For the PDF version of this article, click here.
One of the challenges faced by designers involved in contemporary radio design and debug is being able to accurately measure portions of a signal path that are not connectorized. For instance, how does one measure the insertion loss (S21) of a filter whose input end is coupled to a diplexer, and the output to an LNA? Or, perhaps, the output power and linearity of a power amplifier (PA) is in question, but it happens to connect to a diplexer on one end and a Tx driver on the other. The answer to these scenarios (and many others) lies in the careful use of a semi-rigid micro-coaxial cable; commonly referred to as a “pigtail” (Figure 1).
Semi-rigid micro-coaxial cable can be obtained from many sources (www.micro-coax.com) in many characteristic impedances. For 50 Ω applications, they can range from small and flexible [8mil outer diameter (0.203 mm)] to stout and robust [250 mil. outer diameter (6.35 mm)]. What makes these cables so versatile, however, is that they can be soldered directly into a signal path and used as a temporary means of measuring key parameters — even if the circuit was not designed to accommodate such measurements.
Hence, if the insertion loss of a SAW filter was in question one would first break the circuit on the input and output sides of the device. Then, two micro-coax cables would be soldered to the board, allowing a simple two-port test of the device. This measurement is a true accounting of how the device is performing on this particular board. If, for instance the PCB layout was somehow incorrect, and S21 was more than the device manufacturer intended, it would be evident. Applying the same technique to a PA allows testing and optimization of the circuit without outside contributors present. Transmitter spurious can be eliminated from the measurement, mismatches at the input and output can be identified, layout issues become much more apparent, and a full range of tests can be applied to diagnose problems.
To get the most out of a pigtail, some guidelines should be followed. First, consider the frequency at which the measurement will be made. A handmade cable with coax connected to an SMA edge-mount connector may be acceptable at frequencies up to 2 GHz. Above 2 GHz, these become unpredictable in return loss and impedance due to the SMA to coax transition (Figures 2a, 2b). For higher frequencies, a ‘sealed’ cable assembly is probably more appropriate (Figures 1, 3). Sealed cables can sometimes be found at local used parts and equipment dealers. These assemblies, when cut in half, yield two pigtails each.
While keeping overall cable length as short as possible minimizes insertion loss, it is also extremely important to minimize the amount of center conductor protruding from the cable end. Any excess here will degrade return loss of the pigtail significantly.
Another item to consider when setting up for a measurement is the dc voltage level where the pigtail will be connected. Keep in mind that many spectrum analyzers (and other equipment) can be damaged by dc. Therefore, always try to break the transmission path so as to place a dc-blocking capacitor between the pigtail and the device under test.
Finally, don't be shy about grounding. If there is solder mask in the way, get out the old Xacto knife and make room. The more shield to ground-plane contact the better, for both RF reasons and mechanical. To have the ground connection break-loose during a measurement would be more than a nuisance; it could mean a pulled PCB trace or broken blocking cap.
With these guidelines, some micro-coax, and a little creativity, the hidden parameters of your transmission path no longer have to be a mystery. These inexpensive tools allow us to optimize impedance matches, locate previously unsuspected insertion losses, isolate portions of a system, and genuinely ‘know’ a radio design better than ever.
ABOUT THE AUTHOR
Tracey Chavers is an associate RF applications engineer with Maxim Integrated Products. He holds an A.S. in Avionics Technology and is pursuing a BSEE part time.