InsertionVSWR of Chinese 1-9 balun module

This article documents an InsertionVSWR test of a cheap Chinese 1-9 balun purchased for less than <\$5 on eBay (shipped).

Above is a 1-9 balun, it would seem to be a clone of the Noelec 1-9 balun. The balun is a compensated voltage balun with the secondary centre tap grounded for these measurements.

The measurements were made with a NanoVNA-H4 v4.3 running DiSlord v1.1.0 firmware.

The 450 ohm load comprises 1kΩ and 820Ω 1% 1210 chip resistors soldered in parallel for a DC resistance measured at 450.7Ω (Rigol DM3058E).

Above is a screenshot of NanoVNA-APP showing the VSWR plot.

Above is the extracted VSWR plot.

Above is a plot of R,X at the reference plane (the SMA saver on the NanoVNA).

The measurement is end to end including the module wiring etc. The performance does not look too wonderful, I have certainly made bare transformers that are much better than this module, but that is not a fair comparison.

Wait a minute… performance is awful!

The fact the frequency for minimum VSWR is so low hints that it may be overcompensated. Let’s explore that idea with Simsmith.

Above is a Simsmith model with the magenta plot derived from the measurement s1p file, and the green plot is with a shunt capacitance of -165pF, ie 165pF removed.

Let’s remove it with a soldering iron and measure it.

Hmmm, 150pF removed, way overcompensated. The Noelec schematic shows CR1 in this position, apparently a Kyocera AntennaGuard varistor, but no part number.

Above, with the compensation capacitor removed. I would not try to improve that with a compensation capacitor, it is not needed at these frequencies, though it may help at higher frequencies.

The minimum InsertionVSWR is still not wonderful, the next suspect is the turns ratio.

The impedance transformation mid band is $$\frac{450}{36}=12.5$$, suggesting a turns ratio around 3.6:1 (note that transformation is affected also by flux leakage).

Let’s measure the inductance of primary and secondary with an LCR meter at 10kHz where strays have little effect.

Lp=21.4µH, Lp=275µH, suggesting turns ratio $$\sqrt \frac{275}{21.4}=3.6$$

There is every sign that the turns ratio is not 3:1 as implied by the 9:1 impedance ratio label, but is more like 3.6:1, and the impedance ratio more like 13:1, and that the nominal impedance transformation is 650:50.

Since the turns should be integer numbers, perhaps it is 3t:5+5t.

Chinese Quality… but of an oxymoron really!

The module could be used as a test fixture, OSL calibrated with loads on the DUT terminals, and the correction process of the VNA papers over the less than ideal transformer behavior to a large extent.