After the failure of the Paraset Pixie, I decided to give up for a bit on any Pixie-related projects. I was frustrated, and rightfully so. I more than likely got a dud but for as cheap as they run, there's not much of a loss there.
However, one does have to appreciate the simplicity of the Pixie circuit. You can spend hours and hours on research, but it all boils down to a simple direct conversion receiver paired with a simple CW transmitter. The work is placed on the final amplifier transistor, as this functions as the power amplifier in transmit, and acts as the mixer on receive. Clever, isn't it? You can see the idea here from the RU-QRP group:
Unfortunately, the Pixie is an abhorrent choice of QRP radio to use. It's crystal controlled, meaning that it lacks frequency agility. The choice of direct conversion on receive leaves the pass band wide open as a barn door, meaning that QSOs will be several kilohertz above and below your spot frequency. The variable potentiometer only helps up to a point. However, this doesn't mean that it is automatically destined for the trash bin.
The Plan
I was inspired on a QRZ forum that I asked a question on about changing the Pixie's frequency. Originally, I was going to make a simple VXO centered around a 28.224MHz crystal, but I soon realized that this was a beacon-only frequency - no one would be looking for CW QSOs here or around it! One user suggested I make a DDS VFO from an AD9850 or an Si5351. So I think that's what I'll stick with - both have advantages and disadvantages.
I also am planning on fitting my project with an OLED display and a rotary encoder to display frequency. This won't be a traditional Pixie obviously, as I will be adding a DDS, a more powerful amplifier transistor, a roughly 800Hz audio bandpass filter, and so much more! On top of that, this will be a 10 15 meter radio - I can take it truly anywhere, dipole and everything!
Problems, as always
I ran a sample sketch on the 9850 module I picked up a few years ago, and unfortunately, I must've gotten a dud. This is a huge problem as this module cost me over $20 back during COVID, so I lost out in comparison to today's money. Luckily, I may have a solution.
When I was searching for another solution, I was referenced to the Silicon Labs' Si5351 module. This chip works from as low as 8KHz up to 160MHz - what a spread! I'll only be using 28 to 28.3 21-21.1MHz give or take, but you can use it on all of the ham bands, even 60 meters and as far up as 2 meters. These things are so cool; if that's not cool enough, the Si5351 is controlled via an I2C serial bus, meaning that only two wires are required for communications to and from the microcontroller!
Before I get too in-depth, let's take a look at how the Pixie works. Above is the block diagram of a Pixie transceiver. Refer to it as necessary through these next few sections.
Transmit
The Pixie transmitter is turned on when the power amplifier transistor, an NPN-series, is grounded by the key. When this happens, the oscillator sends the desired frequency through an NPN buffer stage and into the power amplifier, which is then filtered by a band-specific LPF. The transmission is a continuous wave, or CW, and the receiver is turned off.
Pixies output at least 250mW, at most a solid watt. It all depends on the power transistor used. For example, you may get more power output by using a metal-cased 2N2222 than with a plastic-cased 2N2222 or the stock S8050, like what you get in the Chinese kits. More power can be achieved with an outboard amp, but you're more than likely better off with using the stock configuration with more voltage if the transistors can handle it. After all, where's the joy of oscillation if you're not QRP? Better yet, QRPp?
Receive
When the key is open on receive, the oscillator remains free-running, which is why you'll hear a tone if you place the Pixie within the vicinity of an SSB receiver. This provides the local oscillator for our direct conversion receiver. This mixes and beats with the incoming frequency to produce both CW and SSB, depending upon where you are frequency-wise.
The receive audio is amplified by an additional amplification stage, usually an LM386 or basic transistorized audio amplifier. Crystal earpieces can also work without amplification. Due to our receiver being direct conversion, it's receive pass band is quite wide. You'll hear dozens of CW QSOs happening at the same time, which for the unseasoned operator with a poor ear and impatience, can spell out frustration and wasted time.
