Re: PRBS23 test pattern -- RE: [802.3ae]802.3ae PRBSsareupsided own
John Irwin wrote:
> You could be correct, although the SONET/SDH testers that I have used you
> would ordinarily define the mux structure to be tested (i.e. VC4, VC3, VC12,
> VC11, ...). These testers them demultiplex the payload from this mux
> structure (i.e. remove all frame overhead), and then test the underlying
> payload for a pre-programmed data structure (i.e. PRBS). Once the multiplex
> is defined the testers that I have worked with would expect a continous PRBS
> sequence to exist within the payload of the selected multiplex. Again, I
> would be concerned because periodically resetting the the PRBS would tend to
> disrupt this sequence.
10GE is dealing strictly with concatenated payloads. There are no virtual channels
to extract, just one fast pipe. The test pattern was designed for pattern based
tester where you must program the value of every bit in the repeating sequence.
If the PRBS were allowed to "float" in the SPE (as you suggest) you would not
have a repeatable, predictable pattern (although a floating pattern would repeat
a couple times a day). Resetting and truncating allows a clear definition of
every single bit at any given time.
Using a "Pure SONET Tester" to receive the pattern, you don't have to tell the
receiver that the SPE has PRBS, all we care about are errors (as indicated
by the BIP)
I don't know of a simple test pattern which can be used in all types of high
speed BERTs. If you can think of one, the committee is still open for
suggestions.
--
Tim Warland P. Eng.
Applications Engineer
Quake Technologies (613)270-8113 ext 2311
Shake it!