Version 4
Figure 1: RX Path v4 Flowchart
The fourth revision of the RX path circuitry is mostly a refinement of the previous revision. The overall RF path remained very similar, with slight changes in components and ordering. The enclosure selected for this design resulted in a smaller and easier-to-use board than revision 3. The major changes introduced in this fourth revision are:
- RF component ordering was modified so the first RF switch is ahead of the amplifiers. This is to ensure the amplified signal is always low enough that the amplifiers operate safely.
- Several components were swapped out for equivalent components that were more affordable and more suited for this application. The limiters were changed to pin diodes, the attenuator was changed to a lower-power model, and the second amplifier was changed to a cheaper gain-block amplifier (instead of a low-noise amplifier).
- LED indicators of power levels and TR logic state were added
- The board was designed to fit within a smaller extruded enclosure, with all I/O connections interfaced through one edge of the board
Testing of the circuit was generally successful, and the circuit worked as expected. The only issue that arose was regarding the input limiter - the selected limiter didn't clamp the input power to a low enough level to protect the circuit. Changing the input limiter to the same as the output limiter solved this issue.
Following successful testing, v4.1 boards were installed at the Saskatoon radar site for all receive paths. On future site visits, v4.2 boards will be installed at the remaining SuperDARN Canada radar sites.
Figure 2: RX Path v4.0 3D Render
Figure 3: RX Path v4.0 Board Renders
Figure 4: RX Path v4.0 Enclosure Front Panel
Circuit Schematic: receive_path_schematic_v4.0.pdf
KiCad Project: v4.0
Noise figure (NF) is the measure of how much a device degrades the signal-to-noise ratio (SNR). Low-noise amplifiers typically have an NF value close to 1.0 dB, while other amplifiers may have NF values greater than 4.0 dB. To determine how important the NF value of each amplifier is to the overall NF of the circuit, the following equation was used (Friis' formula):
where
Three different amplifier configurations were tested:
- Case 1: Both amplifiers LNAs:
- Overall NF: 1.10 dB
- Case 2: LNA first, non-LNA second:
- Overall NF: 1.11 dB
- Case 3: non-LNA first, LNA second:
- Overall NF: 4.00 dB
The first amplifier has the greatest impact on the overall noise factor of the circuit. Since a high-quality low-noise amplifier is pricier than a regular amplifier, we can save a bit of money by using only one low-noise amplifier per circuit.
Design change: The second amplifier was changed from the LHA-13LN+ low-noise amplifier to the GALI-4+ gain block amplifier. This reduced the overall gain from the pair of amplifiers from 48 dB to 39 dB. With a fixed attenuator, the gain can be reduced to ~30 dB, the same as the previous system. The price reduction is also significant - from ~$20 CAD down to ~$4 CAD.
Some cheaper alternative components were found for the limiters and attenuator. These alternatives perform just as well for our application; the price difference is just due to either lower power specifications or different manufacturers.
Input limiter: The RLM-521+ limiter from Mini-Circuits was replaced with the CLA4610-085LF PIN diode limiter. The limiter was selected to limit the circuit's input power to similar levels as the original limiter. The cost decreased from ~$40 CAD down to ~$3 CAD.
Output limiter: The RLM-33+ limiter from Mini-Circuits was replaced with the SMP1330-005LF PIN diode limiter. The limiter was selected to ensure output power from the circuit was within the specifications for the N200s. The cost decreased from ~$25 CAD down to ~$1 CAD.
Attenuator: The QAT-8+ attenuator was replaced with the GAT-8+. The main difference between the two devices is power rating - QAT-8+ was rated for 1.4 W, while the max power the attenuator would see within the circuit is only 250 mW. The GAT-8+ attenuator is rated for 500 mW instead. The cost decreased from ~$20 CAD down to ~$6 CAD.
Since an RF switch will be placed ahead of the amplifiers in v4.0, a proper investigation is needed to determine how video leakage (i.e. switching noise) from the RF switch will impact the circuit. Video leakage is listed on the datasheet for RF switches - for the M3SWA-2-50DR switch used in v4.0, the video leakage is specified as 42.5 mVpp (-23.4 dBm in a 50Ω system).
