From time to time I have a need to measure a device which has UHF series connectors.
UHF series connectors are not suitable for high accuracy measurements, and the problem is not simply that they are not ‘constant through impedance’ connectors, but the availability of reasonably priced calibration parts.
A simple solution when using short interconnecting cables at HF is to:
- OSLT calibrate the NanoVNA at its Port 1 and Port 2 jacks; and
- use the e-delay adjustment in the NanoVNA firmware to approximately compensate the short interconnects.
(2) approximately compensates for the time delay or phase delay caused by the short cable but does not compensate for cable attenuation.
Let’s look at an example. Let’s say we want to measure the characteristics of the internal dummy load in an MFJ-949E ATU.
A short (nominally 300mm) SMA(M) to UHF(M) RG400 cable is used for connecting the VNA to the DUT (about $30 for two incl shipping from RFSupplier.com).
So, after performing the calibration process, the short cable is connected to the VNA, the UHF connector collar is unscrewed to give unfettered access to the centre pin, and a small piece of new crumpled up kitchen foil is used to bond the centre bin to the connector body right where the centre pin emerges from the connector body.
The length of ‘uncontrolled’ Zo is about 10mm in this case. I would not recommend using a cascade of adapters for several reasons, the dedicated cable is more likely to be reliable, to give repeatable results.
Above is the display, the arc is due to the time delay or frequency dependent phase delay between the calibration plane (Port 1 connector) and the shorted UHF connector. Of course, all connectors must be clean and in good condition.
By adjusting e-delay until the arc reduces to a spot at the left hand end of the X=0 line, we have approximately compensated the delay of the reflected wave due to the cable electrical length (ie the two way propagation time). If cable attenuation was zero, the ‘dot’ would be on the outer circle of the chart, it is just inside it, the magnitude of s11 at 38MHz derived from the displayed Zin is about 0.99 or just under 0.1dB, just under 0.05dB for the one way attenuation. So whilst compensation is approximate, the error will be acceptable for many if not most measurements where the DUT uses UHF connectors.
Above is the test setup. Make sure you tighten the both connectors, I use a torque wrench for the SMA and a pair of multi-grips for the UHF.
Above is a screenshot of the measurements, scan changed here to 1-61MHz (though the device is only specified to 30MHz visibility of its response beyond that is enlightening).
Note that all four traces are approximately correct with this adjustment, without the e-delay adjustment only one of the traces would be correct.
The approximation is more than adequate for measurement of this DUT, it is hardly a precision dummy load.
An exercise for the reader: what would the edelay need to be to compensate an s21 measurement if two identical cables were used to connect a UHF-UHF DUT?