Here’s the page we think you wanted. See search results instead:


Contact an Expert

N4962A compares incoming Error Detector data bits with its own PRBS pattern, which has the same setting as the PRBS generator used in the PG part. If we use external clocks for both the PG and ED parts, can we run them at different data rates?

The customer asks about creating a 5 Gb/s pattern with the PG and muxing up to 10 Gb/s. Then demuxing down to 2.5 Gb/s on the ED side. There are a couple of problems with muxing the PG signal from 5 Gb/s up to 10 Gb/s. I do not recommend doing this, and here is why:

You would need two PG signals at the input of the mux, and they must be offset by 1/2 pattern length in order to create the same PRBS pattern at the mux output. If this is done improperly, then the resulting signal is a “pseudo-PRBS”, not a PRBS pattern and the DUT is not being correctly stressed. Trying to do this would involve either a splitter or a second PG and A LOT of manual delay adjustment which would not necessarily be the same after changing frequency, pattern, or power cycling. It is also important to note that measuring BER is not a good check of whether or not the mux side is properly set-up because pseudo-PRBS signals still demux back down to PRBS signals. It is very possible to get this wrong, and still see 0-BER. Therefore, you end up checking the pattern on a scope, which makes this impractical for anything other than a PRBS7 pattern.

It would be more straight-forward to provide the PG (TX) with a 10 GHz clock (for 10 Gb/s data) and then use the 1:4 demux and provide the ED (RX) with a 2.5 GHz clock (measuring 2.5 Gb/s data).

I tried out something similar here with a 5 GHz clock to the PG and a 1.25 GHz clock to the ED and it did seem to work okay. Note that this setup does not use a demux; instead the ED is sub-sampling, measuring every 4th bit of the PRBS pattern. The key points are:

  • You must set the clock source to external
  • You should set the N4962A frequency to the ED clock frequency (not the PG rate)
  • You must leave the ‘No Data’ function OFF (this means there is some chance of measuring 0-BER when no signal is present at the ED input).

Otherwise, the rest of the operation is like normal. Please be sure the customer understands that, while this is expected to work, it is not the normal mode of operation for this BERT and is not guaranteed.