Test Setup: To test a switch at the front of the RX path circuit, the ZYSWA-2-50DR SMA-packaged switch was used with the v3.0 circuit - same setup as the last test of the v3.0 board. In this test, we are measuring the video leakage of the ZYSWA switch instead of the M3SWA switch. The video leakage of the ZYSWA switch is specified as 30 mVpp, while the leakage measured with the oscilloscope is shown below in Figure 5. The switch was controlled using the TR signal from an N200 during normalsound operation.
Figure 5: Video leakage measurements
The left image shows the leakage when the switch goes from on (no signal suppression) to off (full signal suppression), while the right image shows the switch going from off to on. The characteristics for both cases are as follows:
- Switch On side:
- Video leakage has an amplitude of 11.2 mVpp. After both amplifiers, the leakage is amplified to 1.80 Vpp (or +9 dBm in a 50Ω system).
- All video leakage and transients fade away within 300 ns. The Largest spike fades within 100 ns.
- Switch Off side:
- Video leakage amplitude varies between 10 mVpp and 50 mVpp, and fades away after 15 μs.
- After both amplifiers, the leakage is amplified to 6.0 Vpp (or +20 dBm in a 50Ω system). Peak leakage fades within 2.5 μs.
The switch-off leakage occurs within the transmit window, so it won't affect the received data. The switch-on leakage would affect the data, but the spikes fade very quickly, so their impact should be small. To see how the video leakage would appear at the output of the circuit, the signal was measured with a test RF signal input of 10 mVpp (-36 dBm) at 12 MHz, shown below in Figure 6.
Figure 6: Video leakage observations at the circuit output within a typical RF waveform
The left image shows leakage within a received RF signal on the switch-on side, while the right image shows the switch-off side. As expected, the video leakage is only visible on the switch-on side (evident by the single spike), while on the switch-off side, the leakage is suppressed by the second RF switch. Thus, video leakage only affects the first 100 ns of the received signal when the RX path switches from TX to RX mode.
Conclusion: Video leakage has minimal effects on the received data. The bigger concern would be potential damage to RF components after the video leakage is amplified by the two stages of amplification. From the measurements, this leakage is amplified to a maximum of +20 dBm right before the final RF switch. Since the RF switches are rated for +30 dBm input, no damage should occur. Thus, the RF switch video leakage is not a concern for this design.
LED Indicators: The following LED indicators were added to the circuit.
- Two LED indicators for the +/- 5V power rails were added to be visible outside the enclosure. The following board-mounted LEDs were chosen, which mount a typical PIN-diode LED at a right angle: WP934EB/2SURDK
- Two surface-mount LED indicators showing the TR signal state. These LEDs will only be visible when the enclosure is open, and will be used for testing/debugging. The colours are blue for TX mode, and green for RX mode.
The resistors to modify the LED brightness would be determined during testing.
Jumpers: The following 2.54 mm pin jumpers were added to modify the TR signal input. The jumpers are 3 vertical pins where the centre pin is shorted to one of the outer pins to select the state.
- One jumper selects what source the TR comes from - either SMA or differential input. This allows components for both options to be populated on the board. In regular operation, a jumper must be installed to select one of the two inputs.
- The other jumper allows TR to be manually fixed high or low (TX or RX mode). This jumper is only for debugging and should be left unconnected during regular operation.
A pull-up resistor was added to the TR signal as well, so the circuit is pulled into RX mode by default.
For v4.0, a different enclosure was selected: 1455L801BK.
Figure 7: RX path v4.0 enclosure
This enclosure has several advantages over the box + lid enclosure chosen for v3.0:
- The extruded enclosure design allows the PCB to slide into the enclosure and fit snugly, allowing for a smaller enclosure
- This design also allows for easy input/output interfacing with the edge of the PCB
- The enclosure also has a removable top panel, allowing the circuit to be seen while it is mounted within the enclosure.
