Introduction
RF engineers have found ways to go around the problem of missing calibration kits for pigtails. A classic one is the port extension supported by most VNA models, which is merely a delay compensation with possibly a loss compensation. The issue with that is that the delay is not a constant, and this limits the max frequency of this method to about 2 GHz as was demonstrated by a T-check measurement discussed in our previous blog post.
Another method applicable to pigtails is to build a DIY calibration standard set consisting of open-ended, shorted and load-terminated pigtails, Figure 1.
Figure 1: DIY calibration standards
Applying these standards is SHORT-OPEN-LOAD -calibration takes the reference plane to the pigtail tip, so we don’t need to care about pigtail losses or SMA-to-cable discontinuity, as they are all part of the transmission line chain from the VNA hardware up to the pigtail tip. The terminations at the tip are the calibration standards, and we can use ¥ Ohms, 0 Ohms and 50 Ohms as the calibration models, and have the problem solved – or?
The argument given is correct per se. There are two factors that limit the applicable frequency range of this method:
- The short standard is not 0 Ohms, and the load standard is not 50 Ohms. Instead, the open standard is very close to a real open circuit.
- A typical lot of pigtails exhibits variation between the pigtails, so if you just pick three, they are more or less inconsistent.
The question is: how much does it matter? What is a useful upper frequency limit? If we want to apply the T-check method, we must build a through-standard as well, in order to carry out 2-port calibration on the VNA. Following the same idea as in Figure 1, we can make two pigtails “kiss” each other as closely as possible, and this way define a through-standard with nearly zero delay, Figure 2. So, let’s look at what we find out, and how does it compare with other calibration methods.
T-Check results
There are variations in how different VNA models carry out the 2-port calibration, but the model we used, PicoVNA 108, can automatically determine the electrical length of the through standard. In our case, the calculated length is 1.72 mm which is actually very close to the edge-to-edge distance of the two pigtail outer conductors. We then measure the PCB T-Checker, and compare the results with previously obtained results, Figure 3.
Figure 2: Close-up of pigtail DIY through standard, enclosing copper shell partly open
There are variations in how different VNA models carry out the 2-port calibration, but the model we used, PicoVNA 108, can automatically determine the electrical length of the through standard. In our case, the calculated length is 1.72 mm which is actually very close to the edge-to-edge distance of the two pigtail outer conductors. We then measure the PCB T-Checker, and compare the results with previously obtained results, Figure 3.
Figure 3: T-check parameter cT comparison due to various calibration methods
The result is interesting: the DIY kit provides a very good and flat cT response up to about 2.2 GHz, but the quality of DIY calibration rapidly degrades at higher frequencies. When we compare port extension and DIY calibration, the results clearly favour the latter, and neither is reliable above 2.5 GHz.
Conclusions
The DIY calibration kit is a very good option for pigtail calibration up to 2 GHz, but its accuracy may be severely compromised already at BT/Wi-Fi frequencies. The DIY kit quality can be somewhat improved by pre-selecting electrically equal pigtails for the kit – which may require measuring several tens of pigtails – but eventually the kit quality is limited by the termination model inaccuracy. Dicaliant calibration utilizes a carefully developed and optimized digital twin model, ensuring steady cT response up to 8.5 GHz.