Maybe someone here can help me with a problem we have with our PSTN gateway.
We receive 63 E1's channelized into a STM-1. The E1's are provided by DTAG (German PSTN) and are multiplexed into the STM-1 provided by Verizon. STM-1 directly terminates in a Audiocodes Mediant 3000 which demultiplexes STM-1 and routes E1's respectively.
We see a huge number of Controlled Slips (one in 20 seconds) and it seems this condition causes fax calls to crash.
We got in contact with Verizon and they told us, that they are receiving DTAG's E1's in sync with a PDH clock and transparently multiplex that into the STM-1 which is clocked by a different master clock. (As it is not their policy to generate clock from external sources and PDH can not be used for SDH clock source - that's what they said) [For what I read on the interwebs SDH was specially designed to support relaying of PDH - so I don't really get the point...]
So we receive E1's and STM-1 on different source clocks. Unfortunately Audiocodes tells us the gateway only supports one single clock source, which has to be the STM-1 or any external clock directly connected to the device.
I appreciate any hints or tips on this issue (or any lesson on PSTN clocking) as I am used to deal with IP networks on the very higher layers...
When setting up a system that transports E1s over an STM-1 interface, you can use one clock for the STM-1 interface and a different clock for the E1 interface. These two clocks should be locked to each other and be derived from the same source, but do not have to be which your case is - The E1 clock from DTAG and the SDH from Verizon are not locked to each other.
The source of your slips is coming from the mis-match of these two clock rates. What is happening is that on your Audiocodes receive side, it is receiving the STM-1 and de-multiplexing it into E1s. The differences in frequencies of the STM-1 and the E1 are dealt with [via] pointer justifications happening at the TU-12 path level (A TU-12 is what is used to transport an E1 over a SDH network - you have 63 TU-12 paths - one for each E1). Since these frequencies are different, if your Audiocodes box does performance monitoring, you should be able to look for pointer changes at the TU-12 level to see that these two clocks are different.
Pointer changes are not the same as slips. A slip occurs when the receive clock on the E1 is different from the system clock in your Audiocodes box. Inside the E1 framer of the Audiocodes box, there is a block of memory that the timeslots of the E1 are stored as they arrive on the E1 using the receive E1 clock rate. The timeslots are then readout of this block of memory at the system clock rate. NOTE: This block of memory is called the elastic store or slip buffer. So, what happens when the rate of timeslots arriving on the E1 is faster than the system takes them out. Since there is not infinite memory, the E1 framer will do a controlled slip which would, in this case, would delete an entire frames worth of timeslots. Conversely, if the rate of timeslots arriving on the E1 is slower than the speed the system takes them out, the E1 framer device will create timeslots by duplicating the last frames worth of timeslots. The deleting/creating timeslots during slips will be hardly noticeable to the human ear, but for data services like fax, it will cause the kind of problems you are seeing.
So, the solution to you problem is to get the system timing of the Audio codes box to be locked to the E1 clock rate. That may be difficult, if not impossible, given the features you describe that are available in the Audiocode box. You indicate that you only have two choices for system timing on the Audio codes box: Receive STM-1 or external input.
Since the STM-1 clock is derived from Verizon's network and the E1 clock is from the DTAG network, you cannot use the STM-1 clock because they are not locked to each other.
Thus, you need to provide a clock on the external input that is locked to the DTAG timing somehow. You don't have enough description to determine if that is a difficult thing to do or not.
Finally, you can find a different receive device that allows the system timing to be locked to the receive E1 clock.
The SDH clock will be country or intercontinental in scope for any carrier - in our network we will use our caesium clock sources for stability thank you :)
in reality, 2 different SDH networks would normally be clocked accurately enough that clock slips between different networks are rare
- maybe less than 1 a year?
E1 is "PDH" and the standards allow for 50 ppm clock variation between the source of any circuir and a notional 2.048 Mbps accurate clock. It is common practice to use a different clock on each E1, produced by whatever is plugged into the source port at 1 end (or even different clocks at each end of the same E1)
However - if you are getting a TDM voice / fax signal within each E1, the voice signals in the G.732 64k channels have much more stringent timing constraints, and should all trace back to another "carrier class" accuracy clock.
1 slip every 20 sec is 1 part in 40m or so and you should get much better accuracy than that / lower slip rate
Your equipment should use a "voice" clock from 1 or more of the E1s - but it probably assumes they are all locked to the same clock.
So - your voice / fax stuff is where you want to look
maybe you are getting clock from multiple places and 1 of them is inconsistent (note "wrong" here is likely to be agrued + vary between the telcos....)
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.