This radio was previously described on the old RadioBoard forum under the title “A Simple Superhet”, and is re-presented here in more detail than previously. It is a general coverage broadcast receiver.
The radio features digital tuning and frequency readout with memory channels, a low VHF first intermediate frequency for good image rejection, and broadband circuits eliminating alignment. No unobtainium parts are used. Commercial products are used where there is an advantage to do so.
The radio is constructed modularly. Three 50 x 80mm BusBoard systems SMD prototype boards ( 200 x 100 format ) hold the RF and audio circuits. The power supplies are on simple perfboard. The completed modules and VFO are installed on a breadboard “chassis” designed on TinkerCad and printed on a simple and inexpensive 3D printer with PLA plastic, thus no complicated metal work is required. A rotary encoder is used for tuning, and directly mounted to the front panel, eliminating the tricky mechanical alignment problems associated with obsolete LC or Varactor VFO’s. The main tuning knob is also 3D printed. Volume is set by an electronic attenuator controlled by a simple DC voltage, eliminating cumbersome shielded wiring to the front panel.
The heart of the radio is a commercially produced QRP Labs VFO kit, available for about $33 USD, using the now ubiquitous Si5351 frequency synthesizer chip. The VFO can generate a tunable LO well past 200 MHz, quadrature output if required, tuning steps as fine as 1 Hz, at a square wave level sufficient to drive commercial level 7 mixers without additional amplification. A digital frequency readout and memory channels are provided. Further, an additional fixed frequency local oscillator is provided. The firmware allows the frequency displayed to account for the IF offset from the tunable oscillator. The kit is supplied with a white on blue background LCD, but 5 volt type 1602 LCD readouts are available in a wide variety of color combinations. My radio at this time uses a black LCD with red digits.
The Si5351 has many good features and one, perhaps two, significant drawbacks - there is a degree of cross talk between the oscillator channels that is inherent to the device. The output is a square wave, by definition harmonic rich. This is ideal for diode mixers. The Si5351 has a DC component on its output, so a DC blocking capacitor of 10 or 100 nF should be used in circuits where this DC component cannot be tolerated.
For up converting superheterodyne receivers, the general rule of thumb is that the first intermediate frequency should be at least 1.5 X the highest frequency to be received.
The intermediate frequencies in this receiver are 45 MHz and 450 KHz. Crystal filtering is used at the first IF, and a ceramic filter is used at the second IF. The use of a low VHF first IF, greatly simplifies the construction of the receiver, while enhancing its performance compared to older design techniques. Tunable LO injection is 45 MHz higher than the frequency to be received. The use of the low VHF IF puts the image frequency 90 MHz above the received signal, thus a simple low pass filter is satisfactory to protect a strong first mixer, while eliminating front end alignment and tracking issues.
The low pass filters and diode double balanced mixers used are commercial products by Mini Circuits. The IF amplifiers are 50 ohm broadband amplifiers. A TDA1072 Integrated Circuit handles final IF amplification, AM detection, AGC, and provides a meter output voltage that is used for signal strength indication and an additional AGC loop. Audio output is from a DA7052A BTL Integrated Circuit.
Each of the modules in the radio will be described in further detail in subsequent posts.
This receiver was built expressly for hi fi reception of the big one, 5085 Mhz, and now that it is permanently QRT, with all of the talent and programming having moved to WRMI, I find myself in the unfortunate position (literally) of being in between the two 31 meter beams, and at a less than optimal distance from the transmitter, to add insult to the injury.
So, after a few weeks of observing received signal strength at my lake house QTH, which is typically 50 to 60 dB less than what I was getting, I pulled the four pole 9 KHz filter and replaced it with a four pole 6 KHz filter. The new alternate 4980 KHz frequency puts in a pretty good signal here (although it is beamed at south America, so I am off the back side), but is often compromised by a strong RTTY signal on 4987.5 KHz that would spill over into the edge of my wide passband during QSB of the main signal. The 6 KHz filter has stopped that spill over.
Not to wax philosophical, but whenever one door closes, another one opens. Sensitivity is marginally better with less noise in the passband, enough so that I can now Q5 copy some weak MW stations that were just too weak previously. And since I had no 450 KHz 6 KHz filters, I installed a 455 KHz filter that I pulled from something a long time ago, and I have some SMD crystals for 44.545 MHz, which means with sufficient motivation, I can make a crystal controlled second local oscillator to go from 45 MHz to the second IF at 455 KHz. That could improve performance even more since there is unavoidable cross talk in the Si5351 oscillator channels.
