Is the maximum voltage of high voltage differential SCSI documented somewhere? All I can find in various source like Wikipedia is that it is "greater than 5 volts" ... but how high can it actually be?
I have seen what HVD can do to a single ended SCSI controller ... burns the copper traces right off the circuit board, so there is definitely a lot of power on that bus. But just how much power?
HVD in the SCSI standards was 'according to RS-485', which allows up to -7V and +12V on the differential wiring, though limits the difference signal to the (0V, 5V) range. There is a separate line in SCSI standards for identifying HVD, LVD (low-voltage differential) and single-ended (SE, based on each pair of wires being one dedicated ground, and one open-collector driven). Unless the identification is ignored, a HVD or LVD driver will not engage (the drivers just stay in high-impedance state), so damage is rarely an issue, though function is.
A typical offering used only +5V and ground, like this one AHA3944UWD and would offer ONLY all-HVD device bus function (and power-off if a LVD or SE device were attached). The difficulty arises if an HVD driver attempts to lift the always-grounded second wire of an SE bus device, or applies high overvoltage to an LVD input. A typical driver for HVD (SN75176B) is capable of putting 150 mA into its output. If wired to SE-type ground/common wiring, 25 or so driven wires, that means an SE ground current of 3.75 amps could flow. That'd be nearly twenty watts of fault power dissipated in wires or chips.
