[comp.unix.xenix] Problems with UUCP and Telebits

kory@avatar.UUCP (Kory Hamzeh) (04/07/90)

I am running Xenix386 2.3.2 with the HoneyDanBer UUCP package. My Telebit
is running on COM1 and 1200/2400/19200 bps. I seem to have alot of problems
at 19200 with incoming calls. The uucico log file just says:

	FAILED (conversation complete)

Can anyone give me any hints as to what to look for? Or even send me a copy
of your default NVRAM values for the Telebit?

On outgoing calls, the problem also occurs, but far less frequent.

Any help would be greatly appreciated.

Thanks,
--kory



-- 
-------------------------------------------------------------------------------
Kory Hamzeh             UUCP: avatar!kory or ..!uunet!psivax!quad1!avatar!kory
                    INTERNET: avatar!kory@quad.com

campbell@Thalatta.COM (Bill Campbell) (04/08/90)

In article <143@avatar.UUCP> kory@avatar.UUCP (Kory Hamzeh) writes:
>I am running Xenix386 2.3.2 with the HoneyDanBer UUCP package. My Telebit
>is running on COM1 and 1200/2400/19200 bps. I seem to have alot of problems
>at 19200 with incoming calls. The uucico log file just says:
>
>	FAILED (conversation complete)
>
>Can anyone give me any hints as to what to look for? Or even send me a copy
>of your default NVRAM values for the Telebit?
>
The problem is most likely in the uart on the COM1 port.  These are
typically brain-damaged parts that work OK on Messy-DOS because DOS
doesn't process interrupts, but fail miserably under a real operating
system.  The original IBM releases of Xenix specified maximum port
speeds of 2400 baud on a *10 FOOT CABLE* :-) largely because of the
port problems.

>On outgoing calls, the problem also occurs, but far less frequent.
>
This is on incoming data, not calls because you are under control more
of the time and the machine is not being asked to process the incoming
interrupts as frequently, particularly since the Telebits are doing
most of the ACKing.

I don't remember the part number of the later uart chips, but replacing
your COM1 board is probably your only alternative to get as high as
9600 on COM1.

I am running Telebit+s at 19,200 with no problems at all on a Specialix
8-port board, but have had problems on tty1A or tty2A.

-- 
....microsoft--\                    Bill Campbell; Celestial Software
...uw-beaver-----!thebes!camco!bill 6641 East Mercer Way
....fluke------/                    Mercer Island, Wa 98040
....hplsla----/                     (206) 232-4164

larry@nstar.UUCP (Larry Snyder) (04/09/90)

In article <3714@thebes.Thalatta.COM>, campbell@Thalatta.COM (Bill Campbell) writes:
> I don't remember the part number of the later uart chips, but replacing
> your COM1 board is probably your only alternative to get as high as
> 9600 on COM1.

16550A - but the driver needs to support the FIFO buffer in the chip
in order to do much good.

When I was running SCO Xenix (2.3.2) '386 with a dumb 8 port Comtrol
Hostess board - I had 4 high speed modems running full bang 
(1600 cps on the HST modems, 1400 on the PEP and 940 cps on the
V-Series) without *ANY* problems other than the loading - but the
modems could keep up without problems.   Under Unix (386/ix) with
their supplied drivers (even the X5 upgraded one) I could barely keep
one modem running at 2400 baud without loosing characters.

My solution - smartcards (that's a whole story in itself).  
Currently I am running with a Computone Intelliport smartcard with
a 80186 processor to handle the IO.  I now have the same throughput
under 386/ix that I did using the dumb card under Xenix - except the
system isn't overloaded - as a matter of fact - with all 4 modems
going full blast - response is such that you can't even tell the
serial lines are in use.



-- 
...!iuvax!ndmath!nstar!larry  -or-  larry@nstar

larry@nstar.UUCP (Larry Snyder) (04/13/90)

> Not necessarily; the 16550AFN will use it's buffers regardless (ie. it 
> won't loose characters), though the interrupt overhead won't be 
> reduced unless the software explicitly uses the FIFOs - particularly 
> the transmitter.

The FASY drivers are drop in replacements for the 386/ix drivers and
work excellent with 16550AFN chips.  The latest version is available
here on nstar.

> Agreed, but they're very expensive.

not really - I am currently running with a Computone Intelliport -
and I must admit it took a couple of months to get the correct drivers
working with one of the 3 sets of proms they mailed me - but for around
$500 6 smart ports that can products 1650 cps without any kind of load
on the CPU isn't bad..


-- 
...!iuvax!ndmath!nstar!larry  -or-  larry@nstar