But another anomaly has reared its head - like Coildog experienced with his ceramic filters - my filter is off frequency by about 1 kc. It has the same i/o impedance as my 9 KHz filter, which was spot on, so it's not an impedance matching problem. I looked at the specs, and sure enough, the specification for nominal center frequency on these inexpensive four pole ceramics is +- 1 KHz, so this is within spec. Other than annoying, it's not that big a deal - I just adjusted my second LO by 1 KHz to compensate. Easy to do with the Si5351. But it annoys me, so I ordered another group of five of the 455 KHz 6 KHz filters, and hopefully one will be on center frequency. If I get two or more on a common center frequency, I might try putting two in series.
Or I might ditch work on the simple communications receiver, take what I have learned from building this receiver, and make a higher performance broadcast receiver that is better equipped to deal with these annoyances.
After about fourteen months of fairly heavy use, I had my first component failure - the TDA7052A BTL audio amp. The symptom was intermittent audio output, restored by a slight adjustment of the volume potentiometer.
I incorrectly diagnosed this as a failure of the volume control pot, as it has a DC voltage applied to it which I suspected as having caused the failure of the pot. Replacement of the pot did not cure the problem, so I replaced the chip, which restored normal operation. I substituted a 100K linear pot for the previous 10K audio taper, to at least reduce the current in the pot, and spread the control range out, which was pretty compressed in the prior iteration, but not annoying enough to replace it for that reason alone.
In the development phase of the receiver, I stressed the chip pretty heavily exploring its voltage range, at one point making it so hot I feared it would unsolder itself. I probably should have replaced it then, but it continued to work so I let it go.
Replacement of the chip has restored normal operation. Even though these chips are rated for relatively high voltage operation, I do not recommend this. I think 5 - 6 volts DC is optimum.
In addition to the 1980's era books by Collins Radio Co, ("Single Sideband Systems and Circuits"). Dr. Rohde ("Communications Receivers, Principles and Design"), and Wes Hayward ("Introduction to Radio Frequency Design") there are a couple of good books by a less well known author:
Radio Communications Receivers, Cornell Drentea, 1982
Modern communications receiver design and technology, Cornell Drentea, 2010
The first three titles are probably only available in print, but the copyright on Mr. Drentea's books does not look to be vigorously enforced and they are available as pdf downloads. I have a pdf of the 2010 book, and it is 487 page tome that is terrific, so it is out there somewhere and well worth looking for. Mr. Drentea's credentials are top notch, and it seems pretty clear to me that he is a ham, like most of us. I am reluctant to post the pdf here because it is clearly marked as copyrighted.
The answer to the original question "How much gain is too much at 455 KHz?" is "It depends".
The overall receiver gain for good sensitivity probably needs to be at least 100 dB, but there is no law that says you have to get all that at RF. You really need only as much RF/IF gain as is necessary for the amount of AGC control range desired, if AGC is even desired. That could be as little as one IF stage, or lots of them.
For additional context, the issue I was investigating with broadband IF amplifiers was basically "what is the minimum number of IF tuned circuits required in a single-conversion superhet with a regenerative detector?" I thought I might get it down to one -- an LC filter on the mixer output, followed by broadband IF amplification, followed by a crystal-controlled regenerative detector at IF -- but I never succeeded with the "broadband IF amplification" stages in my trials, which admittedly were not carefully designed. I got tired of trying and failing, and finally just went with the old-fashioned way of making every IF amplifier tuned.
Well, I'm not entirely sure I agree with his premise, that untuned amplifiers necessarily lead to an increased background noise level.
In my particular receiver, it may be important to note that I used low noise bipolar transistors - mmbt5179, the smd equivalent of 2N5179. The strip is fixed gain, with attenuation, as needed, handled by the PIN diode attenuator at the front end.
My IF amplifiers are split over two frequencies, 45 Mhz and 450 KHz, and are filtered at each end - I have a 20 KHz two pole crystal filter at the entrance to the IF strip, and a 9 KHz four pole ceramic filter at the tail end of the untuned IF strip.
I am a believer in the benefits of a filter at the tail end of an IF strip, even an LC filter. This is pretty well known, I think, and probably mentioned in every textbook on superheterodyne receiver design.
I have also found tuned IF strips to have a tendency to oscillate. I intend to try common base amps, as you suggest, the next time I make a tuned IF strip from bipolar transistors,
At the moment, I have a decided preference for broadband designs.
Following up on previous discussion in this thread about untuned vs. tuned IF amplifiers: I commented earlier that I didn't have much luck with untuned IF amplifiers, whereas the receiver of this topic is using 2 untuned IF amplifiers with success.
