Back in 2023, I published a 3D-printed tape measure Yagi antenna design för “fox hunting” on the 2m band (144 MHz). While the antenna worked (and still works) fine, I realized that it would be better to have the tape measure pieces mounted with the broadside towards the front rather than up. That way, they more gracefully bend when they hit branches and trees.

Another change I wanted to make was to the spools one can roll upp the antenna elements onto. They were previously designed to be printed as two separate pieces and glued together. While this makes the 3D printing easy without any need for supports, the gluing is a bit messy. I therefore wanted to come up with a design that does not require any gluing. This instead requires some support material, but by printing the clamp that goes around the spool “in place”, most of the supporting is done by the clamp.

Yet another novelty is that I wanted it to be possible to disassemble the antenna to make the pieces short enough to fit in a reasonably sized suitcase. I therefore cut the pipe at a location that ends up in the middle of the holder for the driven element and let the holder also clamp the two pieces of the pipe together.

Here is a link to a zip-file with the 3D models for printing:

Link to a zip-file with 3MF models of the parts:

Here is a drawing with dimensions:

The shortest element is the director (towards the main lobe of the antenna) while the longest is the reflector. The “driven” element, connected to the receiver, is the one in the middle.

According to YagiCAD, the radiation pattern of the antenna is as follows:

For the boom, I use a piece of 16 mm outside diameter PVC electrical conduit pipe (“VP-rör” in Swedish).

For the elements, I use 25 mm wide steel tape measure from Biltema, e.g. part number 16-2931.

Throughout the design, M3 screws and nuts are used to hold the pieces together. Stainless screws and nuts should ideally be used to prevent corrosion over time.

The front element (director) is held by these 3D printed pieces that clamp onto the pipe as well as to the tape measure:

The rear element (reflector) is held by similar parts. The main difference is that the tape measure is reversed relative to the end of the pipe compared to how it is at the front . The reason is of course to hold the tape measure in the orientation that makes it bend most easily when one runs forward and it collides with some object.

The parts holding the driven element are a little more complex:

The red piece serves both as a clamp for the coax cable and as a cover to mechanically protect the connections between the coax and the wires leading to the antenna elements. Under it, there is a pocket for making those connections:

There are channels for routing the wires from the pocket to the antenna elements as can be seen in this cross section:

The coax (RG174) going into the pocket and connecting to the antenna elements may look like this with the cover/strain relief removed:

I did not use a balun between the differential antenna elements and the unbalanced coax, but it seems to work fine anyway. The pocket is otherwise big enough to fit a little PCB with a balun, or some other balun solution. Maybe I will try that in the future.

These parts serve double duty as they both hold the receiving antenna elements, but also can hold the boom together if it is split up into two pieces to allow it to be disassembled into two segments:

To reinforce the joint between the boom segments, there is a little plug with a ridge around the middle that fits into the two pipe sections:

The flat part of this piece lets it be easily printed without supports. There is no need for it to be a complete cylinder.

I also made some modifications to the handle. It is now a little simpler, perhaps lighter and the clamp pivots around a 5 mm steel pin. (An M5 screw might also work.) There is still a mechanism with a wedge pulled by a threaded M4 rod to tighten the grip around the pipe:

For the knob to hold on securely to the threaded M4 rod, two nuts should be tightened against each other inside the knob:

The grip around the pipe is however not as secure as one might hope. The pipe is slippery, so to avoid having the handle easily slip and rotate around the pipe while running, I made a little peg that can be glued onto the top of the pipe (preferably using epoxy). The gap in the handle clamp fits around the peg and prevents the handle from rotating. There are two holes in the peg where one can insert pieces av 3D filament to reinforce the glue joint. Corresponding holes obviously need to be drilled into the pipe as well.

When the antenna is used, the handle is pushed forward so that the peg ends up in the gap and prevents the handle from rotating. During storage, the handle can be slid away from the peg and rotated to the side to make the antenna more compact.

The spools and clamps that allow the tape measure elements to be rolled up for storage are perhaps the most complex parts in this design. As mentioned previously, they can now be printed as one piece, no glue required, but at the price of using supports during printing. So there is still some manual work to do to remove the supports, even if the messier gluing is avoided. By printing the clamp that keep the the tape measure from unrolling from the spool in place around the spool, it will serve as part of the supports and save some material and print time.

The spool and clamp together look like this:

Six sets of spools/clamps are needed for a complete antenna.

To help when rolling up the tape measure onto the spool, there is also a hexagonal pin that fits in the center hole of the spool:

Only one such pin is needed per antenna as it can be moved between the spools.

