Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [10GMMF] TP3 Discussion



 

Some thoughts from me on the question of separating out the dynamic testing. There is clearly merit in the arguments for both doing this, and also in retaining a single test. However, on balance, my view is that separating the test can provide an advantage that is worth having …

 

On the validity of separating out the dynamic testing

 

For our (Phyworks's) EDC IC, adaptation involves digital processing, and the adaptation speed is governed by digital parameters. For this reason we do not expect variation in adaptation speed with temp, supply, ageing, or from sample to sample. If this is also true for other EDC solutions, one could make the case that dynamic behaviour needs to be verified only during design validation, and need not be measured for each sample.

 

Aspects of a receiver that manufacturers will likely wish to test for every sample will include the various component gains, linearities, bandwidths and noise contributions. It seems to me that these aspects may be covered by one (or perhaps more than one) static test.

 

In the above I have referred to the dynamic "behaviour" of an IC. Whilst this is not the same thing as the resulting dynamic channel "penalty", I believe that a reasonable working assumption is that we may separate out the dynamic testing. Of course this assumption will require justification in due course. - I agree that caution and care are required.

 

On the desirability of separating out the dynamic testing

 

Ideas have been suggested for a fairly straight-forward static compliance test (e.g. use of a Bessel filter as the ISI generator). The situation may change, but I believe the proposals for dynamic testing currently appear more complex (perhaps a topic for discussion during our September meeting). This is the helpful way around - the more complex test being the one that may be required only for design validation.

 

In summary

 

In summary, it does seem reasonable to assume that we can separate out the dynamic testing, and also that this would be a useful thing to do. Having said this, the assumption does need to be verified, and the usefulness becomes less if we find that there is little cost implication.

 

Regards,

Nick Weiner.