[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: PCI On-card Trace Length
- To: "PCI SIG Developers Forum" <pci-sig@znyx.com>
- Subject: RE: PCI On-card Trace Length
- From: "Ingraham, Andrew" <Andrew.Ingraham@hp.com>
- Date: Fri, 28 Jun 2002 17:42:46 -0400
- Resent-Date: Fri, 28 Jun 2002 14:47:43 -0700
- Resent-From: pci-sig@znyx.com
- Resent-Message-ID: <DAD0D.A.5AF.TiNH9@electra>
- Resent-Sender: pci-sig-request@znyx.com
- Thread-Index: AcIdOmLI+TCLmuebSfi1TPCLxGyvpgBoiyTR
- Thread-Topic: PCI On-card Trace Length
This is an interesting question. I don't know if you'll find someone to give you an "electrical justification." I'll offer my thoughts; just remember that it's my opinion, nothing official from any past or present employer or the PCI SIG.
Keep in mind that PCI does not dictate a limit on the number of devices or connectors. What determines that number, may be what the system (motherboard) designer can reliably make work. It is her/his responsibility to assure that both the lower-32 and upper-32 halves of the bus work.
A consequence of allowing the upper-32 traces to be longer, is that there might indeed be fewer 64 bit connectors than you could have of 32 bit connectors on the same bus. And you might see motherboards with one or two 32 bit slots on the end of a 64 bit bus. Or one with an on-board 32 bit device on the motherboard.
Can you somehow take advantage of a 64 bit motherboard to make your traces longer on the lower-32 half? I don't think so. It wouldn't be "PCI compliant," because PCI compliance requires 1.5 inch or shorter traces on those signals, on any expansion boards, no matter where used.
Even if you ignored the "PCI compliant" question, the motherboard designer might have counted on your card having no more than 1.5 inch long traces on the lower-32 half. On some systems it might not matter. On others it would.
Regards,
Andy
> ----------
> My question involves PCI 2.2, and the 1.5 inch limit for trace lengths
> on PCI add-in cards for 32 bit buses, and the 2.0 inch limit for the 64
> bit extension signals. Electrically speaking, why is the 64 bit bus
> trace length allowed to be longer than the 32 bit bus signals.
>
> Physically, one can see that the traces must be longer just to reach all
> of the connectors, but what electrical justification allows the longer
> length for just that portion of the bus. Are there less taps allowed on
> 64 bit buses, similar to less taps allowed at 66 Mhz?
>
> Does that mean that if there will be a limited number of add-on cards
> (and slots) allowed per PCI bus, that the 32 bit bus trace length can be
> lengthened to the same limit as the 64bit bus, and still be "PCI
> compliant"? Many thanks
> - c.neumann
>
>