LZ1AQ active loop performance evaluation

Created: Jan 2025. Modified: Feb. 2025.

-----------------------------

Introduction

The LZ1AQ is an active loop aerial for the HF band. The associated web-page is very thorough in detailing design considerations and calculations [1], but is jarringly quiet on the preamp's performances. This information shortfall has prompted some third parties to evaluate and report some important parameters such as gain and input impedance, but they are far from exhaustive [2, 3].

As knowledge of the preamp's performances is necessary for (a) checking homebrewed prototypes, (b) improving on the original design and (c) comparing against competing designs such PA0FRI [4], MA0AYF [5], and MLA-30 [6], etc., this article aims to bridge this knowledge gap. To this end, we performed simulation and measurements of the said preamp and then, collates the results here. Finally, to facilitate comparison, we tabulate the most important results of popular designs such LZ1AQ, PA0FRI, MA0AYF and MLA-30.

Material and method

The evaluation samples were built according to the circuit in fig. 1, but with the following deviations:

1. Replaced the original LAN socket with a coax connection.

2. Instead of the original supplying power via another LAN wire pair, the preamps were phantom powered via the output coax connection.

3. Output transformer L4-6 wound on a slightly bigger toroid than original. We used what was available in the junk box.

4. Instead of the specified plastic packaged PN2222A, we substituted with metal can 2N2222A.

5. Instead of the on-board voltage regulator, the pre-amp's supply is regulated externally.

Fig. 1: preamp circuit (reproduced from LZ1AQ website)

The prototypes were assembled on a double-sided FR4 PCB measuring 50 x 50 x 1.5 mm. All components are through hole.

Fig. 2: unpopulated PCB (left) and assembled prototype (right)

Fig. 3: part-list

To predict the performances and to facilitate circuit tweaking in the future, we developed an equivalent circuit model for simulating RF performances (fig. 4). The 2N2222A Spice model originates from Motorola, and is dated 1993. The RLC component models do not include parasitics - hence, they will become less accurate at higher frequencies. The PCB traces are not modelled as we think they are insignificant at HF.

Fig. 4: Simulation circuit

During testing, the experimental samples are phantom powered via a homebrew biastee (fig. 5). To measure the gain, the preamp input is connected to a dummy aerial (aerial simulation network) that replicates the impedances of a 1m diameter loop [7]. Other measurements omit the dummy aerial and are directly fed with 50 ohm sources.

Fig. 5: Test setup for gain showing dummy aerial connected to the DUT input and biastee to the output.

Results

Measured and simulated gain show good agreement. Modelled gain (dotted black in fig. 6) is slightly higher than experimental because the former doesn't account for various circuit losses. The model's maximum error is < 3 dB.

The experimental gain is very consistent between samples as demonstrated by the almost overlapping solid traces (fig. 6). The experimental gain varies <1 dB between the 3 samples.

The gain peaks at ~28.5 MHz due to the combined resonance of the dummy aerial and preamp's input filter. Curiously, below the peak frequency, the gain decreases at a rate of ~9 dB/ octave, resulting in very low gain in the lower half of the HF range - this makes the LZ1AQ an anomaly among loop preamps (see fig. 13 in appendix).

Fig. 6: Measured and simulated gain show good agreement (sample size, n = 3). The experimental samples are also very consistent with a gain variation of <1 dB between samples.

Measured and simulated input impedances (S11) don't agree perfectly, but they do exhibit the same trend over 1-30 MHz (fig. 7). Again, the experimental traces overlapped, pointing to excellent consistency between samples.

Fig. 7: Although measured and simulated input impedances don't agree perfectly, they show the same trend over 1-30 MHz. The inter-sample variation (n = 3) is negligible

Measured (solid black) and simulated (dotted red) output impedances (S22) agree well over 1-30 MHz (fig. 8).

Fig. 8: Measured (solid black) and simulated (dotted red) output impedances (S22) showing reasonably good agreement.

The simulated 1-dB gain compression (P1dB) has a 1.5 dB error at 5 MHz which then gradually worsens until reaching 4 dB at 30 MHz (fig. 9). The error in the model is likely due to the simple component models which don't account for increasing losses with frequency. The experimental P1dB which exceeds 15.5 dBm is head & shoulders above the competition (fig. 12). The LZ1AQ's high P1dB confers blocking immunity.

Fig. 9: Measured (solid blue) and simulated (dotted black) P1dB. The model has a 1.5 dB error at 5 MHz which worsens to 4 dB at 30 MHz. The experimental P1dB exceeds 15.5 dBm

The modelled output 3rd order intercept point (OIP3) has almost zero error at 5 MHz, but increases to ~5.5 dB at 30 MHz (fig. 10). There are two likely reasons for the error: (a) the component models do not account for frequency dependent losses, and (b) the model assumes perfect balance between top & bottom amplifier sections, but the real-world prototypes have unmatched transistors, asymmetrical PCB layout, and / or inequal windings in the output transformer (see fig. 15 in appendix). Nevertheless, the experimental OIP3 of >29.5 dBm beats the competition by a wide margin (fig. 12). Furthermore, the LZ1AQ's relatively low gain translates to a high input IP3. The combination of low gain & high linearity makes the LZ1AQ resistant to adjacent channel interferers.

