Cores FAQ

Can I use a single RF line with an isolated ground as a balanced pair? 
What are the inductances for GP membrane ISS loads? 
What is the dielectric constant of the polyimide in PPD cores?

Can I use a single RF line with an isolated ground as a balanced pair?
This is not recommended. Our standard method for handling differential/balanced signaling is to provide two independent RF lines (for example RF1+ and RF1-) that are delay matched. Standard delay mismatch is ±10 ps. Custom delay matching of balanced line pairs typically provides delay mismatch of less than ~±1.5 ps. 

Using independent signal lines for a differential pair does not force the waveform balance of an ideal, balanced transmission line. In practice no such ideally balanced line exists. Standard test system practice (and the method we normally use for Pyramid Probe cards) is to provide two matched independent signal lines and rely on the terminations and signal sources to force signal balance. Ground currents will approximately cancel, providing most of the benefit of differential signaling.

The single RF line is an asymmetric transmission line - the signal and separate ground - and will have differing parasitic capacitances loading the signal and ground conductors. This results in distortion of the symmetry of the waveforms and loss of balance. A separate 50-ohm line over a common ground performs better.

In our experience, the simplest and most robust solution is to use independent 50-ohm lines right to the IC. Any attempt to transition to a balanced transmission line in the probe causes more harm than good. The differential drive is provided externally by the signal sources or by a balun (balanced transformer).

For uncoupled 50-ohm transmission lines, both even and odd mode characteristic impedances are 50 ohms. This means that the differential signal is in an environment equivalent to what it experiences in a balanced transmission line. The coaxial grounds are connected to the common analog ground at the membrane. The balun or test equipment provides the connection at the other ends of the cables. Net ground currents will be equal and opposite for equal and opposite signals, and will cancel.

What are the inductances for GP membrane ISS loads? 
This depends on the orientation of the resistor and probe. Using 150-µm pitch GSG probes we get:  

R extending away from probe Lterm = -29 pH

R returning under probe Lterm = -77 pH          

R across probe contacts (either) Lterm = -47 pH 

What is the dielectric constant of the polyimide in PPD cores?
The simple answer is 3.5. This is subject to some variation, for reasons described below.

Polyimide absorbs water. Published numbers indicate up to 2 or 3% absorption at 100% humidity. Published variations in dielectric constant indicate about a 10% increase in relative dielectric constant at maximum water absorption.

Since impedance changes with the square root of dielectric constant, a 10% increase in dielectric constant results in only a little over a 3% change in impedance.

BUT, people don't usually test wafers underwater. Within the normal operating environment of a test floor, the humidity variance, hence the impedance variation, is small. So far, not a single customer has observed this to be a problem. In fact, this is why people calibrate. All test instrumentation, cables, sockets, etc., have similar variations with humidity and temperature. (Admittedly though, polyimide may be worse than most others with regard to humidity.)