here "reflection" means a one-port device with high reflection (ex. ro>0.95) and known phase of reflection. the reflection phase should be known with a (lambda/4=pi/2) accuracy.
the reason should will be clear if you draw the NA error modeling flow diagram (see: **broken link removed**)
In the TRL method, you need to solve some nonlinear euqation to obtain the phase of the reflection standard. This equation could have two solutions from a pure mathematical point of view. You need to specify which one is correct (i.e. physical) to determine uniquely the correct "error model".
I think they are talking about the reference plane.
In other words, the reflection standard should not
be too long. And apparently 1/4 wavelength is
good enough for errors.
consider the following analogy
x=Sqrt(-1) has two roots: x=+j and x=-j in mathematical sense, but only one root is physical and the methdo needs your help to choose the correct root. If you say the root should have a phase >0 and <180, then it picks +j....
Hi,
Thank you very much for all your reply.
Still confused: This reflection standard (only single length) actually cover very wide frequency range. so the "1/4 wavelength", the wavelength here is corresponding to which frequency? Thx