[SI-LIST] : VMEbus transceivers

O'Donnell, John (o'donnj@aydin.com)
Thu, 16 Apr 1998 17:23:40 -0400

SI Practitioners,

What is the state of the art in commercially available VMEbus
transceivers (i.e '16245-type bus transceiver in surface mount 48-pin
TSSOP packages)?

I have a VME-hosted 6U board that provides a bridge between the VMEbus
and a side bus implemented on the VMEbus J2 Connector rows A and C
using Futurebus technology (i.e BTL transceivers). Activity on the two
buses is asynchronous with respect to the other bus, meaning that the
VME bus transceivers can turn on to drive a master transaction to the
VMEbus at any time during receipt of a slave transaction on the side
bus. The ground lift, or ground bounce produced when the VMEbus drivers
turn on is sufficient to cause data glitches on the local bus outputs of
the side bus BTL transceivers. When the VMEbus driver turn-on happens
to coincide with the sampling edge of the side bus data strobe used to
capture side bus slave data the amplitude and width of the data glitch
is sometimes enough to be captured in the side bus slave data latches.

The board design was based on a heritage design which used 74FCT16245
bus transceivers to interface to the VMEbus. These devices produced
humongous ground bounce effects and serious data error problems on the
side bus. These have since been changed to 74ABT16245. This change has
produced a marked improvement in the side bus data integrity but there
are still occasional errors traceable to ground bounce effects when the
'ABT drivers turn on. Is there a better technology available for driving
a 21-slot VMEbus backplane with standard resistive terminations such
that the driver di/dt is restrained enough to result in a tolerable
ground bounce effect?

John O'Donnell
AYDIN Telemetry