This is the PCB for the QSD2 receiver module for the T41 "Software Defined Transceiver". The PCB was designed using the open-source design tool Kicad 8.
The QSD2 evolved from the original T41 V010 design which remains the most recently published receiver module for the T41 Software Defined Transceiver.
The goals of the QSD2 are to improve receiver performance, and in particular, the performance in the 10 meter amateur radio band. Please note that the designer of this project does not own a laboratory full of expensive test equipment. Much of the verification of the receiver performance has been done "on the air". There are no guarantees or other warranties of the performance of this receiver module. What you see is what you get!
Gerber files for PCB fabrication are included in the gerbers folder. A PDF of the schematic (qsd2.pdf) is included for quick viewing of the circuit design.
Please note that the components used in this design increase the cost compared to the original design. However, the board is easier to build. A link to a public Digikey BOM is included so that the prospective builder can evaluate the cost before proceeding.
Version 1.1 is a minor revision of Version 1.0. Here is a list of the changes:
- 5.0 volt regulator is added.
- Removed unnecessary 3.3 volt and 5.0 volt supply decoupling components.
- Removed the option for 3.3 or 5.0 volt quadrature generator. 5.0 volt rated components are required.
- RF input layout is optimized for slightly improved maximum frequency response.
- The IQ output connector is rotated 90 degrees to match the QSE2 board.
- Low ESR bypass capacitors are recommended to improve noise rejection. These parts are included in the revised BOM.
The original QSD uses a divide-by-4 quadrature LO generator circuit. This is a robust circuit which produces high-quality quadrature local oscillator waveforms. However, the circuit has a trade-off and that is the requirement for VHF range frequencies in the higher amateur HF bands.
For example, reception of a SSB signal at 28.510 MHz requires a frequency of:
(28.510 MHz + 48 kHz) * 4 = 114.232 MHz
Note the 48 kHz term which is the IF frequency.
This frequency is applied to the quadrature generator which is composed of a dual flip-flop integrated circuit. The circuit is powered from the 3.3 volt supply.
The flip-flop device is a type 74AC74. The specification sheet is here:
https://www.ti.com/lit/ds/symlink/sn74ac74.pdf
The specification of interest, fclock with Vcc = 3.3 volts, is on page 4, and the maximum is 95 MHz. This is insufficient to support the 10 Meter band.
One page the same specification is shown for Vcc = 5.0 volts, and the maximum increases to 125 MHz. This shows a method to increase the frequency range, however, a matching 5.0 volt specified multiplexer must be used. A 5.0 volt multiplexer is in fact available at no cost penalty.
QSD2 resolves the high frequency problem by using a divide-by-2 quadrature circuit. The frequency is now:
(28.510 MHz + 48 kHz) * 2 = 57.116 MHz
This is well within the frequency specification of the 74AC74 flip-flop IC even with Vcc = 3.3 volts.
Once again you don't get something for nothing, and the divide-by-2 quadrature circuit requires a little more effort.
Divide-by-2 requires a clock signal and the inverted clock signal. This is easy enough to provide using an inverter integrated circuit:
https://www.ti.com/lit/ds/symlink/sn74lvc1g14.pdf
So providing the inverted clock signal is easy. However, this solution creates a new problem! The inverter creates a small delay in the inverted signal. This affects the quality of the quadrature. However, this is easy to compensate for. Simply add another inverter to the output to "match" the delays. So we have to add two these inexpensive devices in addition to the same flip-flop device used by divide-by-4.
The divide-by-2 circuit was simulated and found to produce acceptable quadrature.
For optimal quadrature circuit performance, the coaxial cable between the Main board and the QSD2 should be as short as possible. Note that the output of the Si5351 synthesizer IC is 3.3 volts. This is used to drive 5.0 volt logic. This is not ideal, but it has been shown to work in practice.
Another complication of the divide-by-two is that the circuit must be properly reset prior to application of the local oscillator signal. If the reset is not executed, the circuit may produce inverted quadrature or it may not function at all.
Thus a reset signal must be provided to the QSD2 board.
This turned out to be easy to implement. The QSD is designed to be powered from the 16 pin ribbon cable which connects all of the radio's modules together to a common power supply. There are unused pins, and one them is used to provide the reset signal. A single wire is tacked onto the back side of the Main board, between a GPO pin of the Teensy and the unused pin on the 16 pin IDC connector. That is all!
The software modification is equally simple. The reset is toggled prior to the application of the local oscillator signal at radio power-up. That is all that is required to guarantee that the divide-by-2 quadrature circuit is properly initialized. This is a common and practically universal requirement for digital circuitry.
There is no change to the calibration code. In fact, the divide factors can be mixed. For example, a divide-by-2 QSD2 can be used with a divide-by-4 QSE. As long as the correct divide factors are used, calibration will work perfectly.
