Hi,
Barry Hufker wrote on Sat, 11 July 2009 23:06 |
Klaus, For people such as myself, would you explain what is being shown in the graph?
|
The green trace in the top plot is the key point. It shows the varying "capacitance" that the amp sees vs. frequency with an unterminated cable of this specific lenght, which is not accounted for if one only does "lumped element" analysis, using the cable's LCR paramters (C would be 0.5nF/m which gives 5nf total for that 10m length). BTW I'm not sure if that simulation is really correct in absolute (quantities) terms (I even cannot recreate now the values I seemed to read from the plots, so much said
), but with an RF impedance analyzer one will see what's really going on which surely would be close -- and that has been done by people.
@John:
One problem that can still arise with bench testing is that one can happen to miss that specific problematic capacitance, unless you have close enough spacing between the tested values. A 1-3-10 or 1-2-5-10 ratio sequence could easlily miss that critical values. This is a place where simulations can come handy and today's delevopers ususally have pretty good sims of their amps generated and optimized by "iterative feedback" from sims-vs.-measurements cycles.
Of course I completely agree that any amp shall not be sensitive to any reasonable cable parameters, but in "HiEnd" realms this is not always the case.
- Klaus