The short M2.5 screw that is used as a hook for the tip of the tape measure when rolling it up is barely visible above.

With supports (green), the spool and clamp may look like this in the 3D slicer:

All other pieces print without support.

A spool in use (with the pin in place) is shown below:

I printed the spools using red filament to make them more easy to spot in most environments and thus reduce the risk of losing them.

The tips of the tape measure elements need to be chamfered at a shallow angle to fit into the spools and a hole should be made near the tip to fit the M2.5 “hook” in the spools. The cut edges easily become very sharp, so to protect from injuries, they should be covered by sports tape:

A photo of the entire antenna with a receiver in a box also attached to the boom is shown below.

The receiver with the elements rolled up looks like this:

I maintain a set of simple ARDF (amateur radio direction finding, or “fox hunting”) receivers of the type Compact 3,5F from OK2BWN for use by beginners. A problem with some of the receivers is that the gain cannot be set low enough, which means that the audio becomes unbearably loud close to a transmitter. I could not find any schematics online, so I decided to reverse-engineer the receiver so that I could figure out what to change to allow lower gain.

The receiver is built on a single-layer board using some through-hole components on the top side and surface mount components on the bottom. The discrete SMD components are large (1206 size), so the resistors have their values printed on them. This all makes the tracing of the circuitry relatively simple. I photographed both sides, flipped the photo of the top side and overlayed them. Then it was “just” a matter of sketching out the circuitry, a process that involved a couple of false starts before it all converged to something that looks like a direct conversion receiver.

Even before starting, I had a good idea of the general architecture as the receiver contained a single SA612 mixer with an analog VCO and an LM386 audio amplifier, but no crystal or ceramic IF filters (so it was not a superheterodyne). Since there was only one mixer it also could not be an image reject receiver, hence receiving both the desired and undesired sideband, which is also evident when tuning the receiver as a CW signal appears for two different tuning settings close to each other. Another thing that was obvious from using the receiver and looking at the parts inside, was that the signal from the E-field antenna was combined with the signal from the H-field antenna by a few turns on the ferrite core when the “antenna selector” switch was pushed to power the E-field amplifier located on a separate board.

Below are the photos of the top and the bottom of the board, as well as a combined photo where the flipped top side components are shown somewhat transparently over the bottom side photo.

Below is the schematics I came up with. As far as I know, it is correct, but I give no guarantees.

An annotated photo of the bottom of the board is shown below.

I have not bothered to desolder and measure unmarked components like capacitors, so apart from the electrolytics, I do not know their values. I wrote 100 nF on capacitors I think only serve the function of DC-blocks, supply decoupling or low-pass filtering of the varactor bias, but this is just a guess. They could have some other value, although that is unlikely to be critical. The capacitors in series with the varactors might theoretically have some low value to adjust/reduce the tuning range, so there the 100 nF guess could be more wrong.

There are a few mystery components in the circuit. The four-pin through-hole part (Q1A, Q1B in the schematic) near the antenna is unmarked, except for a white stripe. Based on its place in the circuit and the components around it, I have guessed it is a double JFET in cascode configuration. While there are indeed dual JFETs available today, I could not find one with this package. My guess is that it is an obsolete part. Also, the varactors are unmarked, but the conclusion they are varactors is more solid than the guess regarding the dual JFET.

A JFET cascode amplifier is a good choice as a low-noise-amplifier for a ferrite antenna as it has high input impedance and low noise. Here it has a tuned drain circuit which increases the gain at the cost of some added complexity (and risk of oscillation?). The Q is limited by the 1.5 kohm input impedance of the SA612.

I suppose having a tuned RF amplifier was necessary to get enough gain for weak signals. The other gain elements are the mixer (nominally 14 dB) and the LM386. The RF amplifier is the only adjustable gain element in the receiver, which is perhaps an unusual choice. The input gate is DC-biased to 0 V via the ferrite antenna while the source is connected to a decoupled (C35) resistive network around the gain potentiometer, which allows changing the operating point of the JFET to select a suitably steep part of the ID-VGS curve to get the desired gain.

Adjusting the minimum gain

Now that we have the schematics, what do we need to do to allow more attenuation/less gain? The obvious place to look is near the existing gain control circuit. When potentiometer R4 is at its maximum counter-clockwise (CCW) position, the gate-source of JFET Q1A is maximally back biased to operate at the shallowest I_D-V_GS point and thus lowest gain. But we need it to be lower still.

