Introduction
This article presents a review of a EFHW transformer using a Jaycar LO1238 core, a pack of 2 for $8 at Jaycar stores (Australia). The LO1238 is a 35x21x13mm Toroid of L15 material (µi=1500). Boxed up, it is probably safely capable of about 5W continuous dissipation.
The design is described at 1:49 EFHW transformer using a Jaycar LO1238 core – design workup.
Implementation
Above is the internals of VK4MQ’s balun. I would not use the pink PTFE tape, the balun core is extremely low conductivity and it is doubtful the tape helps.
Note the winding layout, the turns are close space rather than spread around the core, and it is an autotransformer, a 21t winding tapped at 3t. Both of these reduce leakage inductance which is the enemy of broadband performance. The compensation cap is 100pF silver mica.
Above is a pic of the rear which adds a clamp plate to secure the balun to a mast. The bolts are not connected directly to any balun conductors.
Evaluation
To evaluate the transformer performance, the following circuit configuration was used. The transformer is an autotransformer with 21t winding input tapped at 3t, giving a nominal impedance ratio of 1:49.
The 2400Ω series resistor plus Port 2 provides the transformer with a load of 2450Ω, 49 times 50Ω. We can see from this model of an ideal (ie lossless, no flux leakage) transformer that |s21|=-16.9dB.
Measurements were made using a Agilent VNA with 12 term correction.
Analysis
Above is the graph of |s21|. It is a little lower than -16.9dB, and InsertionLoss can be calculated InsertionLoss=-16.9-|s21|.
InsertionLoss is interesting, but Loss and MismatchLoss are interesting, perhaps more interesting… more later.
Above is a graph of |s11|, or -ReturnLoss. It reveals the extent of input mismatch caused by imperfections in the transformer, particularly flux leakage and magnetising impedance.
Above is a graph of InsertionVSWR, another metric for the goodness of the impedance transformation.
I usually analyse these .s2p files with a script in Jupyter.
Above is my graph of VSWR and ReturnLoss derived from the same measurement file.
Now to the ‘other’ loss metrics.
I will use the meanings explained at On insertion loss.
InsertionLoss, MismatchLoss and Loss can be calculated from the measurement .s2p file.
Above is a chart of InsertionLoss, Loss (labelled TransmissionLoss for clarity), and MismatchLoss (the gap between the two).
Heating of the core and windings is a result of Transmission Loss. Examining the chart, readers can see that at 7MHz, there is very little MismatchLoss, the InsertionLoss is due almost entirely to TransmissionLoss (core and winding loss).
A SimNEC model for analysis
A SimNEC model was created to read the same .s2p file and plot the interesting measures calculated from s21.
Above is a screenshot of the SimNEC model.
Above are the calculated plots:
- blue: InsertionVSWR;
- magenta dots: Loss (transmission loss, core and wire loss);
- blue dots: InsertionLoss;
- red dashes: dissipation with 100W Thevenin source; and
- cyan dashes: load power with 100W Thevenin source.
Note that ham transmitters are not usually well characterised as a Thevenin source.
Summary
A nominal 1:49 transformer with InsertionVSWR<1.5 from 2.7-25MHz, and Loss<0.7dB over that range. It should be capable of about 5W continuous dissipation, so suitable for 50% duty cycle to almost 100W. It would handle high duty cycle modes like A1 Morse code and FT8 to 100W PEP, would not even get warm on 100W PEP SSB with processed speech.
NanoVNA
The accuracy of |s21| depends on the input impedance of Port 2. 12 Term correction corrects errors in input impedance of Port 2.
Note that most if not all NanoVNA variants do not do 12 term correction in the box, nor do most PC clients, though it is possible to perform a calibration and 12 term correction of measurements (eg using Scikit-RF).
There may be other ways to try to correct for input impedance of Port 2 (think about taking a s11 scan of Port 2 and importing it to the L element).
Download
SimNEC model incl data files: EFHW-LO1238-3-21-VK4MQ.7z .