An Experimental Method Of Receiver Design
Ok, so this could end up being a post of epic proportions. So here goes it. Actually most of what I am going to write about as been the culmination of a long process of learning and knowledge gathering, where I am going to attempt to bring all those threads together into a whole. At this point, the jury is still out on whether I can actually do it, I have not yet even completed the build, but rather, am going to try and document the entire process from somewhere in the middle to the finish.
To go from the start I would have to explain in detail every little design decision I have made and why. Suffice to say, I have all those details buried deep in my brain, but I am not here to argue with internet experts, I am just trying to work this shit out for myself and understand the process better so that I might eventually build something worth using more than one and perhaps something that could replace the Icom as my daily drive.
So lets start this in the middle. Here is the board I designed, laid out and had manufactured. It is nothing revolutionary, it is a single conversion superhet. The mixer IC’s are gilbert cell types used in cordless phones. The IF amps are SMD version of the 2n2222 and the crystal filer is a 4 pole Cohn Min Loss design using 12mhz computer grade crystals.
The reason for the rubbish level crystal filter is that this is not going to be my forever receiver build and design, it is first and foremost an experimental board whose purpose is to learn as much about superhet design as I can and to try out some methods test and measurement based around using the nanoVNA for much of the measurement, especially for impedance.
So first things first, fire up the crystal oscillator and find some crystals that are close enough to each other to make the filter with. Don’t you love my oscillator, yeah even I used to homebrew like its 1932 and do Manhattan construction on copper clad board like some kind of peasant who could not design a PCB in some ecad. HAHA,. Actually, seeing this board has been getting a workout of late, i have designed one up and will get it made next time i send of boards. Might as well make something pretty looking 🙂
So here are my selection of crystals. The numbers are the last 4 digits of the frequency readout, so the split between the highest and lowest crystal is 13hz, which is a pretty tight grouping, typically i am looking for a spread no more than about 50hz. Crystal filter building can be hit and miss, but I am trying to work on a simple and yet repeatable method of making filters even with the rubbish computer grade crystals we can get now.
Having a fairly accurate frequency counter really helps here to get crystals as closely matched as you can. This one was rejected because it was 7 hertz different from the rest.
Now I know a 4 crystal filter is going to be kind of rubbish, but, when all you are trying to achieve is learning and a method of design, test and measure, it will be good enough for that. So the above image is the Z plot and 21 gain plot for the filter terminated so of course it looks rubbish.
The 2 markers show kind of where the passband is and if you look at the Z lot you can see that the impedance is all over the place with a minimum of 75 ohms and a maximum of 230 ohms across the passband. This is not going to be very easy to get a match with, but like i has been doing with the other crystal filters I have been attempting to build, I will split the difference and attempt to match the output of the IF amp to maybe something like 150 ohms and just see how the losses look. If its less than 10dB i will be happy enough.
The design goals for the IF amps was to have 20dB of gain. They were designed in LTspice because that hurts no parts LOL. Ignore the transformer ratios because they will need to be measured to see just where things are at. But, as for the gain of the amp, well its pretty damn close to having 20dB gain.
The input signal was 53mV RMS. Ignore the p-p voltage its actually displaying wrong. Shit scope what can i say.
With the output of 520mV RMS and a little math that looks something like L = 20 × log (voltage ratio V2 / V1) and you get 20dB gain. Bloody dead nuts on.
Measured the input impedance and its about 85ohms.
Measured the output impedance and its about 40 ohms.
And this is where we are currently at, needing to wind some transformers to see if all this measuring, testing and thinking like a big boy actually means anything in reality. But for now, things are looking ok and we have all the numbers we need.
RF IN is 50 ohms.
Mixer Input and output is 1500 ohms.
IF amp is 84ohms in and 40 ohms out.
IF Filter is 150ohms in and out.
And this post is 900+ words and 11 images HAHA and we are not even ready to feed some signal into it and see if it actually works. But hey, this is what learning is all about. Measure once, cut twice HAHA. Only time will tell now if i have my shit together.
Rather than play software and try and get quadrature working on the ESP32, i busted out the Arduino and jumpered it in with dupont wires.
I do not know why my scope channels are so off, swapping the probes about gets the same values in reverse, for some reason, channel 2 is measuring lower than channel 1. MEH
Oh and this is what I am working on. Voltages all work out, i just need to put some signals into it and see if i can get anything intelligible out of it. The phasing receiver is a mix and match of stuff by plenty of others. I did not know what to set the gain of the polyphase input amps, so i just winged it and went with 10X. I have seen everywhere from 1x to 50x so its anyones guess until i measure whats coming out of them. I probably have 5v p-p to play with in the opamps.
So it sort of works. Not very good, But i think i can make it better.
This is the 2nd time I have assembled this board. First one was not very good on 40m. Made a few changes to the transformers and its much better, its even still got some gain left at 20m.
The drive level is equal to the output of an SI5351. All that is left to do now is to assemble the WSPR Brick, put it on the air and watch the spots come flowing in HAHAHA.
You would think that I would not make such simple mistakes when laying out a board. But yeah I put the RF in to the filter to the top of the divider, not into the middle of it where it belongs. Anyway, not a total disaster, cut a track and solder in a bodge wire and its fixed until its broken.
Figured while I am at it, to update the schematic and gerber files and put them all up here for anyone interested in building their own, now that I know things are right and should work without much thinking. The filter is a double tuned circuit, the design software for this type of bandpass filter can be found in EMRFD and I have it for download here on my blog.
This has now opened up some possibilities. A complete remote control home brew station perhaps? I think it can be done. I think I can even stream audio over wifi quite easily with a MEMS mic and an I2S sound card on the other end and an audio streaming library, then feeding the audio into an SSB transmitter. More to come, but next up will be to expand my 2 button interface into something more resembling a bunch of buttons on a 1990’s looking website HAHAHA. I think i will have to learn some QT programming or Python or Ruby and make a nice looking interface.
Ok first off, there is no full schematic for this board, I have already moved on to adding new stuff to it and changing how some parts of it work so I can add in new features. So the best i can offer is this extract that shows where the pins on the micro go to.
Secondly, I wont be releasing my code, the reason is that I do not program using the Arduino IDE and teaching everyone how to program in GCC, to install all the tools required, and uploading the code using an external programmer is beyond my ability to teach anyone let alone everyone. There are just to many things that can and will go wrong and it will only make my life miserable trying to help everyone.
That said, the ESP32 can be programmed in Arduino and the actual basic code is not that hard to write yourself. Turn a encoder and it steps through the relay combinations.
With these relays, I would not put much more than 50w though them. Though the data sheet says they will take 1A but that makes them marginal at 100w. As it stands I think 50w is a very safe bet.
Relays: Omron G5V-1 5V subminiture.
Transistors: PDTC114ET pre-biased SOT23
Caps: 1206 or 1210 Ceramic Multi-layer 1Kv or higher rated, for tuner, 50V 100nF for bypassing. 1uF across the coils of the relays.
Voltage Regulators: 5V and 3.3V 1117 variants SOT223 package.
Toriods: T68-2 T50 would be suitable also, its just what I had, 2,4,8,12,20, 24 and lots of turns.
Top side of the board as 3d rendered.
Bottom side of the board, 3d rendered.
The board as built by me.