#: 20210 S1/General Interest
    15-Aug-94  16:11:36
Sb: #20208-RS232 response time?
Fm: Simon R Ashby 100111,2173
To: Ken Scales 74646,2237 (X)

>19200baud timing calcs....
Thanks for your interest, Ken. In fact, we specify the 3ms timing requirement
from the receipt of the character by the host hardware. This takes out the
initial latency assuming a cable link (no modem/satellite delays.) Furthermore,
the timing period is to the first character in the string; the data is required
to follow in a continuous burst.
I hope that reassures you.
TO ALL:
The problem (it transpires) lies with the BIOS (" R**stone CIOS"?) in the 68020
I/O coprocessor, NOT the main 68040 running OS/9. Apparently when the host
arranges an interrupt from  the co-pro when a serial char arrives, it is
subject to the 25ms co-processor BIOS polling loop. There isn't even a way to
wire in a wire hardware interrupt from a port pin: so that's it.
It always amazes me how large systems insulate their programmers (compared with
our little PCB80528 uP, an 8052 with an extra 256 bytes of ram on board!) A
resident operating system is only a help when the designers have anticipated
EVERYTHING that you might use it for. Perhaps that is why Windows is popular --
by allowing you to break the rules it allows the legion of hackers to make it
usable.
(what a hideous thought!)   All the best        .. Simon A