One of the pitfalls of making component measurements is using the wrong impedance measurement technique.
In this post I am using a NanoVNA-H4 and measuring a 1K and a 3K resistor to see how accurate the resistance can be calculated by connecting it to CH0 with a SMA test jig (shown in previous posts) and doing a S11 measurement. The NanoVNA hardware measures the reflection coefficient Gamma (¦£) and firmware on the device calculates impedance by the following equation Z = Zo * ((1 + ¦£) / (1 - ¦£)). Zo is 50 ohms and Z and ¦£ are complex numbers. Z calculated using this method is in the form R +/- jX which is known as a "serial impedance" representation and can be plotted on the NanoVNA device by selecting the appropriate traces. A plot of the 1K SMD 1206 resistor using this standard method is shown below. Actual DC resistance on a DE-5000 LCR meter was 997 ohms. Note how the calculated value is very close to 997 ohms at 10 kilohertz but quickly drops to 950 ohms at 250 MHz. At first glance it seems that the NanoVNA cannot do a good job of measuring this component at higher frequencies and some users might put this down to being so far away from the 50 ohm system impedance of the NanoVNA hardware.
However a real world resistor is not a pure resistance but has inductance in the leads and a stray capacitance in parallel with the resistance. This is shown in the simplified model below. In the case of a small SMD resistor this lead inductance is very small and the capacitance across the resistor is the dominate reactance. The first plot was a "series impedance" measurement of the 997 resistor which is a standard way of plotting impedance but it does not represent the actual physical component. An alternate form is the "parallel impedance" which is R // +/-jX and unfortunately is not available for plotting on the NanoVNA itself. Some PC packages like NanoVNASharp MODv3 and NanoVNA app by OneOfEleven do have this calculation capability and it is very useful. The parallel impedance plot below takes into account the parallel capacitance across the resistor and we can see the resistance is fairly consistent across the frequency range and deviates by a maximum of 10 ohms from the 997 ohms at DC. The next plot shows the capacitive reactance and if we calculate it by C= 1(2*pi*freq*X) it varies from 0.14 to 0.15 pF which is a very small.
The last test measured a SMD 1206 3K resistor on the same SMA test jig. This part has a DC resistance of 3003 ohms. Serial and parallel impedance plots are below. In a similar fashion the serial impedance method results in a poor estimate across the frequency range but the parallel method is much better and is off by up to 100 ohms maximum or 3.5% which is fairly reasonable.
I hope this post provides some insight into the measurement capability of the NanoVNA-H4 at higher impedances.
Roger
Roger Need
