Re: Bounce from 5V to 3.3V signal switching

[Andy Ingraham <ingraham@wrksys.ENET.dec.com>]

Tom Wilson <tom_wilson@qmail.newbridge.com> asked:

> I have the following question. If you have parts on a PCI bus that operate at
> different voltages then you could potentially experience a significant bounce
> (to as low at 0V) when a 5V part stops driving the bus and a 3.3V part starts,
> switching a signal from 5V to 3.3V.  How would one handle this bounce?  Is it
> a problem in the PCI spec? 

Irv Negrin <IDN@nms.com> replied:

> You should never have 3.3V and 5.0V devices on the bus at the same
> time.  The spec defines separate 3.3V or 5.0V signalling environments
> where all cards that can be plugged in must be compatible with this
> environment.  3.3V signalling cards will only plug into 3.3V connectors. 
> 5V cards only in 5V connectors.  They are keyed differently.  Universal
> cards can plug into either but must automatically be compatible to the
> environment by connecting it's PCI VDD reference pins to the VIO voltage
> input.  When plugged in a 5V slot, VIO is 5V.  When plugged into 3.3V
> slot, VIO is 3.3V.   

Not so fast.  You can have 3.3V-powered devices on a PCI bus using the
5V Signaling Environment.  I suspect that Mr. Wilson is referring to a
3.3V device designed for the 5V Signaling Environment; they are not
that uncommon.

In answer to the original question, having the bus switch from 5V down
to 3.3V is not that much different from switching from 0V up to 3.3V.

PCI allows 10ns switching time for the voltage to settle to a valid
level again, whether a wire is going from high to low, low to high,
or high to high makes no difference.

Regards,
Andy
èÕ