"M.I.5¾" <no.one@no.where.NO_SPAM.co.uk> wrote in message
news:452c91b9$1_1@glkas0286.greenlnk.net...
>
> "Barry Watzman" <WatzmanNOSPAM@neo.rr.com> wrote in message
> news:452A492D.8000308@neo.rr.com...
>> There are lots of devices that don't work. The real problems come in
>> when the software for the device require that the device be on a specific
>> COM port and when that same software then tries to DIRECTLY program the
>> UART (the 8250 equivalent) by writing directly to 8250 hardware
>> registers. Poor design? Absolutely, but that doesn't change the fact
>> that the device won't work with a USB to RS-232 converter.
>>
>
> The software that supports USB-RS232 devices provides a software emulation
> of the I/O port and the interrupts that a real COM has. The actual
> converter has a real UART - it may not be directly accesible through a
> direct write, but the controlling software apparently intercepts the write
> and translates it to the UART in the adapter. Maybe these devices vary,
> but the one I have (a cheap unbranded £9.99 model) works perfectly even if
> you write your own DOS programme that 'writes' directly to the I/O
> registers. It even apparently generates interupts on the expected
> processor interrupt INT3 or 4 - almost certainly software emulated as
> there obviously can be no physical connection.
>

Further to the above:

I remebered that I had an old Belkin USB to RS-232 adaptor in my junk box,
so I dug it out. It was much larger than today's offerings. So out came
the screwdriver and I opened it up, and there inside among the 3 chips was a
real genuine dual in line 16450 UART (the predecessor of the 16550). The
driver disk actually had 2 drivers, the Windows driver and importantly, a
DOS driver for use in a command window. The DOS driver emulated the device
to appear exactly as a real COM port would. The driver properties even
allows you to select which interupt to use (out of a choice of 6). I recall
the Belkin device worked well, but operated extremely slowly with PsiWin for
some unexplored reason.

Talking to a few people reveals that not everyone has a good time with USB
to RS-232 adaptors, which makes me wonder if, in fact, some of the failure
stories are due to limitations in the drivers rather than any inherent
weakness in the modus operandi of the adaptor.