[comp.sys.ibm.pc] serial port diagnostics and information

buck@siswat.UUCP (A. Lester Buck) (09/17/88)

I suffered a lightning-induced power line surge on my AT clone
resulting in damage to the two serial ports on my multifunction board.
Several of the EIA driver chips were fried, and one of the two
8250 UART chips was left flakey.  [Oddly enough, the one port
with good 8250 but bad drivers still received >500K of netnews
without problems, because Microport Unix ignores most modem control
signals.]  I even discovered the hard way that all diagnostic
programs are not the same.  The IBM AT Advanced Diagnositcs v1.04
delivered with my 8 MHz clone caught the bad 8250 on one port after
the driver chips had been replaced.  But the same board tested
in a Compaq 386-16 (shop where the chips were replaced) with 
Compaq diagnostics found nothing wrong.  Not very reassuring...

This experience left me with several unanswered questions:

1)	My multifunction board is for an AT, and the documentation
	for the board, as well as the IBM Options and Adapter manual,
	specifies 16450 UARTs, but the actual hardware delivered
	has 8250 UARTs, as standard on XT serial ports.
	I won't mention the names of the probable guilty parties, but
	I also have never noticed any problems with these ports.
	What exactly is the difference between an 8250 and a 16450?
	(Other than their street prices of about $8 and $17, respectively.)
	Is there any reason to upgrade?

2)	When I first obtained my AT clone, I guessed at a wrap-plug for
	the serial diagnostics by connecting (25-pin numbers) 2-3, 4-5,
	and 6-20 (i.e. TD-RD, RTS-CTS, and DSR-DTR).  This worked fine
	for IBM AT Avanced Diagnostics v1.04.  The Compaq diagnostics
	mentioned above also check pin 8, DCD, by using a Compaq part
	that wires 2-3, 4-5, and 6-8-20.  When trying to run IBM AT
	Advanced Diagnostics v2.04 with this and a number of other
	combinations (including 6-8-20-22 to test RI), the ports
	always fail diagnostics, though they pass v1.04.
	Does anyone know the official wiring for a wrap-plug for v2.04?  
	Can v2.04 somehow now tell I have 8250's instead of 16450's
	and therefore fail?  What has changed between the versions?

3)	While I am asking, does anyone know the correct wiring for
	a parallel port wrap-plug?  (I haven't experimented with this.)

-- 
A. Lester Buck		...!uunet!nuchat!moray!siswat!buck