ground bounce

[jdharmawan@hns.com (johannes dharmawan)]

PCI experts,
I'm testing a PCI target-only card using Zappa E motherboard with Triton
chipset and writing 0xFFFF_FFFF and 0x0000_0000 continuously to the local
SRAMs. In other words, all 32 bits of data toggle continuously. I write to
sequential addresses to the SRAMs and the PCI Triton chip is set to burst mode.

What I observe is that during data phase in burst transfers, most of the input
signals to my PCI card (e.g. frame*, irdy*, cbe*[3:0]) are glitching from low
to high. I'm suspecting this is some sort of ground bounce effect (I also try
to
write constant data and don't see the glitches).

The worst spike on these signals goes as high as 1.6 V (still below CMOS
threshold). I am unable yet to determine the setup time from when these
signals settle down to the next rising edge of the PCI clock.
Right now, this doesn't cause functional problems on my card.

I also try to run traffic on my PCI Ethernet Card to see whether this
happens to other PCI cards or this glitch is caused by my PCI card.
I don't have a special code to simulate the same testing scenario (write all 1s
and 0s continuously). So, I use 'ftp' and run some traffic on the Ethernet card
and I sometimes capture a bus cycle with frame* glitching momentarily.

Has anyone seen the same behavior? I would like to get your inputs. I would
like
to know what needs to be done (either to the motherboard or to my PCI card) if
this is a problem.

Thanks in advance.


-- 
johannes dharmawan
e��