🚀 go-pugleaf

jeremy@sour.sw.oz.au (Jeremy Fitzhardinge) writes:

>This is just a general query to find out what drivers there are
>for ISDN cards, and if it makes a difference, what kinds.

>We're looking at setting up a Linux box as an ISDN gateway
>machine, and we're wondering if we need to write a driver for
>any card we may get.

>We'd be happy with ahpha, beta test/prerelease versions, and
>could actively help develop/debug a driver.

>Also, what PC ISDN cards available approved for use in Australia?

>	J
I went through this exercise about a year ago. 

At the time Austel gave me a list of ISDN PC cards that were approved.
I contacted all the people, saying that I was prepared to work on a
driver, provided I could publish the source of same.

At that time (the situation may have changed since then) there were only
two reasonably priced cards: the one sold by Telecom, a rebadged
'Stohlman', and a 'Design Two Thousand', marketed by JNA.

(There exist other approved cards: IBM, HP, NCR, ISIS, Mitel, and
JTech - they are all expensive)

The Design Two Thousand actually looks the best (you can use both B
channels for Data - it also operates to some extent with the PC
off), but I never actually got through to JNA what GNU means etc. etc.
(They were enthusiastic, I know a number of people there, but kept
talking non-disclosure etc.). In the end I gave up becuase the cards
cost (ahem) over two thousand dollars.

I'm still interested in the idea myself, but after I put the line in,
buy the card, I then have to write code, which (essentially) helps
JNA sell more cards. Somehow I don't feel as though I'd be coming out
ahead here!

Amongst the most popular cards in Europe are the Diehl (German). There
are Linux driver's for these. I'm not sure if these are dumb or smart
cards (or if they have a range).

These are also reasonably priced (figures of DM 500). The problem is
Austel approval. The testing process was quoted to me by Austel as
$30,000. Private conversation with companies that have actually gone
through the process put the figure at more like $100,000. You have
to sell a lot of cards to make this worth while.... maybe this has
changed since then

Just to round off this long-winded article: I don't recommend external
TA's. You lose too much bandwidth in 'framing'.

For me, probably the simplest thing to do is use V.FC modems. You also
don't have timed local calls.

Regards,
Jamie

In <2rcskl$jqj@inferno.mpx.com.au> jhonan@jolt.mpx.com.au (Jamie Honan) writes:

>Amongst the most popular cards in Europe are the Diehl (German). There
>are Linux driver's for these. I'm not sure if these are dumb or smart
>cards (or if they have a range).

To my recollection, diehl isdn GmBh offers both intelligent and passive
ISDN cards. Could check out some glossy brochures if anyone wants me to.

>Just to round off this long-winded article: I don't recommend external
>TA's. You lose too much bandwidth in 'framing'.

That depends on what kind of line protocol the TA uses. Currently, the
most popular (that I'm aware of, anyway) is V.110, which is a non-framing
rate adaption protocol, and V.120, which is a framing/multiplexing protocol.

V.110 is nice to use bacause of its simplicity and lack of overhead, the down
side is that it is not defined for R interface speeds above 19k2bps. At least
it wasn't the last time I checked. Several manufacturers, including us, do,
however, offer V.110 for speeds up to 64kbps. We've tested it for interopera-
bilty between several brands of TA's, and it seems to work nicely.

V.120 offers increased functionality over V.110, at the cost of some framing
overhead.

In the case of a PC card, I believe that some framing technique is employed
whenever one bonds 2 B-channels for increased bandwidth. This may be 
necessary because of the possibility of path differences (and consecutive
differences in transmission delay) between the two B-channels.

Regards,

Ragnar

Ragnar Nielsen
Technical Manager
ID Communication as

In article <jeremy.769246367@suite.sw.oz.au> jeremy@sour.sw.oz.au (Jeremy Fitzhardinge) writes:
>We're looking at setting up a Linux box as an ISDN gateway
>machine, and we're wondering if we need to write a driver for
>any card we may get.

Probably you will. Someone has done an ALPHA driver for the Diehl series
of cards. Its not very usable but works. Also cards with a hardware
16550A UART emulation (or real chip) on on board work. Some have OS/2
software UART emulators - that won't do.

Next be warned that getting ISDN to talk to other ISDN is fascinating...
One trick is to stuff all the official specs and just run asynchronous
SLIP over the ISDN line like a fast modem. PPP still has a lot of
interoperability problems.

>We'd be happy with ahpha, beta test/prerelease versions, and
>could actively help develop/debug a driver.
I'm still trying to get permission here to do a driver for the Sonix 
PC/Arpeggio card.

>Also, what PC ISDN cards available approved for use in Australia?
No idea sorry.

Alan

>>Just to round off this long-winded article: I don't recommend external
>>TA's. You lose too much bandwidth in 'framing'.

>That depends on what kind of line protocol the TA uses. Currently, the
>most popular (that I'm aware of, anyway) is V.110, which is a non-framing
>rate adaption protocol, and V.120, which is a framing/multiplexing protocol.

>V.120 offers increased functionality over V.110, at the cost of some framing
>overhead.

I thought there was a general move towards V.120, mainly because it allowed
multiplexing. I'm on very shakey ground here... 

>In the case of a PC card, I believe that some framing technique is employed
>whenever one bonds 2 B-channels for increased bandwidth. This may be 
>necessary because of the possibility of path differences (and consecutive
>differences in transmission delay) between the two B-channels.

Using SLIP or PPP and TCP/IP I guess there's always lots of 'framing'.

>Ragnar Nielsen
>Technical Manager
>ID Communication as

Who are ID Comm's? Do you have any PC ISDN cards in Aus?