So a few weeks ago I had sent and email to Pete Juliano asking some questions and showing off the progress of my projects, he commented that my work bench looked way to clean for me to have been productive, well I sent him this picture from this morning when I was hard at working building this project. There is currently CRAP EVERYWHERE.
Anyway, now onto the real story. The first thing I have built and designed that I am actually proud of. It all works as designed and I think i am finally starting to get up the learning curve.
So there is nothing revolutionary here. Its a simple CW transmitter biased Class A all the way. Why class A when there are more efficient biasing schemes for CW? Well, because I want reusable circuit blocks that I can use for DSB, SSB and other modes that require linearity. The theory of operation is simple. We have an ATMEGA328 that controls an AD8950 DDS module, Q3 is a buffer so the gain stages do not load down the DDS. Q1 is a simple switching circuit. When the key is keyed, it pulls the base to ground, 12V can flow from collector to emitter of the PNP transistor, turning on Q4 and Q5.
I had to switch both of these and you will notice on the build board below a mod wire, as originally I had only one of these stages switching, but i was getting massive leak through. In a future build I will look at a better switching method to improve the overall usefulness of the transmitter and think about adding in some wave shaping and all the fancy stuff.
Q4,5 and 6 amplify the 300mv DDS signal to 10V p-p where it is then amplified by a single FET to around 5 watts. Its simple, but works, and getting those gain stages all working together was not an easy feat, I spend a lot of time in LTspice making them place nice with each other.
Here is a close up of the board. You can see the mod wire on the top side, on the bottom side of the board I cut a trace to isolate it.
This is most of the test setup, using only 1 side of the paddles because my straight key has the wrong size jack on it.
Current draw is around 0.7 amps key down.
Power out is 5w.
\
Output on the oscilloscope.
And this is why you should never trust the output on the scope to give a honest representation of the harmonic content of the signal. First harmonic is -40db down and the second harmonics is only -30db down. Not good enough for the kinds of girls I go out with. So I need to spend sometime going over the build and see where added distortion is coming from that was not present in the prototype. When I put it on air later to test it out, I will add an extra 7th order filter to knock those harmonics on the head.
You know what. Its really nice when things start to click and what you build looks and works like the design. I get the feeling that 2020 is going to be a ground breaking year for my home brewing.
Cheers,
Rob.
Nice work, Rob!
Thanks. 2020 looks like the year when everything comes together and I build some things worth using.
Great job!
When you are simulating the ad9850 in ltspice, are you just doing a voltage source with 200ohm series output and a 1v pp amplitude?
And how much of the circuit do you simulate together at once?
I set the input voltage to be as close to what I am using as possible, in this case, 300mv p-p and 200ohms. Then simulate as much of the circuit as is practical. In this case, it was the buffer and first 3 gain stages as well as the pnp switch. So it was most of it.
Thanks. Always curious how other hams are using ltspice.
Yeah Ben, its always interesting to see what and how others are doing thing. LTspice is one of those must have skills i think. It really gets you in the ball park.