The T41EEE (Extreme Experimenter's Edition) software includes the software modification for QSD2. To allow operation with QSD2, the division ratio which is found in the file MyConfigurationFile.h must be modified as follows:
// Set multiplication factors for your QSD and QSE boards. #define MASTER_CLK_MULT_RX 4 #define MASTER_CLK_MULT_TX 4
change to
// Set multiplication factors for your QSD and QSE boards. #define MASTER_CLK_MULT_RX 2 #define MASTER_CLK_MULT_TX 4
T41EEE software is available here:
https://github.com/Greg-R/T41EEE
The T41 is an "Experimenter's Platform", and there is no right or wrong way to solve a problem. It is the experiment; the journey of discovery and amateur radio fun and adventure is the destination!
Here are a few ways to experiment with T41 frequency range expansion:
- Change the bias voltage on the 74AC74 from 3.3 volts to 5.0 volts. Change the multiplexer IC to a part specified for 5.0 volt operation. This changes increases the frequency response of the flip-flop device.
- Change the 74AC74 to a device specified for higher frequencies while leaving the bias at 3.3 volts. The device is the 74LVC74. The specification sheet is here: https://assets.nexperia.com/documents/data-sheet/74LVC74A.pdf.
- Change the divide-by-4 circuit to the divide-by-2 circuit as described above. This circuit is slightly more complex, however, it has large margin to the upper frequency limits of the components at 10 meters. 6 meter operation may be possible. There is no known low-frequency limit to the divide-by-2 (or divide-by-4) circuit.
- The T41 V012 design generates quadrature using an internal phase-shift circuit, such that one clock output is in-phase
and another clock output is shifted by 90 degrees. This is directly applied to the multiplexing (demodulator) device,
and thus this circuit is very simple. This also exploits the maximum upper frequency range of the Si5351 phase-lock-loop IC.
On the other hand, this introduces a low frequency limit and it may not be possible to cover the LF and VLF amateur bands with this scheme.
The QSD2 receiver module circuitry is almost entirely revised compared to its ancestor the V010 QSD.
The discrete bipolar transistor RF amplifier is replaced with a Mini-Circuits integrated wideband amplifier:
https://www.minicircuits.com/pdfs/GALI-39+.pdf
Using this device is easy as it requires only a bias resistor and two coupling capacitors. There is a series of similar devices in the same package. This particular device may not have the optimal gain and intermodulation performance, however, it will be easy to substitute and experiment with other Mini-Circuits amplifiers in the same family.
The problem with many of the integrated RF amplifiers is that they have too much gain! Too much gain will reduce the intermodulation and general strong-signal handling of the receiver. Too much gain will have negligible impact on receiver sensitivity.
A Pi resistive attenuator is added between the output of the RF amplifier and the input of the demodulator. Thus the front-end gain can be adjusted to optimize both receiver sensitivity and intermodulation performance.
The original V010 demodulator circuit is a single-balanced design. The QSD2 is upgraded to double-balanced. This requires an input transformer with a tapped secondary:
https://www.minicircuits.com/pdfs/ADT4-6T+.pdf
The tapped secondary provides the means of applying a common-mode bias to the demodulator device. Using this ready-made transformer means no toroid winding! However, it is expensive: US$6.25.
Note that other MiniCircuits transformers in the same package may work such as the ADT1-1WT+. This particular part has the center-tap on the output side. The layout will accommodate this.
The output of the demodulator uses "differential mode" integration capacitors. The idea here is to balance the output circuit to the highest degree possible. The capacitors should be specified to 1% tolerance. Shunt capacitors are also used in the circuit. The capacitor values in this circuit are subject to future optimization.
"Instrumentation amplifiers" replace the op-amps from the V010 circuit. The instrumentation amplifiers are intended to improve the symmetry and balance of the demodulator circuit, thus improving performance, and in particular the image rejection.
Another benefit of the instrumentation amplifiers is the simplification of the circuit. A single resistor sets the gain. The gain setting resistors should be specified to 1% tolerance.
A disadvantage of the instrumentation amplifiers is higher cost. The AD8226 device cost is US$3.77. Quantity 2 are required.
The power supplies routed through the 16 wire ribbon cable are prone to noise, most likely from the Main board. Extra decoupling components were added to clean up the power supplies. A lower noise floor has been observed of the displayed spectrum versus a V10 QSD board.
The version 1.1 QSD2 adds a 5 volt regulator to the circuit. This was added to make sure the instrumentation amplifiers have a clean 5 volt supply. Another benefit is that power dissipated in the 5 volt regulator of the power supply board is reduced; not by a lot, but at least it is going in the right direction.
The I/Q output connector is changed to surface mount. The RF SMA connectors are the same, or they can be changed to right-angle connectors to improve coaxial cable routing. The 16 pin IDC connector is the same as the V010 board.
The PCB layout was completed using Kicad version 8. This is an open-source tool, unlike the "Dip Trace" commercial product used to design the V010 and other T41 boards.
The layout in the quadrature generator and demodulator areas was very carefully routed for maximum frequency performance. The QSD2 is approximately the same board size as V010, but there is enough area to add additional circuitry if desired.
