One fundamental I review regularly with all my clients is impedance matching. The reason is that proper (or optimized) impedance matching affects so much of what an engineer does in their daily work. Optimizing impedance matching will:
- Improve the accuracy of measurements made using spectrum analyzers, vector network analyzers and power meters
- Allow for better transfer of power in a circuit (i.e. to an antenna)
- Smooth group delay effects (click on the link for a review of group delay and why it’s important)
I sometimes get a raised eyebrow when I mention the last one; not many think of using impedance matching to improve group delay but I’ll prove this point later in the post.
Impedance mismatch is one the largest contributors of error in absolute power measurements. The uncertainty of the measurement due to mismatch is defined as follows:
What does this translate to in terms of actual error based on impedance mismatches? Let’s take a look…
The first thing to understand is that you need to know the impedance characteristics of both the source and load it is connecting to (be it a cable to a power sensor, an amplifier to an antenna, etc.). This is typically given in a data sheet as the VSWR or simply SWR of most any device with a connector. Sometimes this data is presented as the “Return Loss” (RL) value.
The second important concept is the relationship between SWR and RL. This next figure shows translating between the Voltage Reflection Coefficient (which is rarely used in data sheets), Return Loss and VSWR/SWR.
“I’m getting unacceptable variances in my power measurements” is one common complaint I hear from engineers or technicians which often turns out to be the result of mismatch uncertainty. Let’s take the example from figure 2 that the source and load have a VSWR of 1.75 (RL of around 6dB). In this case, the measured values could swing from -0.623 to +0.671 depending on the instantaneous phase; if the absolute value is +10dB, you could be getting measurements with a variance from 9.38 to 10.68 (almost 1.3dB). One example where this would be especially problematic is in measuring fast loop power control on a WCDMA (UMTS) system. The specifications define the accuracy for a 1dB power control step to be +/-0.5dB. Also, when using compresses mode, how quickly the system is able to recover from a comressed frame relys on accurate power measurements. So, having poorly matched connectors anywhere along the receiver chain or using a mismatched test connection to a Radio Communcuation Analyzer, spectrum analyzer or power meter will cause no end of frustration.
The simple solution is to add a precision attenuator at the end of the worst-matched connection. I emphasize precision because it’s critical that the attenuator have excellent VSWR or else you are simply magnifying the problem, not solving it. One source for high-quality, low SWR attenuators is Anritsu Company. While using a higher value attneuator (or “pad”) while reduce the voltage reflections and improve matching uncertainty, even a small value pad will help. This is why I most always recommend including at least a 3dB pad when making measurements with a VNA and power meter as well as having some internal attenuation applied for a spectrum analyzer and signal generator.
OK, let’s talk about group delay. Literature on the subject of group delay variations due to transmission line impedance
mismatches is scarce. The placement of the attenuation is especially critical in this application. The figure below shows the group delay for a shunt 12.57-picofarad (pF) capacitor as the discontinuity at each end of a 50-ohm transmission line. As the frequency increases, the capacitive reactance shunting the line decreases causing larger discontinuities. This in turn increases the group delay peak-to-valley values. Even with the use of attenuators, the group delay shows large variations over frequency.
Once the attenuation pads are placed at the immediate ends of the transmission line, there is a significant reduction in group delay variations.
Next time you are including group delay parameter when modeling a circuit, consider use of properly placed attenuation to improve performance. If the group delay is sinusoidal in nature, this is an indication that pad placement may be to blame.
Feel free to leave comments or questions or contact me for more information on this topic.