Re: 3.3V PCI

[Richard Blackborow <richard@alacritech.com>]

Paul,

Anyone building systems to the Rev 2.2 PCI spec must supply 3.3V to all PCI
slots (regardless of whether the slots are 3.3V or 5V signalling).  So your
issue boils down to whether you care about compatibility with systems
implemented to PCI Rev 2.1 and earlier, where it is not required to supply
3.3V to the slots unless the system has 3.3V signalling (which it will if it
has any 66MHz slots).

It's a tough decision to decide to run everything from 3.3V, particularly if
you think you have a market for your product on 33MHz legacy systems.  If
your product really needs a 66MHz slot, then you don't have an issue, since
66MHz is 3.3V signalling by definition, and therefore the system will supply
3.3V at the slot.

One thing to check is that all your 3.3V components can tolerate the +/- 9%
spec on the PCI 3.3V supply.  I'm running into this problem right now (see
my post from 1/11/01).

Richard Blackborow


Paul Capes wrote:

> Hi all,
>
> We are developing new 64 bit/66 MHz PCI boards on which most of the
> components rely on 3.3/2.5V supplies. Due to the speed and density of
> these products, I am very reluctant to derive these supplies from the 5V
> rail. However, I cannot find any motherboards with more than two 3.3V
> PCI slots. Would anybody care to comment on the approach they are taking
> with their current product development (requiring 3.3V supplied by the
> host).
>
> --
> Paul Capes
>
> Engineering
> Interactive Circuits and Systems Ltd.
> 5430 Canotek Road
> Gloucester, Ontario
> Canada K1J 9G2
> Tel (613) 749-9241
> Fax (613) 749-9461
> Email pcapes@ics-ltd.com