[Asterisk-Users] T1 Sync clarification

[Steve Underwood]

Rich Adamson wrote:

>>>To complete this rather lengthy topic... what happens if you ignore all of
>>>this and just slap a bunch of systems together with no regard to a master
>>>sync source?  The quality and stability of your network will likely not be
>>>as good as what it could be. If your clocks (in each device) happen to be
>>>running very very close to what is expected, your network might run just
>>>fine. But, if one of the clock's frequency drifts around, it could impact
>>>quality via frame slippage and other unwanted events, and if off by a
>>>large amount could even be the source of failures. (Your milage will vary
>>>directly with the stability of your clocks.)
>>>      
>>>
>
>  
>
>>What are the practical effects with in-correct clock sync
>>-like to you hear odd buzzing, or dropped voice or gaps of audio ??
>>    
>>
>
>As mentioned earlier, it depends entirely upon how far off one clock is
>from the clock at the other end of the T1.
>
>If they are off by a little bit, you would see frame slips but probably
>not hear any quality differences.
>
>As the slip rate increases (to some unknown value since I've not tried
>personally to qualify this), the audio would be infrequently interrupted
>from the lost frames. I would expect you to hear it as repetitive clicks 
>of some sort that might be construed as noise. The exact noise would again 
>depending upon how far off the clocks really were. Each audio channel 
>consists of 8,000 voice samples per second (on a normal US T1), so if the 
>slip occurred once/second on average and then recovered, one would probably 
>not hear 1/8000 second of a hickup.
>
>If the slips were 100/sec average, it's likely the end nodes would have
>a hard time recovering from it (best guess), and I would expect noise to
>be apparent.
>
>Others that have more experience correlating slip rates to noise levels
>might have a better description of the noise vs slip rate.
>
>Rich
>  
>
You would only have a fast slip rate if something is faulty. Anything 
complying with the E1 or T1 specs should never have its clock >50ppm in 
error. Anything coming from the PSTN is essentially bang on, as it comes 
from an atomic clock.

Some people have commented about potentially difference clock rates from 
different providers. In practice that doesn't happen. Providers have a 
rhubidium clock in each exchange. These are so accurate, frame slips 
would be a one a year event. However, phase locking between carriers 
usually ensures even that does not occur. The globe's phone systems are 
pretty much all locked together these days.

The older higher order digital links - 8, 34, 140, and 565M in E1 land, 
and DS3 etc. in T1 land - have a bit stuffing scheme that allows 
individual E1s or T1s to be at slightly different rates. This is called 
PDH - plesiochronous (almost synchronous) digital heirarchy -  and was 
very helpful in moving from a totally analogue network to a mixed 
analogue/digital one. Once the network backbones became 100% digital, 
this became a huge liability. SDH (synchronous digital heirarchy, or 
Sonet) was born to solve this. SDH assumes the entire network is 
perfectly synchronised. Drop and insert is *far* cheaper in a truely 
synchronous stream. SDH is the norm for anything new today.

Regards,
Steve