The prototype PCBs were fabricated by PCBWay at a cost of US$1.00 each. Shipping was about US$25.00 for quantity 5 boards.
A public Digikey BOM is here:
https://www.digikey.com/en/mylists/list/K5OIGZF85K
Please note that specific parts may or may not be available when attempting to order. It is the responsibility of the builder to find subsitutes as required.
In general, QSD2 is easier to build than the original V010/V011 series boards. There are a few items to be aware of to avoid build errors.
The most probable error(s) are incorrect orientation of the flip-flop, multiplexer, transformer, and instrumentation amplifier devices. The pin 1 markings on the devices are very difficult to see. The transformer markings are easy to read; this should be clear in the photograph of the finished board.
Here are details on each part with regards to proper orientation of the board. The descriptions are viewing the top side of the board, with the "T41 EXPERIMENTERS PLATFORM" near the bottom of the board in normal left-to-right reading orientation.
The transformer TR1 should be placed with the "dot" at the upper right corner. The textual markings will be upside down.
Pin 1 should be in the lower right corner. This is marked with a small white dot on the PCB.
Pin 1 is at the upper left. The part I used has a bar on the Pin 1 end. So the part is placed with the bar up, towards the flip-flop device.
This part is probably better as a 10 Ohm resistor. A higher value is possible, which would further attenuate power supply noise. The voltage drop across the resistor has not been measured, and the value of R6 is subject to further optimization.
Only U5 has the Pin 1 indicated with a dot on the PCB. However, both parts are oriented the same, with Pin 1 towards the upper left. There is a dot on the parts to indicated Pin 1, however, it is very hard to see. The package is beveled on the Pin 1 side; this is easy to see.
C20 and C21 are "DNP" (Do Not Place). These parts were intended to be used as RF bypass capcitors if necessary. So far, this is not a problem.
There are 4 capacitors on the bottom side. These should be soldered last. I was able to use a hot air gun without any problems with the top-side components desoldering.
R16 (O ohm) is placed on the bottom side in the case of using only one of the 2.5 volt voltage regulators. This modification has not been evaluated yet, so R16 is labelled as "DNP" (Do Not Place).
Take a good look at the placement of the board in your T41 radio. You will want the SMA connectors of the QSD2 in the orientation for easy coaxial cable routing. I used 90 degree connectors as this was the easiest for extensive testing, but not necessarily the best for permanent installation. Also, you may choose to put the SMAs on one side or the other for optimal cable routing.
Please note that the cable from the Main board to J1, which is the receive local oscillator, should be as short as possible.
Links to photos of a fully constructed QSD2 follow.
https://drive.google.com/file/d/1XmEzX6fYmMjZtCuPrj_bD8YfRSP8r1Cy/view?usp=sharing
https://drive.google.com/file/d/1RXFwcmE2U3pgW8C_lg-2-wzK6qM8bwK8/view?usp=sharing
A wire needs to be added to the main board for resetting the dual flip-flop at power on. The wire is connected to pin 0 of the Teensy. Note that I also added a shunt bypass capacitor to ground. A zero-ohm resistor was required to reach ground. The other end of the wire connects to pin 1 of the 16-pin IDC connector. It might be better to tack the bypass connector to that end of the wire. In any case, be careful while soldering the wire because it is directly connected to the Teensy. Follow good ESD procedures to make sure the Teensy is not damaged. Here is a high resolution image of the Main board. The reset wire is blue.
https://drive.google.com/file/d/1hGLFVzCbjHw2UNAi1ffdGSL7jEQ02EJ3/view?usp=sharing
The QSD2 is an attempt to integrate the best circuits from several sources into a high-performance HF receiver module. Here is a list of sources which inspired the design of the QSD2:
- "Digital Signal Processing and Software Defined Radio, Theory and Construction of the T41-EP Software Defined Transceiver", by Albert F. Peter, AC8GY, and Dr. Jack Purdum, W8TEE. This is the ultimate resource for builders of the T41 Software Defined Transceiver: https://www.amazon.com/Digial-Signal-Processing-Software-Defined/dp/B0D1JH5L65
- "A Software Defined Radio for the Masses, Part 4" by Gerald Youngblood, AC5OG. This shows the circuit for the double-balanced demodulator circuit and the use of instrumentation amplifiers on the output circuit. https://www.arrl.org/files/file/Technology/tis/info/pdf/030304qex020.pdf
- The Rod Gatehouse 2022 transceiver website: https://ad5gh.wordpress.com/2022-sdr-transceiver/ Rod took the original T41 SDT design and evolved it into a high-performance digital mode radio. Excellent circuit upgrades!
- "The Lentz Receiver: Tayloe Evolved" by H. Scott Lentz, AG7FF. An interesting discussion of several design features of the "Lentz Receiver" which improve the performance of the typical "Tayloe Detector" style HF receiver. https://www.arrl.org/files/file/QEX_Next_Issue/2023/05%20may-jun%202023/05%202023%20TofC.pdf