Simple SSB Rig: Part 10a - RF Power Amplifier Changes 

Hi Charlie. In my 40 years being in electronics I never had anything to do with short wave RF design. However, last November I passed the exam for F-license and now I am totally hooked up to RF design videos. (Specially when the wife is watching "The Voice"). I am actually mostly interested in how it all works. For my work I have designed wearables at 869 and 918 MHz but I really like the 20m band for the hobby, it's much more fun than all the 0201 stuff on small boards with everything on chip. Your videos are amazing, especially the way you take a piece of paper and explain each and every resistor and transformer. I have built an SSB receiver with a Tayloe detector and used a Teensy with your Hilbert transform software explanation to make a phase shifter. It works amazingly well. Now I try to built a 5 watt 20m transmitter with two times BD139 in push pull setting. I modelled it in LTspice and it seems to work reasonably well. However, if I drive the input transformer (3:1) with a 50 ohm generator LTspice shows distortion on the generator output going into the transformer. The emitter resistor of the BD139's is 2 ohm, if I understand your calculations right that should reveal some 25-30 ohm impedance at the input. Making the generator very low impedance tackles that problem but I wonder what's going on. I used that low-pass filter you explained lately and it gives a perfect, low harmonic output on the 50 ohm output dummy side. If I put a 220pF in parallel with the 50 ohm input the distortion on the input is gone. Somehow I get some interference out of the input stage.... am I right? I calculated with FT37-43, and have more than enough reactance. Can you give me some hints about what I am possibly doing wrong? Many thanks, Harold PA0AHL The Netherlands

congrats on 5k subs! I've been going through all your videos and building my first rig. Got a receiver working and it sounds good! Working on a variation of this circuit now. For the output side I found a trifilar wound xfmr in a 1:2 configuration on the drain (right into a .1 cap) gives the most output. The mosfet sees 12.5 ohms at the drain with a 50 ohm load. I can get 5W out with 12v supply, admittedly with a 10v p-p signal from my function generator. I'm still playing with the drive circuit though. I'm finding a lot of the signal at the drain makes it back to the gate through the gate-drain internal capacitance. Since I want it for 40/20 band use I'm thinking to keep the 50 ohm resistor so it swamps (relatively) the very different reactance of the gate capacitor at 7 vs 14mhz. My next step is to try a common emitter amp to drive the mosfet. Anyway just wanted to share some ideas and say thanks for all this great info! 73 cheers!

He said nice and clear with no distortion! you can't ask for more. well done!

Very nice radio. if you do buy a case, get a bigger case than you need - this way you can open the case and not damage the radio. I damaged a MFJ Cub I built because i kept getting the radio out of the case to tweak it.

Thanks Charlie, I'm just starting to wrap my head around AGC for my rig. I'm thinking separate IF AGC and audio level control. I always look forward to your posts. WA8HTO

Hello Charlie - just to let you I'm copying (ish) your build and look forward to each video. I'm a rank beginner so the simpler the better for me. Keep doing what you're doing! One of the bigger changes I've made is using a RP2040 Pico (£3.90 plus postage) with Python rather than the Arduino - that makes the coding a lot easier for me (I got the Python hint from the YT-er NA5Y). Thanks Kevin M0XYM

Of course you did say you wanted to keep it simple, so yes you are correct with no AGC. I have a real problem when designing things of going crazy with the features! You have to draw the line somewhere, unfortunately for me I keep moving that line which I think is the main reason a lot of my projects are never totally finished.

And for the output. The optimum impedance the IRF510 needs to see at 24V supply is 35R for 10W output. So your 5:6 transformer is about right to match a 50R output, but you have it the wrong way round !

Agree, it should be much more than 5W. Low power output may indicate either an impedance mismatch (the energy is reflected back to the driver) or a low drive level, or both. Personally, I would model the circuit in LTspice to get a better idea of what's going on. The model is usually quite accurate for such kinds of circuits. As a side note, there is a good 10W PA circuit based on IRF510s, that can be found in uBITX schematic. Requires 13.8V and I had no problems reproducing it. Unless the idea is to experiment with new designs, of course.

I'm looking at building my own version of the uBitX. Will have dual conversion with 45mhz and 11.2mhz IF's (found surplus cell phone 45mhz filters on Ebay, and the 11.2mhz SSB filter came from an old CB radio). Will use the si5351 for all oscillators (might use the two 11.2 mhz crystals from the CB rig for BFO/Carrier oscillator). Will re-arrange things a bit, on RX will put IF amps AFTER the filters, and will use separate IF amps / product detector / bal modulators instead of true bi-lateral. I have some MCL combiner/splitters I might use.

Just had a thought of something new that may make the video more interesting when transmitting. I was playing with some online SDRs and noticed the other day some have a record function to them. Listening to both sides of the radio would be cool. :)

Every nice. I have some good power transistors i plan on using for my PA. Got two MRF453 from a CB amp i bought from a thrift store for $5 (mostly to make it work on a different band, plus its something illegal off the streets.)

A good conclusion to a great project. Thanks for making these videos, certainly has encouraged me to have a go. Which rig to box up? A difficult call, the first one would give you two bands in the field but the second would be smaller, lighter and more in keeping with the QRP and field ethos. At the end of the day it's only you who can make the call Charlie! Best 73s.

