A simple HF QRP Transceiver optimized for Digital Modes By WB2CBA Barb – Simple to procure – meaning not effected by chip shortage – Simple to build – 2 modules, 2 IC’s and 4 Mosfets! – Simple to setup and tune – One simple calibration procedure is all needed. – Simple to operate – Plug in ADX MIC to soundcard MIC input and ADX SPK to PC soundcard speaker input and we are good to go with any digital modes Software. – Dirt Cheap – Costs less than 25$ to get all parts and PCB if we exclude ridiculous shipping costs! Any digital tonal mode consists of varying audio tones that change frequency in relation to the data they correlate to. This audio tone generated for example for FT8 with WSJT/X software is passed through an audio band pass filter and then it is compared with Arduino Nano’s Atmega328P processor A/D comparator for start and stop zero cross detection to determine period of that tone. From that period, frequency of that particular tone that is determined and added to base transmit frequency, for example 14074000 Hz for 20m FT8. If the tone let’s say is 1000 Hz then the carrier TX is now 14074000 Hz + 1000 Hz = 14075000 Hz. As it is summed it will be in USB frequency range of any SSB receiver though the signal is not pure SSB signal still any SSB TRX set to USB can’t tell the difference! This tone frequency detection and adding to base frequency is continuously repeated 400 times per second and refreshed until the FT8 tone transmission generation is over. In that case TX stops. What fascinates me about this FSK signal generation technique is to generate a SSB – USB signal without any SSB mixing or filtering which eliminates problems of IMD or phase differences etc that comes with SSB signal generation technique which allows such a simple Transceiver to be conceived with minimal parts. For more info: https://antrak.org.tr/blog/adx-arduino-digital-transceiver/
The signal is not SSB. It is simply FSK, which can be generated directly. It is just that using audio FSK and an SSB transmitter to heterodyne it up to the desired frequencies is a convenient, although Rube Goldberg, way to do it. Direct FSK is also an option. The same could be said of CW, you could key an audio oscillator into an SSB transmitter, or you can key a carrier directly.
Awesome project - Very cool ! I read something on the pcbway forum about using band-switch relays. With Cost in mind, you might like the diode-switching alternative solution like you can find in the SSB-6-in-1 kit
It is of course interesting. It is either FSK or SSB, depending on how you wish to see it. And this is literally true. We may see the FSK connection right off. And it is correct, insofar as it goes. This was the old school way of making FSK. By 1968 or so, we started making filtered versions of FSK. We needed something more flexible than PLLs and so forth. By 1974, we started doing this in pure math (bit slice processors flaming away at 100 MHz) and ramming the results into DACs, headed for complex phase modulators ... these were the first true DSP TX's. Barb's machine follows in their giant footsteps. Yeah, we needed CPMs in 1974 (for VHF/UHF, at least), but that would vanish in time with fast FPGAs and ultra fast DACs, straight to RF. And most folks here know how smartphones do it. Stripped of magnitude information, a complex-valued signal generator, programmed in freq, or "pushed along" in phase, is just "half" of a phasor. See Ron Bracewell's great text, The Fourier Transform and Its Applications, an EE501 favorite back in the day. More senior class these days, but well worth the price. I dog-ear mine all the time. A phasor is complex-valued x(t) = f(t) - j Hi[f(t)], the analytic signal. The arg[c(t)] is its phase in real time t. Mag[f(t)] = (I[x(t)]^2 + Q[x(t)]^2)^(1/2) is its magnitude (back to that below). The arg produces the instantaneous freq of x. Now, just take the "projection" of x, or Re[x(t)] = s(t), something you can pump as real electrons, a.k.a. current, and you have SSB, any direction, up or down. No filters, no double balanced mixers (that's all in the math), let alone complex DBMs. This SSB can "carry," if you will, anything at all; ASSB, FM, FSK (or AFSK as some purists insist), PSK, QAM, all filtered variants, even good ole AM. It is the dream machine. And I say, SSB is all carrier. That old school carrier thing in AM was just a fixed freq component, (absolutely) necessary to keep the modulated information from going negative! Yeah, I know the (yes, great) ARRL Handbook says otherwise! Old, and properly venerated, school. Barb's machine is a simple instantiation of the dream machine, so far reported to be doing FT8 FSK. But it is, by its design, capable of much more, and that is the SSB interpretation of the machine. Maybe he's working on that right now! ;-) Several other QRP rig designers have pushed this basic architecture to voice SSB, using real-time, narrowband phase advance/retard to create the modulation. No filters, no balanced mixers, just a little clock gen chip, high speed programmable, the SiLabs 5351 ($8 in qty!) as in Barb's machine. This all-freq "phone" is a bit Donald Duck of course, but works as the brain handles the rest. [Take that, AI!] The Real part of the phasor is "realized" by translating Q (along with I as an arctan) into an instantaneous argument in radians, fed to the chip as a freq sample (for a few thousand RF cycles). You can go way beyond this with a DDS, handling both arg and mag in real-time. This is where the complete phasor comes back into play. We just take Re[ ] of that "full" phasor and presto, SSB in all its glory. I think everyone sees it (literally, since modern rigs do this, and many of us - incl. me - just stare in wonder). You can even do this straight off with a single DAC, but that route is substantial power at this point in history to do 40m SSB. I think by 2030 or so, we will see the direct SSB route at 40m as QRP. Or maybe even a higher band. I am a great admirer of these QRP rigs, even as I don't own one, programmed as they may be in C, but also maybe even in Python. S/W power. Hiding in my Yaesu, beyond easy reverse engineering in all likelihood. But who cares. We can do any of it ourselves, straight from first principles. Go Barb!
Jameco, $0.39 each https://www.jameco.com/z/BS170-Majo...ansistor-N-Channel-MOSFET-60-Volt_256031.html
Oh ok. Don't know why I didn't check there. Digikey and Mouser have them on order with about a year lead time. Thanks! Also, Antique Electronics Supply evidently has them, as an additional answer to my own question.
It is pure FSK. It is not using audio FSK to heterodyne. Audio is used to determine the tone fr It is pure FSK Yes you are absolutely right. The signal is not SSB. It is pure FSK. Audio tone is used to measure frequency of tone at that instance and that tone frequency is added to the base frequency and the output is pure FSK which changes in every tone change.
I downloaded the gerbers, but haven't had time to really dig into this. I have 3 other projects in que, so it will be a while. But I was wondering -- This started out like most projects and is targeted at HF, but the 5351 is capable of doing 6M and 2M. The 2M depending more on layout. Other modules can go even higher. So, is there a timing or precision issue that would limit the upper frequency?