Since my last article, I’ve been trying to decide on CAD tools. I originally suggested EAGLE but it has been purchased by AutoDesk and they require a yearly subscription for 4-layer PCBs. I intend to create only 4-layer PCBs to make layout easier per the suggestion in Michael Ossman’s video. I also believe electrical engineering needs to move towards open-source tools like the software community. Consequently, I started learning KiCAD, which neckbeards argue over how to pronounce. It’s a free hotkey-based CAD tool with significant recent contributions from CERN that runs on Linux, MAC and Windows. There are many tutorials (see here and here). I started with a YouTube tutorial:
Using the tutorials, I added one part to my schematic that will be the basis of my transmitter and hopefully double as the LO for my receiver: the AD9850 DDS. I got this idea from several different DIY hams on the web, like Stretchyman on HF Underground, VK5TM, and others. A DDS is a divider for a DAC clock which changes the DAC frequency output according to a frequency control word. The number of bits in the frequency control word and the DAC clock rate gives the frequency tuning resolution. The DDS will give any frequency from DC to the DAC clock rate divided by 2. For example, when this DDS is running at 125 MHz, I can output any frequency from DC to slightly-below 60 MHz (Nyquist). Different output frequencies have different spurious and harmonic performance as do different clock rates. These are trade-offs you make when you use a DDS as an LO.
On the transmit side, I intend to use the DDS as both an LO and a modulator. This DDS has a 5-bit phase control register which will allow me to tune to different frequencies and send PSK sequences. PSK31 is a form of QPSK. QPSK requires only 2 bits, so this DDS has plenty of phase resolution for PSK31. Next comes the question of how to make this into a transmitter. The output power of the DDS is +16 dBm or about 40 mW. I’m trying to achieve 5-10 Watts output power so I need a power amplifier that increases the power by a factor of at least 100 to 250 or 20 to 25 dB.
Power amplifier topology
This brings us to the choice of power amplifier topology. Should I use a linear power amplifier (A, B or AB) or a switched-mode power amplifier (C, D, E, F)? Linear amplifiers are far less power efficient but put out fewer nonlinear products like spurs and harmonics. Switched-mode amplifiers are far less linear but far more power-efficient. Since I’m using PSK which is constant-amplitude, I can use a high-efficiency switched-mode power amplifier which gives Watts for pennies. From what I’ve read, hams are using these amplifiers to achieve 100-1000 Watts for about $10 per MOSFET to transmit AM phone. Switch mode amplifiers are explained in The Handiman’s Guide to MOSFET “Switched Mode” Amplifiers by Paul Harden, NA5N. You can find some of these articles on Scribd. These amplifiers require a scope to tune, so while they’re cheap to fabricate, they’re possibly more-expensive to test. These amplifiers seem great if you want to rag-chew with someone on another continent using a cheap DIY transmitter. We are trying to send small amounts of data locally or nationally with as little power as possible relying on the processing gain of our modulation scheme (PSK31) to improve SNR. Consequently, I should also consider linear amplifiers (A, B, and AB).
I have found component selection to be difficult. I must first select the MOSFETs for the final power stage and then drivers, if necessary. These MOSFETs are typically specified at VHF and UHF so test circuits are provided in the data sheet only for those bands. We are working at HF, which might require us to design the matching networks using old-fashioned methods like a Smith Chart. The DDS gives +16 dBm, which is pretty beefy so I may not need a driver. Remember our requirements: 5-10 Watts output power, 8-12 volts bias voltage (VDS). Our choices of device technologies are GaN, which tends to work better at higher frequencies, and SiC which tends to work better below 7 MHz. GaN is driven with higher voltage for higher power and might be overkill for our application, though we want to cover higher frequencies. OgreVorbis designed a class D AM transmitter that outputs 500 Watts using SiC with a goal of covering 40 M (7 MHz). As he reduced VDS to 13.8 V at 3 MHz for 40 Watts, the efficiency fell to 90%, which is still much better than a class A or B amplifier. The power was 10 times higher than we want, though. His board didn’t work above 5 MHz even after a board revision. He is trying to diagnose the problems.
The next installment will discuss part selection and the schematic.