I apologize that I post this Oscilloscope/TDR measurement question here within the NWA section, but I think this question is closer to the NWA topic.
For some weeks now, I'm glad to have an Agilent 54754A based TDR in my lab to analyze PCB structures. On one PCB I analyzed the transition from the PCB wave guide into an RF connector and found a positive bump indicating a more or less inductive behaviour for this section. The system allows the positioning of two markers and calculates the excessive reactance in between these markers. This is really valuable to determine the correct compensation, or to find a way to reduce the excess. For the section of interest I get round about 630nH of inductance if I place the markers left and right of it. So far so good:
Questionable to me is the behaviour that I see if I move the markers closer together to cover only a part of this inductive discontinuity. At a certain point, when the left quarter of the discontinuity is outside the markers the display jumps over and starts showing fF and further in pF values. I'd like to make clear, that all the curve in between the markers is clearly >50Ohm.
The other way around it happens in the following screenshot:
Here the impedance is clearly below 50Ohm (39.6 ... 47.2) and we see 90pH mismatch.
The system is in the market for quite some time and I checked I use the latest software from Aug2011. I'm still searching the solution in between screen an chair fo course.
I wonder if somebody has a similar system available, or has a good explanation for this observation.
I would guess statement 1 is not correct. Presuming this is a step response, to associate a step response with C or L, it must go up and down. If it only goes up, or only goes down, then it is no an indication of reactance but rather change in resistance. So your choice of markers is undersampling the data and makes the results somewhat difficult to understand (or to put another way, too narrow of markers makes the computation meaningless).
As I understand, this functionality is to measure the excess reactance of a discontinuity. The method assumes the discontinuity is surrounded with transmission lines of reference impedance. In the case of excessive inductance, the total inductance of the discontinuity includes two parts, the first part can be used to balance the shunt capacitance of the discontinuity and the combination is equivalent to a short transmission line with the reference impedance and with a short delay, the second part is the inductance left, and is called excessive inductance. So Lexcess = Ltotal - Lbalance = td * Zhigh - Ctotal * Z0 * Z0 = td * Zhigh - td / Zhigh * Z0 * Z0. Where Zhigh is the impedance of the discontinuity and Z0 is the impedance of the surrounded transmission lines. With this assumption to determine the excess reactance, the two markers are required to be placed on the same amplitude, this is also mentioned in a TDR application note. The detailed method can be referenced to a paper named "Measuring parasitic capacitance and inductance using TDR", I believe you can google it.