To interface the I/O of the circuit with the enclosure, the following changes were made:
- SMA connections: extra-long through-hole right-angle SMA connector
- Differential TR signal: 2-element Molex Nano fit connector. This ensures polarity isn't flipped accidentally, and clips into place.
- Power input: Barrel jack with a locking mechanism. Barrel jack ensures power input isn't shorted, and locking plug ensures the connector doesn't fall out.
A custom front enclosure face was created as a PCB (see Figure 4).
Figure 8: Assembled v4.0 board
The v4.0 PCB was ordered from OSH Park. Additionally, a stencil of the PCB was purchased from OSH Stencil for reflow soldering assembly in the lab. The enclosure front panel PCB was purchased from Elecrow.
The v4.0 board was assembled in the lab using the stencil and reflow oven. Some notes from assembly:
- The barrel power jack is not rated for reflow soldering (the casing melted). The SCLF low-pass filter also showed discolouring from the oven. All other through-hole and surface-mount components worked well within the reflow oven.
- LED circuitry was backwards - all LEDs needed to switch polarity from what was indicated on the PCB
- Silkscreen writing was too close to the holes on the enclosure front panel
- LED resistance values were tested with various values. The brightness of each LED was observed for different resistances, and the final values were selected as they were used elsewhere in the circuit:
- R11 (Red), R12 (Red), R13 (Blue): 1.5 kΩ
- R14 (Green): 10 kΩ
Using a Copper Mountain TRVNA, the following RF characteristics were measured for the v4.0 circuit:
- Whole circuit gain: +30 dB
- Whole circuit isolation (switch turned off): -105 dB (measured with +3 dBm input from TRVNA)
- Individual component gain/isolation:
- Switch 1: -90 dB
- Amp 1: +24 dB
- Attenuator: -8 dB
- Amp 2: +15 dB
- Switch 2: -80 dB
Using an oscilloscope with a variable RF input signal to the v4.0 circuit, the sinusoid signal within the circuit was measured for different input powers. This test determines under what input conditions the circuit behaves properly. At each point in the RF path, the RF signal was observed for variable input power levels (ranging from -35 dBm to -14 dBm) to see at what input power level distortion occurs. The results are below (measured up to a max input power of -14 dBm):
| After RF Component | Signal Distortion? |
|---|---|
| Input Limiter | No |
| Low-pass Filter | No |
| Switch 1 | No |
| Amplifier 1 | No |
| Attenuator (8 dB) | No |
| Amplifier 2 | Yes - at 120 mVpp (-14.4 dBm) input |
| Switch 2 | Yes - at 55 mVpp (-21.2 dBm) input |
| Output Limiter | Yes - at 45 mVpp (-23.0 dBm) input |
For the overall circuit, the highest input power where there is clearly no distortion is -24 dBm (40 mVpp).
Additionally, the power level at each point in the circuit was measured to be a maximum of 14.5 dBm (under a -16 dBm input signal), well within the operating conditions of the RF components.
Note: All Vpp to dBm power conversions are done for a 50Ω system
The switching time was measured using an oscilloscope with the circuit connected and controlled by the lab Borealis setup running Normalscan. The switching times were measured to be:
- TX to RX: 30 ns
- RX to TX: 30 ns
Additionally, the video leakage (switching noise) was measured for this circuit. This was measured for two cases: an RF signal connected to the input, and no signal connected to the input (only showing the video leakage). The results are as follows:
- TX to RX: 408 mVpp (-3.8 dBm) spike that disappears after 300 ns. The spike is noticeable for low input power levels (10 mVpp or -36 dBm), but less noticeable for higher levels (40 mVpp or -24 dBm)
- RX to TX: 92 mVpp (-16.7 dBm) spike that disappears after 50 ns. The spike is not noticeable for any input power levels.
Figure 9: Measured v4.0 circuit video leakage
The video leakage spikes when the circuit has no input RF signal are shown above in Figure 9. The left plot shows the RX to TX transition, while the right plot shows the TX to RX transition.