What is the voltage swing around the bias on the IRF510 gate? General wisdom is you need to drive an IRF510 pretty hard, and I doubt you're getting much out of the 2N3904 in a linear mode. If it's (generously) hitting 50% efficiency, it can't operate within specs and be linear with more than about a quarter watt of output (maybe a 0.5W with a mondo heatsink and driven out of spec). Getting about 5W means you are probably seeing about 13dB of power gain -- not too bad for a non-RF transistor. Asking for 10W or more is asking it to perform like something that it's really not. It'll only give you so much for so much swing at the input, and it has that massive 180pF input capacitance to deal with too. A stronger driver might work -- could even just make it one more IRF510 in between, so it's three stages... 2N3904 -> IRF510 -> IRF510?

Old Collins Guy Alc and agc would be nice, Diving into teensy as a dsp alternative, Their so Cheep. got a few for experimenting with, The demodulation and audio Compression etc. and running the whole thing. the 4.0 or 4.1 Teensy has a lot of possibilities got some others , had a DSP 10 2meter in past and lost it because of moving among other things. Seems to me the dsp would eliminate the If Filter and amps and add multi modes.

hello, I think you should add a decoupling capasitor on the ampplifier at the 24V input just befor the RFC. Otherwise your 24V wire participate to the inductance value of this RFC and may also be a big source of RF leakage. As a general rule place as much decoupling capacitance as you can on any volatge source at the point of connection. on4ldy

It's good to keep it simple! get rid of luxury that is not necessary indeed. what struck me is that 5 Watt indeed seems too little to me? when I make an AM Grenade transmitter I thought I was closer to 10Watt. otherwise search with Google for: grenade transmitter maybe it's the amplifier stage, (amplification from transistor?) I used a 2N3866 a little less maybe 2N2219A could be work too! And who knows, you just need an extra staircase behind that 2N2904 ?! And who knows, 2X 2N2904 may still work??? or also pay attention to the capacitors that they do not heat up at the output. Healthy Greetings from the Netherlands! Rob

Whoa. IRF510 has input cap of 180pF, bias RFC is 35uH and transformer secondary is 17uH. These are in parallel so inductance is 11uH. Which just happens to resonate with IRF510 input 180pF at 3.5MHz. So reason for higher output is you have a tuned circuit between driver and PA? For broadband use better to keep the 56R resistor and use the transformer to present a suitable impedance as a load on the driver. Remember the drive 2N3904 is a current source so what you get out is the load impedance x AC current. AC current can at best be that generated by Vcc - Vbe pk-pk... am I right?

At 24V supply a load of 35R should produce 10W output, You just need to match this to 50R nominal load. If you feed output directly to 50R (no transformer) you should get 6W output, One comment I would make is that any MOSFET has a significant capacitive input impedance so that is what your diver is looking at with the bias RFC. I'm following you closely on this as I am struggling also to understand/make a QRP PA. I have the driver done (input -10 to -15dBm, 33-46dB gain, up to +25dBm output, or 4Vrms/300m 50R). Now looking at two output stages IRF530 and 2SC1971 or other.... 12V supply used. For sure I would like to implement ALC, what range?

Well done again Charlie! Regarding adding things to your rigs, I was wondering have you ever tried adding some compression to the audio? I've got a little SSM2167 module which is a mic preamp with compression. I've not had much chance to play with it yet but I was thinking of trying it at the front of my transmit chain to get a bit more punch in the signal. Too much compression can sound terrible but a little bit might just help a bit. Just something else to think about! Thanks again. 73, Nick M0NTV.

I have been playing with transmitters, Class C, using efficiency as my measure of goodness. Closer to theory I guess

Thanks for the update, it’s a great project. I’m surprised by the output transformer on the IRF510, I always thought that the transitor’s output impedance was well below 50 ohms, which doesn’t seem to be the case here. I played around with the cheap Chinese amplifier kit available on eBay/AliExpress (the miniPA70) which are delivered normally with IRF530. I tried out some IRF510 and IRF520, everything else being equal and the output power was definitely the highest with the 530’s Maybe worth a trial ?

For AGC is it possible to send some of the audio into an Arduino analog input for a to d sampling? Grab three digital outputs to a simple decoder chip for an 8 step output which can drive a resistor ladder for a log curve attenuator? Sounds complicated but may be fun to calculate the resistor ladder that can interface with a simple op amp. The a to d also yields signal levels to display on the lcd display. Sounds complex but you could easily model it, build it and display the test log curves on your scope. A to D also yields good data like a simple cw detector to drive a morse decoder.

Revisiting old one is better :) And also because you are using the same blocks with minor changes may be it's time to draw PCB for some o them ;) It will save little time in future but also space for sure.

Fun experiments. Does it seem to you the 2n3904 worked as well as the 2219?

Go full homebrew!

Any chance of installing another band or mode?

How many watts is that heat sink rated for? It looks like, at least, 25 watts. How many watts are you radiating in the form of heat?

First view

Hey Charlie, I really like your content. Do you have a patreon account? I would certainly throw a few bucks your way every month so you can keep these videos going.