Fwd: RE: 32 bit / 3.3V PCI slots?

[Donald Connolly <Donald.Connolly@matrox.com>]

>From: "Ingraham, Andrew" <Andrew.Ingraham@compaq.com>
>To: pci-sig@znyx.com, "'Breuer Thorsten'" <Thorsten.Breuer@khe.siemens.de>
>Subject: RE: 32 bit / 3.3V PCI slots?
>Date: Tue, 22 May 2001 11:44:36 -0400
>
>> I'm working on a 32bit PCI extension board for PC.
>> I was asked to design it as universal board for either 5V and 3.3V
>> signalling environment but I've never seen any mainboard with 3.3V 32bit
>> PCI
>> slots. So adding this feature doesn't really make much sense to me.
>> 
>> Any comments?
> 
>You WILL start seeing motherboards with 3.3V PCI slots, and when you do, any
>PCI expansion cards that are not Universal or otherwise 3.3V capable, will
>no longer be usable in these newer systems.
>
>Revision 2.2 of the PCI spec says that all new expansion boards should be
>Universal boards.
>
>In order to make a clean transition from 5V to 3.3V, which MUST happen
>because chipsets that are 5V-tolerant are increasingly difficult to make as
>the feature sizes decrease, all expansion boards on the market must be
>3.3V-capable, or expect to lose marked share.
>
>Regards,
>Andy
> 

Considering that 5V-tolerant devices are increasingly difficult to make, it seems to me the question for a board manufacturer should instead be whether to design a board to be Universal or 3.3V-only. It is a chicken and egg question, but building Universal boards may not be possible for much more longer and it is imperative that PCs with 3.3V slots become available soon, or for the 5V PCs to dissapear. Just looking at the evolution of 3.3 V buffers in recent FPGA families and we can see it going from full PCI 5V support to tolerance using an external resistor to no support at all. It is about the same story with ASICs.

Ignoring the clamping issue for a moment, I started to wonder why 5V-tolerance is a requirement for today's PCI devices. Virtually all PCI devices shipping today use 3.3V on their output buffers, and it has been the case for several years. So the only things that prevent a 3.3V-only board to work in today's PCs is the 5V connector key and 5V used to pull-up the control signals and interrupts on the motherboard. Just for curiosity, I checked the voltage on these signals on several PCs using 5V connectors and I found 5V as expected, but also some 3.3V and even a couple of machines with 4V pull-ups! That would be just about the limit of what 3.3V devices can withstand. Why would motherboard manufacturers use 4V pull-ups on PCI signals? Considering the presence of pull-ups on all signals that are not driven by a 3.3V device, is 5V-tolerance a requirement for today's PCI devices?

Donald