[Asterisk-Users] Re: Problems with TDM400P card

[Rich Adamson]

> > >I am new to Asterisk, but so far, having looked at this thread (which is
> > >quite long and full of lot's of hard work), the zttest code, the Digium 
> > >web site,
> > >and the spandsp website, and having talked to Digium sales/support about
> > >the TDM, I think the issue belongs the spandsp community.  They may be
> > >able to point out an improvement to the TDM product that Digium may or
> > >may not implement.  Or they may need to adapt to an improvement on new TDM
> > >cards. It does not appear that Digium has promised support to
> > >the spandsp add-on function to Asterisk. I also didn't see where the 
> > >spandsp
> > >project was committed to staying up to date with each new revision of
> > >TDM card - that's what the community is for.
> > > 
> > >
> > Then you think wrong. This is totally unrelated to spandsp. 
> 
> Probably. I was careful to (CYA) point out my conclusion came only 
> from what I learned in the recent discussions and research. 
> 
> > Any FAX 
> > machine, modem, or other timing critical device has the same problems 
> > with these cards.
> 
> I was hoping to learn if this was the case or not.  But are all or some fax/modems 
> not working with recent TDM?
> 
> The system test becomes more complex now:
> 
> The benchmark becomes a wide selection of reputable fax/modems.  Pass 1
> is TDM (including multiple revs of TDM, software and hosts) vs benchmark.
> Pass 2 is spandsp vs benchmark. Pass 3 is TDM vs spandsp.
> 
> It's possible a pattern will emerge from Pass 1 that will obviate the
> need for Pass 2/3.
> 
> > Something is broken, and since it used to work OK with 
> > the hardware I am using now I assume it is the TDM driver.
> 
> TDM not working with fax/modems is a powerful indictment against TDM.
> 
> Also, if spandsp is reliable with spans and not TDM, yet another
> indictment accrues against TDM.
> > 
> > By the way. Where is this spandsp community? I've never heard of them. :-)
> 
> Ouch :) Where's your P.C.?  Rich is a pretty strong advocate.  He was
> using spandsp to slap TDM. I wasn't sure he could do that at first.  Now
> I think its OK, but using modems and faxes to smack TDM will inflict
> much more pain.
> 
> Lot's of users can be organized into a system test program the likes of
> which no private concern can match.  This thread induced a spontaneous ad
> hoc system test using zttest.  Pass 1 described above could be done as
> follows:
> 
> 1. Write up the test procedure.  The test cases are: 
>    - line to line (3 cases: fxs/fxs, fxo/fxs, fxo/fxo)
>    - line to span (2 cases: fxs/span, fxo/span)
>    - line to IP (2 cases: fxs/IP, fxo/IP)
>    for a total 7 conduit cases.  Running the test in two directions for each 
>    conduit combination yields 14 possible test cases for each modem or fax. 
> 2. Specify how to report host, software, TDM, etc. configuration; write a script to
>    compile the information.
> 3. Provide a wiki or something for test reports.
> 
> I've got FXO and FXS TDM on order, so I'll participate. 

Mike, since you're coming into this with not a lot of historic background,
let me see if I can summarize in a rather short way.

I purchased one of the first TDM cards from digium just a fews days before
it was actually announced, and have been using it in test and production
for almost a year now. The card has four active pstn lines from two
different central offices. I've got a very heavy hands-on professional telephony
background (21 years as an engineer with a major telco). I'm very interested
in correcting the problems with this card (if possible), and certainly not
interested in tossing negative comments towards anyone at digium. (Let's
just view it as helping them resolve whatever we can.)

Steve has been active in asterisk development work and you'll see his
name associated with some of the echo canceller work done earlier. He
has spent a fair amount of professional time diagnosing problems with 
the spandsp/TDM, and in all cases that I can remember, the problem has
been with missing data coming from the TDM card (call it frames, octets
or whatever makes sense). And, as you probably already know, any missing
data associated with modem use (fax or otherwise) is going to cause a
major issue.

None of us know much more about it other then...
 - it does not seem to be related to the OS distro
 - spandsp is probably a very reasonable stress test and rather easy
   to implement (and most TDM users will fail, but a few are successful)
 - my testing tried to eliminate as many things as possible, leaving
   other items as potential sources of issues to be addressed
 - jacking with pci latency, interrupts, etc, does _not_ lead to anything
   that one can point to that would improve the odds of it working
 - there is a significant interest by many (as you can tell from the
   posted responses recently on testing) in getting it to work
 - there are more cases of slow/older mobo's working then there are
   faster/newer mobo.
 - the majority of the * deveopers don't want to get involved with
   the card drivers, zaptel, etc, and historically the bench strength
   within digium seems to be a little on the short side in this area
   (not intended to be a dig against digium either)

I think its fair to say the TDM card is operating on some ragged edge
where using it with voice is pretty good, but passing anything else
(eg, modem traffic of any sort) is well below the marginal level.

If the TDM card can't be fixed for whatever reason, that's okay, we
just need to understand its limitations and move on. But, if there
is something that can be tweaked, we'd all like to ID that issue.

If the lack of a TDM card is the only thing holding anyone back from
helping out, I can certainly make a system available to assist in
that process.

Rich