On August 14, 2024, the v4.0 boards were brought out to the Saskatoon radar site for testing. They were installed on the following receive paths:
- Main #00
- Interferometer #01
- Interferometer #02
Antennas IQ plots collected while the radar ran its regular operating schedule showed identical performance as the existing receive path circuitry. The TX pulse suppression looked slightly improved, and the received signal power looked identical.
Additionally, the radar was run in engineering-debug mode to collect some raw RF data to see transmitted/received pulses. Plots comparing the new versus old system are shown below in Figure 10, comparing receive paths for antenna #0 (new) and #1 (old):
Figure 10: Raw RF Plots: New RX path (left) versus old RX path (right)
The new circuit improved the TX pulse suppression by ~5 dB, and additionally suppressed some switching noise at the RX to TX switching boundary.
Following the on-site test, two v4.0 boards were left connected to interferometer channels I1 and I2. These receive paths both worked initially, but on August 23rd (nine days after installation), they both stopped working. This outage lined up with a lightning storm in Saskatoon. Both boards were discovered to have broken RF components (both RF switches and amplifiers) - the input limiter did not protect these components successfully. After some more testing, the circuits were left at Saskatoon again, this time with external VLM-52-S+ SMA-packages limiters attached to the inputs. After this change, the circuits worked and didn't break again.
Figure 11: RX Path v4.1 Board Renders
Figure 12: RX Path v4.1 Enclosure Front Panel
Circuit Schematic: receive_path_schematic_v4.1.pdf
KiCad Project: v4.1
The following small changes were made for v4.1:
- Fixed LED orientation
- Made two separate enclosure front plates, one for each TR input option
- Adjusted silkscreen so no writing is cut off
- Slightly modified PCB dimensions to better fit in the enclosure
For v4.1, enough boards (21x) were purchased to fully install the system at the Saskatoon radar site. The PCBs were purchased from OshPark and assembled in the lab by hand using the v4.0 stencil and reflow oven. Assembling and testing all 21 boards by hand took about two weeks. Some notes from assembly:
- All components except the LPF, +/- 5V LEDs, and input voltage jack were baked in the reflow oven. Many of the through-hole SMA connectors didn't solder correctly within the reflow oven and had to be touched up by hand.
- Reflow soldering the small RF components with tiny pins was troublesome. Less than half the boards worked properly after the initial soldering attempt - many of them had incorrectly soldered components. The troublesome components were determined via testing with the VNA and were re-soldered by heating the board with a solder gun and applying more solder if required. This method wasn't foolproof, as some components had to be re-soldered multiple times.
- Custom cables were made for the differential TR input and power input
- For differential TR: Molex Nano connector
- For power: Special locking power plug
Figure 13: v4.1 assembly
To verify that the boards were assembled correctly, the following tests were done:
- Power on the board and verify the power supplies work (LEDs light up)
- Connect the TRVNA to the input and output of the board and verify that the through connections are valid. The board should have +30 dB amplification and -80 dB isolation. Manually switch the board from TX and RX mode using the TR jumper.
- Connect a TR input to the circuit and confirm the board can switch between TX and RX mode from the external TR input.
Once all boards were properly assembled, S21 and S11 measurements were recorded for each board using the Copper Mountain TRVNA. To verify that all boards have identical gain and phase, the plots shown below in Figure 14 were made. These plots confirm that all 20 RX path boards have the same gain and won't introduce unintentional phase shifts to the received data.
Figure 14: v4.1 VNA measurement comparison
The assembled v4.1 boards were installed at the Saskatoon radar site on October 2, 2025.
- Main array RX path boards were installed within the transmitter boxes, replacing the previously used Mini-Circuits pre-amps. The 15V and TR inputs were connected via the screw terminals within the transmitter. The RX path enclosure was fixed within the transmitter using double-sided hook-and-loop tape.
