Serial Boards
18 messages in this thread
Does anyone have a C= multi serial board (A2232) running with a high speed
modem?
I'm looking into opening a large BBS using USR Dual Standards on each line
The problem is if I use ASDG/GVP boards I will be limited to 8 lines but
not limited on speed on the modems. The C= boards are low cost and contain
7 ports per board. Only thing is, is each port is rated at 19.2K.
Considering this will be a data base board (only file transfers) and that
all the files will be compressed, compression (.42bis) will not be relivent
to the speed of the transfer…thus limited to 14.4K (16.8 USR HST). so in
theory I should have no problems getting 1700 cps using Y-ModemG. But
anyone who has worked with SCSI long enough knows they can't rely on that
alone. With my A3000 I am unable to get past 1650cps. 1680 is my all time
high on the Ami, and my average on my IBM (I do have 16550 though).
I've also heard of the new 28.8 Standard…this would seemingly make the C=
useless in the future. But USR says that you should expect anything past
19.2 unles you phones lines are fiberoptic and blessed by the gods (not
their exact words). Does anyone know anything about this new standard?
Of course the easiest solution would be a serial board with 5 or more ports
capable of 57.6K…but I've yet to find one.
Dave
Elite MicroComputers – via Whap!
Dave,
Yes, I've been using an A2232 for over a year now. It's currently
connected to a PP9600SA and is what was used to put this reply on CIS
(using AutoPilot).
Will that modem do 14.4K ? What is the highest cps you've gotten?
Dave
Elite MicroComputers – via Whap!
Dave,
No, it only does 9600 (my 14.4 will hopefully arrive in a month).
I've gotten approx 920cps when doing file transfers (assuming the… other
end cooperates).
The highest data rate the board will support is 19.2kbps. The highest
rate the modem will support is 14.4kbps. Hence it will be the
limiting factor on all but raw text transfers to similar modems. Then
the 19.2kbps will become the limiting factor, VLT'd probably be able
to provide rates up to close to 1850 or more cps for straight text
transfers and about 1380 cps for compressed files.
Dave,
Considering the construction of the A2232, and it's relatively slow
processor, I doubt that it will support the throughput you're after on
multiple ports simultaneously. Although I don't have the book handy to
check, if I remember correctly, the board has one CPU and 7 UARTS (and an
8520). The CPU runs somewhere around 3-4MHz. It'd probably work OK on
multiple 2400 baud, non-V.42 modems, but not the high speed ones you're
considering.
If you'd like more info, let me know and I'll try to hunt up the manual.
David
David;
There was a pretty heavy discussion of this a few months back (on another
network). The conclusion (if you can call it one) is that the A2232
should handle 7 9600 baud lines at full speed. The timing considerations
showed room to spare but personally I doubt that it would work.
Happy Holidays to you and yours,
-bill
Bill, and others, I've got my 2232 set up with a high-speed modem, a fax
board, a 9600 connection to my printer sharing device, and a couple 9600
connects to other machines in the office.
I ran them all at once and I don't 'see' any bottlenecks… that's 5
9600+ connections at once…
Thanks Gordon for sharing that info. On the "back burner" is a
project that will need quite a few serial lines all running at once
and I have been a bit concerned that the machine would not be able to
handle it. -bill
Sounds good enough for me…I guess I'll pick one up.
Might you be able to 'Hack It' with a faster CPU, or would that mess up the
timing of the ports?
Dave
Elite MicroComputers – via Whap!
Dave,
My _guess_ would be that there'd be more to it than just putting in a
faster CPU. I've wondered about the same thing (hacking it with a faster
CPU) before, and dismissed it for the reason you mentioned. There's not
enough information supplied with the board to enable one to write a new
driver for it.
The board would support the baud rate well enough when not under load.
The question would be when you'd hit the upper limit of its capabilities,
and what would happen to your system at that point. Output data probably
wouldn't be a problem, since you'd just have additional time between
bytes. Where I'd expect the problem would be in losing input data.
Probably the only way to be certain would be to try one out, but if it
doesn't do what you need out of the box, I'm not at all certain that
anything could be effectively done to it to make it work for you.
David
Well this would be a 'download only' BBS….so It might not work out to
bad.
Dave
Elite MicroComputers – via Whap!
Using GVP boards in four slots will give you eight ports. I was also under
the impression that the TwinX could be configured with 4 ports (2 serial
modules with 2 ports each), using two slots for eight ports.
-Chris
This TwinX….who makes it? Do you have a phone # for them?
Dave
Elite MicroComputers – via Whap!
Chris:
Using the TwinX board with the serial modules makes each card take up 2
slots, so you'd still have 4 slots used. Why not use the C= 2232 card and
have 1 slot = 8 ports? (7 on the card + internal serial = 8 ports) I have
the 2232 card in my machine, and love it.
// -Mike Schiller
\X/ AMIGA! On AutoPilot in sunny Fort Lauderdale
What speed are you running it at…I am opening up a multi line BBS…it
will be download only so I don't have to worry about input to the cards.
Read my old message at the begining of the tread….the point is this it
has ports…but with it handle 14.4 (16.8HST) on each one simultaiously.
I like the C= board, but the answer to my prayers is a 7 port board with
speeds at 57.6K. Remember I don't want to have to upgrade when the new
28.8K standard comes into effect. You would think the C= boards wouldn't
handle it (19.2K max) but the word from US Robotics is you shouldn't expect
anything past 19.2K unless you have (really good) fiber optic lines.
Dave
Elite MicroComputers – via Whap!
I believe the 2232 will do 19.2kbps on all 7 ports with no problem.
It would be nice if it could go faster, but then it is what it is.