Click here to go to our main page on coax
Click here to go to our main page on heat and temperature
New for June 2020: we've updated our coax spreadsheet with the ability to look at loss with temperature, find it in our free download area. The pull-down menus of conductors now include temperature coefficients of resistance (TCRs). In some cases we could not find a value so you will get a "no data" error for weird conductors like carbon. Happily, aluminum, copper, and silver all have TCR data.
While we were at it, we check our coax temperature model against a Microwave Office (MWO) model, shown below. In this case we used rho for copper. In MWO for some reason rho is referenced to rho for gold so we plugged in 0.686 which is the ratio of the two values. The TCR is 0.00404 ohms/ohm-C, and the equation for "RhoCopperTempC includes the temperature effect which is swept by the SWPVAR list of -54, 25 and +85C. Why did we pick those temperatures? They are well-known minimum and maximum values for military airborne equipment, with room temperature in between.
The diameters of the coax in this case are for 0.047 inch semi-rigid coax with pure PTFE dielectric, but to add confusion we used millimeters for length units. Note that in many cases, semi-rigid cables use a foamed PTFE and the result will be less loss because the center conductor becomes larger in cross-section. Although the spreadsheet will allow you to enter surface roughness for center and outer conductors, we assumed perfect surfaces.
Here is a plot of loss for the MWO coax model at the three temperatures. More accurately, it is a plot of S21 magnitude, so it comes out as a negative number.
Below our spreadsheet plots the same coax geometric and material set at three temperatures; here, loss values are positive. You will have to mentally invert one of the graphs to compare them, but they are very similar. Using two tools showing the same result will give you more confidence that both solutions are correct.
Why don't linear EDA software tools include temperature variation built into circuit models? Probably because microwave engineers don't demand it. Join us in bugging them about this for future releases!