Clocking Issues with VWICs

Question

We connected a 3825 to a Nortel PBX today and had some weird clocking issues. We're using VWIC-2MFT-T1 modules and I think I need to learn more about how clocking works with these things. The PBX didn't seem to like it when the router provided clock. The engineer working with us told us that he thinks the clock on the router is not stable enough to provide clocking.
Have any of you had any issues using a router to provide clocking to a PBX? It seems like it shouldn't be a problem but we kept experiencing slips on the Nortel side if the router was providing clocking.
This creates a problem for us. We will be connecting one port to a frame relay data circuit and that port must be set to "clock source line". If the router can't provide clock to the PBX then that port also must be set to "clock source line".
I ran across a tech note that says the following:
"When both ports on a VWIC-2MFT-T1, VWIC-2MFT-E1, VWIC-2MFT-G703, or VWIC-2MFT-E1-DI interface card are configured as clock source line, then they must be set to the same clock source. If they are not set to the same clock source, timing slips can occur. Each port on the VWIC2-2MFT-T1/E1 and VWIC2-2MFT-G703 interface cards can be set to independent clock sources for data applications."
Are we going to run into problems since we don't have the VWIC2 modules?
I don't understand the way clocking works with these modules anyway. With only one port connected to the PBX we were seeing slips even with the router was set to "line". We fixed that by using the "network-clock-select 1 T1 0/0/0" command. I don't know what that command actually does but it stabilized the clock and eliminated the slips. Weird. I need to go do some research on this issue because we need to get it figured out and settled soon.
Any thoughts?
Thanks!
John

Answer 1

The 2800 and 3800 routers have internal tdm buses that can be used to tdm switch between any ports on the box. You use 'network-clock-select' to pick the clocking source for the internal tdm bus. This clock source is then used for every interface that is identified via the 'network-clock-participate' command. The best you can do in your case is to clock off the frame relay network and see if you get slips on the T1 to the nortel, or clock off the nortel and check for slips to FR. Make sure the nortel is set up correctly to either take clocking from the T1 or provide clocking on the T1 - you may not have a choice there depending on protocol used. Best practice (and what you may end up needing to do) is to separate the T1s onto different VWICs so that clocking can be independent. Even with a VWIC2 you only get 1 clock source if using voice (with data can run ports independently).
The internal clock on the router, like the internal clock on the PBX, should only be used for fallback as they aren't nearly as accurate as what is used in the cloud. For this reason you create a synchronization plan that cascades the clock from the cloud. In your case the nortel would clock off the PSTN (I'm sure already being done) and cascades clocking to the 3825. When you get more PSTN trunking into the IP side of things you would want to swap - have the 3825 clock off a PSTN T1 and then cascade clocking to the Nortel.
hth,
Don