circuit on the 3v3 supply. We manufacture this board in quantities without any problems, so we would like to investigate why this could have happened, mainly for peace of mind.

We removed the chip, and there is less than 70 ohm between all the vccio4,vccio2 pins and GND. So we suspect the chip must have fused internally somehow. However, vccio1 shows no short.

The chip 's 3v3 and 1v2 are supplied from two low power regulators (LM317L variety) which can supply no more than 100mA each, so we are a bit puzzled.

We suspect two possibilities:

1) A momentary short between one of the I/O pins and the main 5V supply (which can supply high current drive)

2) A software bug might have caused eight of the Cyclone I/O pins to output directly to the output of a 245 CMOS buffer (also fed from the

3v3 current limited supply) . With both FPGA and 245 pins outputting straight into each other. Not sure which way round (possibly cyclone=low 254=high) This situation may have lasted for a long time.

Could anybody with experience of these devices suggest whether any of these could have caused the fault??

TIA Ted

Reply to
edaudio2000
Loading thread data ...

Were you using an FPGA? Often the manufacturing faults can get in the way even if the device initially appears to work. The balls underneath can be shorted or malformed to where they're nearly shorted. The ROHS solders aren't perfect and in cases where there's almost no clearance, dendrites can still be an issue; it's less of a problem for good clearances and clean boards in limited humidity if I understand the process properly.

If it's a single instance of a problem with an FPGA, chalk it up to "probably" a manufacturing issue unless you have 100% x-ray inspection through a reliable QA process. If you have the problem more than once, look a little further. Are ESD handling procedures strictly adhered to?

- John_H

Reply to
John_H

Thanks John,

I am not too convinced this was a manufacturing or solder short problem. The FPGA chip is fed from a 100mA limited power supply, and cannot imagine how any combination of solder shorts would have caused the chip to fuse so solidly internally.

In any case, the board had been working for a while when it failed suddenly during tests. So I imagine some form of external "spanner in the works" may have been responsible. There is a 5V line on the board (high current capacity) but that's all.

I suspect an ESD failure might have caused the chip to stop working, or make some of the I/O pins become not operational, that's all.

The puzzle is that the chip went "fused short" i.e. less than an ohm between the 3v3 vccio4 and gnd (The 1V5 supply is OK).

And quite a solid short it is, If I apply 1V to vccio the chip draws a lot of current and gets warm..

This is all very odd...It would be interesting to know what might have caused this.

Regards

Ted

Reply to
edaudio2000

There are failure analysis companies around. We have used maser. You will probably have to send them a good and a bad device plus some socketted fan out pcb.

Curve tracing supplies and signals will show which if any pins have been damaged by ESD.

Diagnosing to an on device issue can be done by removing the package surface exposing the passivation layer of the silicon. Cove the passivation layer with liquid crystal. Apply voltage and watch for changes of colour under a microsocope.

It is a tricky business at the best of times.

Reply to
Andy Botterill

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.