> receiving device. If a
> single-ended device, such as a CD-ROM, is connected to a
> differential host
> adapter (e.g., AHA-2944UW), or a differential device (e.g., a
> hard drive with a
> D suffix)is connected to a single-ended host adapter, such as
> the AHA-2940UW,
> *irreparable damage can be done to either or both devices*.

This is simply corporate CYA, is misleading, and might be considered
fraudulent; no HVD device that complies with SPI-2, and no LVD/MSE
device that complies with SPI-3, can cause any kind of damage to any
SE device, because they are not allowed to drive any signal other than
DIFFSENS until the mode of the bus has been determined (DIFFSENS is
driven with 1.3 +/- 0.1V at 5(nominal)/15(max)mA by an LVD/MSE device,
and pulled up to +5V(nominal) through  a diode-blocked 1Kohm(typical)
resistor by an HVD device).

> Emphasis in the data above is mine.  Connecting Differential
> and SE devices on
> the same bus is the IT version of Russian Roulette.  Most
> times nothing will
> happen, but every once in a while. . .

If any damage occurs with a device advertised as SPI-2 or SPI-3
compliant, the manufacturer will be liable for the damage, since the
device was being operated within the parameters of the applicable
specification. Mixing any combination of compliant SE, LVD, and HVD
devices on a bus is less "Russian Roulette" than mixing two brands of
detergent in your washing machine. There are clearly-specified
requirements to ensure that the SCSI devices work together; there's
nothing preventing the two detergents from combining to create
something fun like sulfuric acid that eats your clothes.

It's unfortunate that these false "warnings" against mixing SCSI
devices continue to be propagated.


Steve Dirickson   WestWin Consulting
[log in to unmask]   (360) 598-6111