I found a discussion on another forum that briefly explains some of the risks of using untuned IF amplifiers, at https://www.vintage-radio.net/forum/showpost.php?p=1465261&postcount=4 . Quote:
I think that is partly what I was experiencing when I had very noisy results with untuned IF amplifiers. Furthermore, my use of an unbalanced mixer meant that there was a very large LO signal still present at the mixer output, which may have driven the following untuned IF amplifiers into saturation, further increasing distortion and noise and obliterating the comparatively tiny IF signal that should have been amplified.
Tuned IF amplifiers vastly improved the performance (using 2N3904 transistors running at 1.2 volts) in my experience.
I would guess that untuned IF amplifiers require more careful design than tuned IF amplifiers to keep the noise within acceptable levels. On the other hand, tuned IF amplifiers may be able to provide acceptably low-noise gain in a wider variety of sub-optimal conditions -- such as impedance mismatches, low signal levels, noisy transistors, and transistor bias -- thanks to the narrow-banded resonant gain of a tuned amplifier. Creating a chain of several tuned IF amplifiers brings with it a risk of oscillation, but I think the easiest way to avoid that is to use tuned common-base IF amplifiers, which all but eliminate unwanted oscillation thanks to the shunting of the instability-causing collector-to-base Miller capacitance to ground.
I'm afraid I am not yet familiar enough with the NanoVNA to start a thread on it. I haven't really used it much yet. The main reason I bought it is for adjustment of the small coupling loop on a small transmitting loop antenna, for a perfect 50-ohm resistive match (and 1:1 SWR) at the desired frequency. This is a pretty fiddly operation requiring small and precise geometric adjustments to the coupling loop's size, orientation, and position. I had given some thought as to how to do this with completely homebrew equipment (noise bridge and dip meter), but I finally realized it would be vastly easier just to use a NanoVNA, so I bought one. As a test, I grabbed a USB cable and rolled it up into a 3-turn coil. By hooking a small coupling loop onto the NanoVNA and holding that loop inside the 3-turn coil, I could see that the loop was resonant around 50 MHz, and by looking at the real-time SWR graph, and adjusting the size and position of the coupling loop, I could very quickly achieve a perfect 50-ohm resistive match (1:1 SWR). That quick experiment gives me confidence that I will be able to do the same thing with an actual small transmitting loop antenna.
I'm also hoping I can use it to measure crystal parameters for building a crystal filter.
I should have been more clear - I still use the 2M FM transceiver ...
At the time of construction, I used the dip meter as an RF wavemeter. The TX strip was from a kit ( GLB Electronics ) a classic frequency multiplier type from 12MHz ish to 2 meters, so it was important to make sure the multiplier stages were being tuned up on the right harmonic which the dip meter was well suited for, as well as for tuning the crystal oscillator tanks, the RF output tank, etc.
I still use it as a wavemeter for a quick check for high level spurious emissions, etc.
I was never successful in using it to reliably check the resonant frequency of tuned circuits, likely for some of the reasons you mention above. Since I tend to use toroids for tuned circuits, which are self shielding, I don't think dip meters get a good coupling into the field, unless the toroid is so large that the dip meter coil can be inserted through the middle of the core.
Dip meters would probably warrant a thread of their own, because, as you note, they can be tricky to use and give misleading indications if you are not careful.
I would be interested in a thread on your NanoVNA and it's capabilities for the hobbyist. There is a VNA adapter for the Red Pitaya but it is a bit pricy ( I'm cheap when it comes to hobbies ) and the price of the NanoVNA looks real appealing.
By comparing with my NanoVNA, I discovered (as has probably been long known by more experienced experimenters) that dipping an LC tank, having a toroidal inductor, by using a shorted link into the toroid can cause a huge shift in the resonant frequency. I had an FT50-61 toroid with maybe 30 turns on it, resonated with a small plastic variable capacitor.
First, I tested the resonant frequency by placing a single link turn through the toroid and connecting it to the NanoVNA's input. I assume the NanoVNA has a 50-ohm input, so the link winding is essentially terminated in 50 ohms, which is then transformed up by the square of the turns ratio (1 link turn : 30 main turns), so the tank is not loaded heavily. The resonant frequency was around 2 MHz.
Next, I tried putting a shorted turn through the toroid - a 5 cm wire formed into a circle and joined at the ends. I placed the sense coil of my dip meter inside of the shorted turn and near the wire. I got a good dip -- but at 5 MHz! This is way, way off from the NanoVNA reading.
