[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: pci compliant devices?



Stuart Adams wrote:
> 
> > The way it works is that FPGA pads can be configured to support
> > many different signaling standards.

The chips with banks of I/Os (Virtex, Virtex-E and Spartan-II) will
allow other I/O types in the same bank, provided there isn't a conflict
(e.g. HSTL and PCI can't share the same bank).
 
>  This is not true for 9 out of 10 FPGA/CPLD's which claim
>  PCI compliance. Most of these have a single output mode
>  (3.3 volt) and no clamp diode. In small print there is
>  often a disclaimer something like "outputs are PCI
>  electrical specification compatible (no internal clamp
>  diode)"

Xilinx has the 3.3 V clamp diodes on all devices for which we have
claimed 3.3 V PCI compliance. In the first devices, the XLT family, we
specially bonded out the clamp diodes for the user to attach to 3.3 V.
In subsequent families (XLA, Spartan-XL, Spartan-II, Virtex, and
Virtex-E) we make the clamp diode user programmable. In some cases (XLA,
Spartan-XL) you turn it on for all I/Os. In newer chips it's selectable
on a per I/O basis. Spartan and 4000/E/EX devices never supported 3.3 V
PCI. 

Universal PCI (3.3 V and 5 V PCI) can be done with Spartan-XL, XLA,
Virtex, and Spartan-II devices. To do this you have to load a different
bitstream depending on the PCI bus voltage (Vio). Virtex-E does not
support 5 V PCI. 


Jim McManus
Xilinx PCI Applications Engineer