I previously attempted to replicate the Elecraft K1, but found that the inherent oscillation was directly producing a mixed signal using an LC mixer, and its frequency drift was faster than the VR2. I then considered whether it would be possible to convert the inherent oscillation into a PLL or DDC.
After researching, I found that the Si5351 seems suitable from all aspects, but the biggest drawback is that it outputs a square wave, while I want the oscillator to be able to vary over a wide frequency range. This makes designing the filter very difficult.
Additionally, I considered using the Elecraft K1 with its fully surface-mount components. However, the Si5351 is both SOP (Surface Mount Device) and has a very dense pin arrangement, which makes soldering challenging.
I also looked into the option of using an MC1648 oscillator + MC145151 (not 114514) for phase locking, which would be fully surface-mount and could theoretically produce a sine wave. However, I discovered that the MC145151 has only a 14-bit frequency resolution, meaning that if I wanted to output a 30MHz signal, the frequency step size would be an astonishing 1831Hz. This is unacceptable for shortwave communication.
Finally, I saw the Elecraft K2's PLL chip, MC145170, which was available on Taobao for a relatively low price. However, even the surface-mount version cost around 40-50 RMB, which was too expensive for my budget, so I decided to abandon this option.
I was really frustrated, so I made some compromises. Instead of directly generating a sine wave, I could use a patch or a chip to do it. At the very least, I could get a pure sine wave. My next investigation revealed that the AD9851 DDS chip is actually quite good; it outputs a basically pure sine wave (the system's harmonics are high, but they can be easily filtered), and it supports both parallel and serial digital input, which is acceptable. It looks promising, so I started drawing the circuit board. When I looked at the datasheet, there were even example circuits! A good circuit, but I want to make some modifications (ding, ding, ding, ding, ding).
After reviewing the datasheet, I identified that the "LC" block was likely the output filter. The resistors R6 and R7 probably served to match impedance. Also, R4 and R5 were for voltage division, providing a square wave to the comparator below. However, this wasn't necessary, so I removed them. As for R8, it seemed to be for voltage division to IOUTB. Since we only used IOUT, I also removed R8. This resulted in:
After a few days, the components and board arrived. I started soldering. But after soldering one plug-in, it started smoking! I immediately cut off the power. After cutting off the power, I quickly inspected the board with a magnifying glass... It seemed fine, so what was going on? I checked and rechecked, and finally my attention focused on the flux in the lab:
I cleaned it thoroughly with alcohol. Now, when I plug it in, it doesn't smoke or produce any waveforms.
After four to five hours of debugging the timing diagram without finding any issues, I reviewed the datasheet and discovered that this device is supposed to operate in parallel mode by default. To switch to serial mode, you need to first input a byte using the parallel mode... Damn it, I had left all these parallel ports unused to save on IO. There was no other choice but to use jumper wires... It took me an hour to manually solder jumper wires with a 0.5mm pitch. After soldering, I powered it up and still didn't get any signal.
I had to check each pin of the simulation circuit one by one, and I found that... how could there be a crystal oscillator without a waveform? No, it's because the amplitude is too small. Did I damage it when soldering with a hot air gun? I tested a new crystal oscillator, and the waveform was still small... In addition, I also discovered that the crystal oscillators I bought now are 3.3V powered, while I plan to use 5V later, so I gave up on this one and bought a new one.
Another two or three days passed, and the components arrived. I soldered them in, but as expected, there was no waveform. There wasn't much else to do but consult the datasheet. Suddenly, I noticed that the datasheet specified an output impedance of 120k ohms for IOUT. That's a lot! Apparently, it's a current-type output... If I remove R8 from IOUTB, then I have to remove it. Continuing with the wiring, this time there was some output, but it was on IOUTB and the distortion was severe; IOUT had a waveform, but the amplitude was very small, only a few mV, drowned out by noise.
I was completely stuck, so I decided to buy a pre-made module and test it. After three days, I received the module, and as soon as I connected a microcontroller, a waveform appeared... This confirmed that the previous bugs were still present. I carefully modified the circuit and then sent it for PCB fabrication.
The debugging process went relatively smoothly this time. Aside from one inductor that was faulty, there were no other issues. Replacing the inductor allowed for normal output. I saw the most normal sine wave output from my hand-built circuits since then.
Now, we have a reliable frequency source for K1, and I hope there will be fewer hardware bugs in the future hhhh.
闽ICP备2021006864号-7
闽公网安备35020602002794号