The Si5351
The Si5351 is a synthesized clock source with three individual and configurable outputs. It has quite a frequency range, about 8KHz to 160MHz. Interfacing it with your favorite microcontroller such as the Raspberry Pi Pico or the Arduino Nano is simple and easy, thanks to its I2C capabilities. Created by Silicon Labs, it has been around since 2010. It's a 10-year-old design that is still popular and widely used. It is older by 5 years than the AD9850 but beats it with a frequency range four times higher. It is also significantly less expensive in bulk than the AD9850 boards.
That's not to say that the Si5351 has its drawbacks, and the most prominent one is its waveform. The Si5351 puts out square wave clock pulses, which are not good for RF use due to a lack of harmonic filtering. Square waves are very rich in harmonics and will cause chaos on the ham bands if filtering is not implemented. This is easily fixed by creating a square to sine converter using resistors and capacitors suited for the frequency range in question. In essence, all a square to sine converter is is an RC (or LC!) low pass or band pass filter. You may find you don't even need this, as some radios have internal circuitry and buffering that cleans up the waveform. On top of that, a crystal oscillator will generally put out a clock signal, which is also close to a square wave.
The Si5351, despite this drawback, is an excellent choice to substitute crystal oscillator circuits and VXOs in crystal-bound circuits. I know this takes away from the simplicity of QRP circuits, but remember, we're expanding on the Pixie. You can get the Si5351 here on Amazon. A three pack only costs $13, whereas a three pack of AD9850s would cost well over $50!
The Circuit
Here is a rough circuit diagram/block diagram I have sketched up based on what I think the circuit will look like:
The receiver is the bare front end, with no audio filtering. Right now, it will let in almost 3 or 4KHz of bandwidth through. We need to create a passive or active audio filter down to around 800Hz or so, whatever you feel comfortable working CW with. Some operators prefer wide band, some prefer a narrow band.
Right now, the LPF in the diagram above is rated for 40 meters. This obviously isn't static, as many Pixies have been rebanded to work as low as 160m and as high as 6m. Below is a table of common LPF values I found at KC9ON.com - another helpful Pixie resource! Values don't have to be exact (330pF instead of 315pF) and you can also wrap your own inductors too, or stack multiples in parallel.
One other thing - experiment with Q2. The Chinese Pixies supply these kits with the S8050, and while this works, most people can only get anywhere from 250 up to 650mW. I'd recommend trying out power transistors such as the BD139 or even a 2N2219. Metal-cased NPNs will dissipate heat better and as a result, push out more power. Whatever you use, make sure it's heatsinked properly.
For my LPF (10m or 28MHz), I'm going to use 2 100pF capacitors with a 270nH inductor. For my LPF (15m or 21MHz), I'm going to use 2 390pF capacitors and a 250nH inductor. The cutoff frequency is around 21.45MHz even, so well within band limits without any spurious emissions. I'll more than likely pipe this through a BNC connector up to a dipole antenna, as this should provide ease and efficiency of radiating my signals. An EFHW would also be a great choice.
10-5-2024 EDIT: How about a BPF instead???
Most if not all Pixie kits are subject to broadcast interference. This means that rigs on 80 and 40 meters may have an issue trying to tune out nearby AM or even shortwave stations. A low pass filter will provide attenuation - only on the frequencies higher than the cutoff frequency. A band pass allows a set amount of signals to pass while attenuating signals above and below the set band pass. This means that on 80 meters, local AM radio stations will be highly attenuated or will completely disappear! Kanga kits has posted this PDF file describing a simple BPF for existing Pixie kits. Here's the added bonus - it will attenuate your desired low pass frequencies almost 10dBm more than the stock Pixie LPF kit. So, if you're suffering from AM station QRM, try using this to make your Pixie more bearable to use!
Working Crossmodes
If you can't tell, I'm an avid follower of Peter Parker, VK3YE. He has very good and resourceful information on any ham radio circuitry you can think of, including the Pixie. In one of his Pixie videos, he worked another ham on crossmode, that is, he worked an SSB operator with a CW transceiver with great success!
I've always wanted to know if working crossmodes is as practical as it seems. After all, us ham radio operators are scientists - we perform several on-air experiments with old and new technology! I think working crossmodes sounds like a fun idea to try, and who knows? It just might catch on in the next few years!