- Interferometer RX path boards were installed on the interferometer pre-amp plate, same place where the previously used pre-amps were installed. A Mini-Circuits SMA-packaged limiter was installed on the RF input of the circuit to compensate for the input limiter and protect against overvoltage due to lightning. These were not necessary for the main array paths as the high- and low-power TR switches protect the circuitry.
The newly installed RX path boards are shown below:
Figure 15: v4.1 installed at the Saskatoon site
Figure 16: RX Path v4.2 Board Renders
Circuit Schematic: receive_path_schematic_v4.2.pdf
KiCad Project: v4.0
The only change in this revision was the input limiter; it was changed to the same device as the output limiter (SMP1330-005LF). This was decided after completing the following limiter testing and analysis.
Test setup: Use the signal generator (AFG) of an oscilloscope to generate a maximum RF signal (12 dBm for our oscilloscope) into the limiter and compare the input/output waveforms
- Port 1 of the oscilloscope: Output of the limiter
- Port 2 of the oscilloscope: Tee connected to AFG and the input of the limiter
- AFG of oscilloscope: Generate 2.5 Vpp (12 dBm) 12 MHz signal
The limiters tested:
- VLM-52-S+: SMA coax in-line limiter
- VLM-33W-2W-S+: SMA coax in-line limiter
- RLM-521-2WL+: v3.0 input SMD limiter
- RLM-33-2W+: v3.0 output SMD limiter
- CLA4610-085LF: v4.0 input SMD limiter, pin diode
- SMP1330-005LF: v4.0 output SMD limiter, pin diode
The results of the tests are shown below in Figure 17. Two measurements were taken for each limiter: the maximum input power where no distortion is visible in the output waveform, and the output power under an input of 12 dBm. Specifications from each limiter's respective datasheet are also listed in the table.
Figure 17: RX Path v4.0 Enclosure Front Panel
Things noticed from these tests:
- The v4.0 input limiter has a much higher spec'd maximum limiting output (+34 dBm compared to <15 dBm). This may be why the circuits failed during the lightning storm - the limiter "limited" the heightened power, but not to a low enough level to protect the circuitry
- RLM-521 limits the RF signal approximately 10 dBm earlier than the rest of the limiters
- All other limiters let the signal pass through until ~7 dBm, and immediately start attenuating/limiting the signal at this point. The maximum limiting output for these limiters is all <15 dBm, which is within the operating specs for the RF components of this circuit.
This analysis concludes that the input limiter chosen for v4.0 (CLA4610) doesn't limit enough, and allows too large a signal through in special circumstances. From the analysis, the output limiter (SMP1330) seems like it would work as the input limiter also.
The v4.2 board was ordered from Elecrow, instead of OshPark. The board was also ordered to be assembled by Elecrow, with all surface-mount components sourced and soldered onto the board by Elecrow. The price of this was much lower than purchasing the boards and components separately and assembling them ourselves in the lab. It took 2-3 weeks for Elecrow to assemble and ship the boards.
Once we received the boards, the following assembly was done:
- The boards were slightly too wide for the enclosures, and had to be sanded down to fit
- All through-hole components were ordered separately and manually hand-soldered to the boards in the lab. The SMA connectors had to be properly lined up to fit within the enclosure front plate.
For this batch, 50 boards were purchased and assembled. All of these boards were tested and measured using a TRVNA, with S21 and S11 measurements recorded. Beyond verifying that each board works, the goal of these tests was to measure the phase shift across all boards. Using this phase data, a subset of the 50 boards can then be selected to be used at a radar site to limit the overall phase difference. This was done to get two sets of RX path boards for two different radar sites. The magnitude and phase differences for these two sets of RX path boards are shown below in Figure 18.
Figure 17: RX path v4.2 phase/magnitude variation
By taking a subset of the assembled boards, the 20-board sets for each site have minimal phase/amplitude variation across the 20 boards (approximately less than 4 degrees phase variation for both sets). This ensures that minimal phase change is introduced within the RX path circuitry.
These two sets of v4.2 RX path boards were installed at the Rankin Inlet and Clyde River radar sites in July 2025.