V_GS(off), i.e. the (negative) voltage from gate to source of a JFET required to turn the transistor fully off is not a well-controlled parameter. It can easily vary by a factor larger than two (sometimes larger than five) between devices with the same part number. As discussed the way gain is adjusted in this receiver is largely by adjusting V_GS, so it is no surprise that different receivers will have different minimum gains.

To reduce the minimum gain for a receiver, it is fortunately quite easy to make a modification that results in a more negative V_GS when the gain pot is maximally turned CCW. Just reduce R1 that together with the gain pot R4 (and R3 and R6) forms a voltage divider from +5V. A lower R1 value means the source will be biased to a higher voltage when the gain pot is at max CCW, but it will have very little effect when the pot is at the opposite end.

I found that parallelling the 33 kΩ R1 with 100 kΩ to get 24.8 kΩ resulted in a decent minimum gain for most receivers I modified. One receiver that was not quite as bad received 130 kΩ across R1 (26.3 kΩ total).

Below is a photo of a board with a blue 100 kΩ resistor soldered on top of the 33 kΩ R1 that was there originally.

Further comments and observations

The antenna is wound using two parallel and mirrored windings. This is supposed to eliminate any possibility of direction error. A direction-finding ferrite antenna should have a null exactly when the magnetic field-lines enter perpendicular to the core, but if there is a single winding on the core with some distance between the start and end of winding, there will be one effective turn over a part of the broadside ferrite core, leading to a slight offset in the null. How big a problem this is in practice is a bit unclear, but it generally is considered good practice to use two windings in parallel, wound symmetrically in opposite directions to cancel this parasitic broadside turn.

There are three varactor-tuned circuits in the receiver that need to be simultaneously tuned to the same frequency as set by the tuning knob: The antenna, the low-noise amplifier and the local oscillator. Getting all three circuits to agree on the frequency for all settings of the tuning dial might have been a challenge when designing the receiver. There is a trimmer capacitor across the antenna that can be adjusted to ensure the antenna agrees with the LO for at least one frequency and the two inductors are also trimmable to allow alignment on at least one (mid-band) frequency. And maybe it is not a big problem if the stages are slightly out of tune near the band edges.

The shielding of the receiver is a bit relaxed. There is a copper clad board acting as shields on each side of the circuit board and ferrite antenna. The shield at the front panel is only connected to the electronics via a thin wire and the shield in the bottom of the box does not seem to have any secure connection at all to the rest of the circuitry. This is not good shielding practice, but perhaps it is good enough at 3.5 MHz. The PCB seems to have provisions along the edges for better shield connections, but maybe this turned out to be unnecessary.

There are shield walls above and below the ferrite antenna, solidly soldered to the copper clad board. Having a good electric shield around the ferrite antenna is crucial for getting distinct nulls in the antenna pattern.

The way the front-back ambiguity of the figure-eight pattern of the ferrite is resolved in this kind of receiver is a bit shaky. When pressing the “antenna selector” button, the E-field antenna signal is injected into the ferrite core and picked up by the ordinary receiver chain. If it is in phase with the signal from the H-field antenna, the total signal gets louder, but if there is close to a 180 degree phase difference, the signal gets weaker, at least if the signals have close to the same amplitude. So by comparing the signal strength with one broadside of the ferrite towards the transmitter with the signal strength with the receiver rotated 180 degrees (which gives a 180 degree phase shift/inversion of the H-field signal, but does not affect the E-field signal), one can figure out if the transmitter is ahead or behind. The big problem however is that the signal strength from the E-field antenna is highly dependent on e.g. the height above ground, so it may not have close to the ideal amplitude and thus not be close to cancelling out the H-field signal when they have opposing phase. And particularly in close vicinity of the transmitter (in the near-field), the two parts of the field may have an unexpected phase and amplitude relationship. This may make it hard to distinguish which orientation of the receiver produces the strongest signal. Trying this at different heights, from knee level to above head level, to get different E-field strengths is often necessary. It would be better if the phase could be compared more directly, but this would add significantly to the complexity of the receiver, which is not an option for a low-cost receiver like this.

The LC-oscillator built with the (now obsolete) SA612 mixer is a configuration I have not seen before. It is not one of the topologies suggested in the data sheet.

A somewhat dubious component choice is the 10-V rated electrolytic C26 connected across the 9-V battery. This is a little too little voltage margin for my taste. A cap with a rating of at least 16 V should be used in that position in my opinion.

There is a through-hole 78L05 5-V regulator on the board, powering parts of the circuitry that benefits from a stable (not dependent on battery discharge level) voltage and also circuits that are not 9-V-tolerant.