On 09/12/2010 09:06 AM, Jacques Goldberg wrote:
Sorry to disappoint you Jonathan
The rate reported by wvdialconf is not the transmission rate! It is
(supposedly, never checked, irrelevant) the rate at which the software
communicates with the serial connection to the modem chip located in
the PC.
You will see the true rate when you launch a wvdial command.
wvdial will report the word "CONNECTED" and the rate, and it will
never reach the theoretical 56kbauds limit.
Gotcha.
Just to tease you, though, I am currently working flawless with a
serial connection at 460,800 bauds. But not the way you think of.
I use a Xilinx Spartan 3AN FPGA development board as a data
acquisition system. It has an RS232 port which can go even at least
twice faster than 460,800. I have connected it to a 10-15$
RS232-to-USB adapter with a shielded cable 20cm long only or it won't
work, and the USB side of the adaptor is connected to my laptop with a
40cm shielded USB cable or it will not work.
Actually while pushing the above to its limits to match the data
acquisition system specifications, I "discovered" that the 115200
bauds limit which everybody knows about is just what was set in the
initial version of National Semiconductors's 16550. There are more
recent versions of this chip which could easily work at 800 kbauds but
they cost 1 or 2 dollars more per piece, a tremendous saving for PC
manufacturers.
This is quite interesting to me. About three years ago I was asked to
try to replace the server of an old multi-user system; all I was told
was that the server was running Windows 95. I found some hardware with
Windows 98 drivers available, ghosted the hard drive to the new
hardware, and completed a Windows 98 in-place upgrade. And then when I
ran the server binary, I was in for a shock.
It was DOS-based, not Windows-based. Last updated in 1984. Clearly
trying to use COM1: in an RS-483 mode. Deep memories in dark storage
opened up, a book I had half-memorized in 1986; RS-483 was a standard
serial-port mode then, the book described RS-232/RS-422 of that time. I
looked at the remaining bits of the old server -- someone had attempted
to work it before me -- and found just a basic ISA 16550A board, nothing
unusual at all.
The server process had been running successfully in a DOS shell under
Windows 95 for at least ten years, so I figured maybe Windows 98 might
be OK too. It had to also cooperate on a 10based2 Windows 2003 network,
so I didn't want to try to take it to FreeDOS, which I might have
otherwise. The big catch, though, was the Dell hardware I had found to
rebuild it all on, had no discrete 165**0* chips at all; its serial
ports must have been part of larger multifunction chips. But to sum it
up, it did work, and very well. The most interesting thing is, when I
later looked up RS-483, I learned that it is still in use to synchronize
many large-arena PA systems...and that it can communicate up to 1 Mbps.
I wonder what kind of cable is needed for the latter.
J.E.B.