Of course, there are many modifications left to be covered. One such includes the previously-mentioned audio filter. Paired with the LM386, this becomes an active filter. For simplicity's sake, using a simple 2-element RC low pass filter should help drastically reduce the amount of audio coming through the speaker. Aim for a cutoff frequency of around 700 to 800Hz. Some good values include 10kOhm and 22nF for a cutoff frequency of ~723Hz.
Stay tuned for an update as I embark on this adventure!
UPDATE (11/25): Keep the Buffer Circuit
A few months ago, I posted a question to the Pixie QRP Facebook page about adding on an Si5351 to the Pixie. Originally, my plan was to bypass the oscillator buffer (remove the 9018 and the entirety of the Colpitts circuit) and just inject the Si5351's signal directly into the amplifier/mixer transistor. However, the more I researched adding on a DDS to a Pixie, the more I realized an oscillator buffer may not be a bad idea.
An oscillator buffer outputs the same voltage that is fed into it, but that's not all. A buffer can help condition and clean up our dirty 5351 signal, match the impedance needed to feed the oscillator output into the amplifier/mixer, and prevent feedback.
In all practicality, an easy buffer can be comprised of an NPN and some resistors, which can be found on board the Pixie already. We will need to remove a few components in order to do this. Compare the two circuits below:
As you can see, the circuit above is just an enlarged version of the red area in the modified Pixie schematic.
This modification can be found on KC9ON's webpage and KM4NMP's webpage. It requires the addition of one 47K resistor between Y1 and D2.
Above is the jist of what you need to do - don't install C3, C7, Y1, W1, D2, R6, and C8. Place a 47K resistor between the ground side of the D2 footprint and the input side of the Y1 footprint. Across the resistor, solder a short piece of coax (RG-174 for example), with the center to the input side of the Y1 footprint and the shield on the ground side of the D2 footprint.
In addition to this some other questions were answered, leading me to abandon constructing this circuit on 10 meters due to low power output reported in several builds others did. Instead, I will construct it on 15 meters. 15 meters has always been fun for me ever since I started operating on HF and it is a daytime band, so I can use it more when I'm awake. In addition, 15 meters has CW access for Techs, Generals, and Extras - everyone can use it!
For a 15 meter low pass filter, you will need to use 2 390pF capacitors and a 270nH inductor. Just using these three components will suppress the second harmonic down to -23dBm - a much better improvement than the Pixie's stock filter! And you shouldn't need a capacitor in parallel with the inductor as the suppression should be more than sufficient for QRP.
If you can't get your hands on a 270nH inductor fast enough like me, you can always place 4 1uH inductors in parallel to get close to this value. In fact, 250nH appears to have a better effect on 15 meters. Placing inductors in parallel work the same principally as resistors do, and in fact you'll have an inductor rated at 1 watt solid if you use 1/4W ones - not bad for QRP!
(Possible) Additional Functions for the Nano or Pi Pico
In addition to just providing the brains for the Si5351, I would like to implement automatic zero-beating with minimal delay in keying this radio. Will this idea work? No clue.
Basically, when one of the digital pins is driven LOW (shorting a Digital pin to ground via the CW key), this will trigger the Si5351 to shift about 600 - 800Hz to match the other operator's signal without having to keep tuning around and changing the frequency by hand. The microcontroller will also output keying via a simple NPN transistor circuit to provide isolation. Of course, an optocoupler could be used. One such circuit could be this one:
You could also implement a simple CW keyer if you plan on using paddles too. Ernest, PA3HCM, has a phenomenal yet simple CW keyer constructed from an Arduino. See my previous several blog posts for the information.
One other modification I'd recommend is some form of sidetone using a PWM pin on the microcontroller going to the headphone socket. This will greatly reduce parts count and is not difficult at all to implement. Just generate the sidetone as the transmitter is keyed.
Of course, if that's not enough, you could also add memory functions, a configuration menu, or even WSPR functionality - anything really that you'd find on a commercial rig! Okay, I might be getting carried away with that one but if it pushes out a watt or so, I think it deserves these modifications.
No comments:
Post a Comment