[asterisk-users] PRI Test Lab

[Matt Florell]

On 2/13/09, Kevin P. Fleming <kpfleming at digium.com> wrote:
> Matt Florell wrote:
>
>  > Can you tell me where the setting is to force Master timing on Digium
>  > cards per port? I really didn't think Digium cards had the ability to
>  > force Master in this way. I've tried to do it with channelbanks before
>  > and couldn't force it to master, whereas I can get it to work with
>  > Sangoma.
>
>
> I'm not sure that the terms you are using match what I'm used to hearing
>  and telling people, so let me try to describe how our cards work and you
>  can tell me how that matches up with what you want to do.
>
>  Digium multi-port T1/E1 cards use a single clock source for transmitting
>  on all connected spans. That clock source can be on the onboard
>  oscillator, or the recovered clock from one of the spans, and the source
>  selection can change dynamically based on the configuration (in other
>  words, you can set the first span as 'highest priority', the second span
>  as 'second priority', etc). If a span is selected to be the clock source
>  for the entire card, and then that span goes into red alarm, a different
>  clock source will be chosen, until that span recovers.
>
>  So, what this means is that each span port that is configured to be used
>  is *always* transmitting a bit stream, and due to the nature of T1/E1
>  signaling, that bit stream includes a clock. A device connected to that
>  port will *always* be able to recover a clock from that bitstream if it
>  chooses to do so. For channel banks, other servers, downstream PBXes,
>  etc. this is a common configuration, and the device will derive its
>  clock from the bitstream it receives from the Digium card.
>
>  In cases where the system admin is using spans that are generated from
>  'upstream' devices, where the card should slave its transmitted clock to
>  the recovered clock from that span, then the card can be configured in
>  this mode. Once a span has been selected to be the clock source for the
>  card, *all* the spans on that card will use that clock source for their
>  transmitted bit streams.

I guess I'm not explaining myself very well, so I'll describe exactly
what happened and how my problem was solved. We had a digium quad port
T1 card with 3 carrier T1s plugged into it and one channelbank. After
a few months of everything running just fine on the system the
channelbank would go red alarm after a few hours of the server being
on, if we reset the server, channelbank or even just unplugged the
crossover T1 and plugged it back into the channelbank it would work
again for a few more hours. The carrier told us it was a timing issue,
so I began to mess with the timing settings and after a week of making
changes and waiting to see if they would work, none of the changes to
any of the timing settings in zapata.conf would do anything. At this
point I swapped out the Digium card with a Sangoma card(because the
quad T1 cards were cheaper than a new channelbank) and the same thing
happened. I emailed Sangoma support and they suggested I try the
forced Master clock setting for the channelbank port to see if that
would help, and after figuring out exactly how to set it up I put it
live and no more red alarms on the channelbank.

My understanding is that this setting lets you ignore the timing
signal coming from the other end of one of the ports, and the card
will take another timing source from a port that you specify and
"force" it to be used as a timer on that first port.

If my explanation makes no sense I apologize, but this is how it
happened and the problem was solved.

MATT---