Fig 10: Simulated & measured OIP3 vs. frequency show good agreement at low frequencies but diverge at higher frequencies. This error is due the model optimistically assuming a perfect balance between top & bottom amplifier sections.

Like OIP3, common mode rejection ratio (CMRR) is highly sensitive to unmatched transistors and PCB layout assymmetry. Hence, this parameter was not simulated because the perfectly balanced model will predict an unrealistic number. The experimental samples achieve > 24 dB CMRR over 1-30 MHz (fig. 11). The CMRR reaches a peak of 33 dB at 3 MHz and then declines to 24 dB at 30 MHz.

Fig. 11: Experimental CMRR declines from 33 dB at 3 MHz to 24 dB at 30 MHz

The LZ1AQ offers a mixed-bag of performances compared to its peers. It has the lowest gain, but the highest P1dB and OIP3 of all (fig. 12).

Fig. 12: A survey of active loops shows the LZ1AQ has the lowest gain, but the highest P1dB and OIP3 of al

Conclusion

The LZ1AQ's most important characteristics have been evaluated and reported here. The performances are evaluated by both simulation and measurement. The simulation model is reasonably accurate for gain & impedances but is less accurate for P1dB and OIP3. However, future iterations can conceivably improve on the accuracy by incorporating frequency dependent losses and circuit imbalance.

Compared to the competition, the LZ1AQ has superior blocking immunity and linearity, but lesser gain.

Appendix

The LZ1AQ is an oddball among active loops because of its drastic gain decline below the 29 MHz peak (black trace in fig. 13), whereas PA0FRI, MA0AYF and MLA-30 have relatively flat responses over 8 - 30 MHz.

Fig. 13: The LZ1AQ's gain drops significantly below the 29 MHz peak, wheras the other preamps maintain relatively flat responses between 8 - 30 MHz

Initially, we used tantalum caps in C1 & C4, because they are more compact than the alternatives. Unfortunately, they have high Effective Series Resistance (ESR) which detrimentally increases the input impedance (fig. 14). Replacing with film caps solves the problem.

Fig. 14: Abnormally high input impedance Zi when C1 & C4 utilize high ESR tantalum caps. (n=3).

To validate the hypothesis that the degraded OIP3 at higher frequencies is caused by imbalance between the top & bottom amplifier sections, we inserted an inductance L3 between Q2 and L5 in the circuit model. When L3 = 0.6 uH, the modelled result approximates the measurement (fig. 15). An inductance of this magnitude likely originates from the transformer windings

Fig. 15: Simulation shows that a 0.6 uH parasitic inductance between Q2 and L5 can degrade 30 MHz OIP3 by ~5 dB. This result agrees with the measured OIP3

The error in the modelled P1dB can be reduced by accounting for the output transformer's loss. The loss is implemented by connecting a series RC network R3 in parallel with the output transformer (fig. 16) - of course, this is a massive over-simplification of the transformer loss.

Fig. 16: Improved model that accounts for output transformer’s frequency dependent loss using series RC, R3

Previously, the modelled P1dB was unrealistically flat over frequency. With the addition of the loss network, the modelled P1dB attains a similar slope to the experimental. As a result, the maximum error Is reduced from ~4 dB to ~2 dB (fig. 17).

Fig. 17: Model’s max error Is reduced from ~4 dB to ~2 dB

References

[1] C. Levkov, “Wideband active small magnetic loop antenna”, 2011. [Online] Available: http://www.lz1aq.signacor.com/docs/wsml/wideband-active-sm-loop-antenna.htm

[2] M. Ehrenfried, G8JNJ, "Broadband loops", [Online] Available: https://web.archive.org/web/20230327034702/https://www.g8jnj.net/loop-inductance

[3] E. Sharp, "LZ1AQ Loop Amp and Bias Tee", [Online] Available: https://drive.google.com/file/d/1iSsCVMf3OHZnzMIGeOVt3Q-_P-8oBD33/view?

[4] F.H.V. Geerligs, PA0FRI, "Active loop antenna for reception", 25 Jul. 2023. [Online] Available: https://pa0fri.home.xs4all.nl/Ant/Actieve%20ontvangst%20antenne/Active%20loop%20antenna%20for%20reception.htm

[5] Des, M0AYF, 'Active loop antenna for HF', transcribed by M0LMK. [Online] Available: http://www.m0lmk.co.uk/2015/02/14/active-loop-antenna-for-hf/

[6] MegaLoop MLA-30+. Can be bought from various online shopping platforms

[7] "Aerial simulation network / dummy aerial for active loops", [Online] Available: https://sites.google.com/site/randomwok/Home/electronic-projects/aerials/aerial-simulation-network-dummy-aerial-for-active-loops

Page updated

Google Sites

Report abuse