One is a 430TX mobo made by some defunct company and doesn't seem to
have any documentation available for it on the web. The other is an
FIC PA-2007 (VIA VP2 chipset) that uses a 5V-only Analog Devices brand
chip to convert between RS-232 and TTL levels and contains its own
charge pump circuitry to generate +10V and -10V for the RS-232 side.
You're right, of course. I have a FIC PA-2012 that uses the same (?)
chip, an ADM213EARS.
Here is the datasheet:
http://www.analog.com/UploadedFiles/Data_Sheets/385450995ADM206E_7E_8E_11E_13E_c.pdf
I've been aware of such 5V-only chips for quite some time (Maxim also
makes them), but until I saw the Analog Devices datasheet I was
struggling to explain why a designer would opt for 8 charge pump caps
and a more expensive IC instead of simply taking advantage of the
+/-12V supply rails. It appears that this IC is powered from the 5VSB
rail and has a low power shutdown mode in which two receivers remain
active for wake-on-interrupt monitoring.
In any case, your original statement that the motherboard doesn't
"use" these negative rails can be seen to be ambiguous. A PCI-only
board would not need the -5V rail since the PCI spec makes no
provision for same. However, the -12V supply *is* bussed to the PCI
slots, so the motherboard does actually use it.
When I bought this mobo used, its RS-232 ports didn't work except at
slow speeds because two .1 uF surface mount capacitors for the charge
pump were missing. I thought they'd been knocked off during
installation of a PCI card, but apparently they had vaporized when the
previous owner plugged or unplugged a parallel printer or serial
device with the power on. I know that the parallel port could do this
because my friend later unplugged a printer from this mobo and caused
the same capacitors to explode.
This is bizarre. I can't understand how the charge pump caps for the
RS232 driver IC can be affected by a failure of the parallel port. On
the PA-2012 board the parallel port is connected to a Winbond W83877F
multi-IO chip. I would think that your PA-2007 would be similar. In
any case, for a capacitor to vaporise, I would think that it would
have to experience a large overvoltage. I can't imagine that an
inductive spike from a data cable disconnection would have sufficient
energy to cause this failure (?). If anything, I would expect the I/O
chips to fail well before any passive component. Having said that, I
*have* seen failures in open collector drivers (eg 7406) in 20mA
current loop implementations of RS232 when the data cable has been
disconnected with power applied.
- Franc Zabkar