Finally, I tested again with the NanoVNA, removing the shorted turn. The tank was resonant at 2 MHz. Then, I inserted the shorted turn, and surely enough, the NanoVNA showed that the resonant frequency jumped up to 5 MHz.
I didn't realize that a shorted turn has such severe consequences. So this can be quite misleading if using a shorted turn to couple from the toroid into the dip meter.
I have 2 dip meter circuits, one more sensitive than the other. The more-sensitive dip meter can sometimes, but not always, dip a toroidal LC tank if the coil is placed near the wires connecting the L and the C, if there is enough magnetic coupling to the wire (which depends on the frequency, and the physics of the dip meter's sense coil). The less-sensitive dip meter cannot dip a toroid in this fashion.
I had thought that using a (shorted) link coupling onto the toroid was an easy way to get enough coupling to the toroid to allow a good dip to be seen on a less-sensitive dip meter. But the dip obtained in this way is (or can be) way off target. So, I'm now wondering if there's a better way for a less-sensitive dip meter to effectively dip a toroid. I briefly tried inserting some series resistance (I think around 220 ohms) into the link winding, to reduce the loading on the tank, but then I couldn't get a dip. Maybe a 100-ohm variable resistor in series with the link winding might allow adjusting the tank loading until a sufficient dip can be obtained.
Yes, let's keep that idea in mind. Way back in the days when Yahoo Groups was still in existence, there was a group called regenrx devoted to regenerative receivers, and there was some talk about the bandwidth at critical threshold. It finally occurred to me that I could actually measure that with simple test equipment:
Bypass any final low-pass audio filtering circuitry on the regen.
Connect the regen's AF output to the microphone input of a PC sound card.
Adjust the regen to critical threshold at the frequency of interest, say 7100 kHz.
Run an AF spectrum analyzer on the PC, with a "peak hold" function.
Place a strong BFO near the regen, tuned slightly below the regen's frequency, say at 7090 kHz. The BFO's radiation turns the regen into a direct conversion receiver.
Place a VFO (like a GDO) near the regen, and slowly and manually sweep it from the BFO frequency upwards, from say 7090 kHz to 7200 kHz.
The regen will receive the radiated signal from the VFO and amplify it via Q-multiplication, with maximum amplification at the resonant frequency of 7100 kHz. At the same time, the BFO causes the regen to convert the Q-multiplied RF directly down to AF. By sweeping the VFO from 7090 to 7200 kHz, we generate an audio tone at the regen's output from 0 kHz to 110 kHz. Then, on the PC's AF spectrum analyzer we see a range from 0-96 kHz, corresponding to the RF frequencies of 7090 to 7090+96 kHz. The "hold peak" function of the PC spectrum analyzer records the maximum signal seen at each frequency, which will be at a maximum when the VFO sweeps across that frequency.
These are not laboratory-grade measurements (the PC sound card will not respond to very low frequencies; there is an unspecified amount of coupling among the VFO, BFO, and the regen's LC tank; the VFO output will vary slightly with frequency), but still it is a quite useful way of getting actual numbers (and pretty graphs!) of the selectivity of a regenerative receiver. That selectivity of course goes down the drain in the presence of strong signals, but that's another topic (and indeed, this selectivity reduction could probably be measured with similar techniques).
Anyway, it's been over 10 years since I did this kind of measurement, but it might be fun to do it again and make a YouTube video.
I agree completely.
I have a dip meter. When I got out of college, I built a 2 meter FM transceiver with little more than some VOM's, that dip meter, and an ancient 10 MHz oscilloscope. I still use it ....
Another great piece of test equipment is the receiver in the Yaesu FT-817.
A thread on inexpensive test equipment and how to leverage it would be great.
The ancient 2 meter rig - my first "big" project .....
My first SSB rig from about the same time. It was originally in a nice cabinet before I dismantled it. I got it out a couple of years ago and hooked up the receiver on a breadboard just to see if it still worked. It did .....
Then I got interested in airplanes and quit building for a couple of decades ....
That said, you can still do a lot with simple, homebrew test equipment. In particular, a good GDO (grid dip oscillator, or dip meter) is pretty handy. When used as a dip meter, it can measure the frequency of a non-energized LC tank through energy absorption. When connected to headphones, it can serve as a general-coverage direct-conversion receiver, which allows you to measure the oscillation frequency of a nearby oscillator (or scan for spurious oscillations). These two features can be used for superhet RF-LO tracking alignment. Furthermore, the GDO can also serve as a signal source (optionally with modulation) for receiver testing and alignment. I built a sensitive GDO by adding an automatic sweep function (frequency modulation), similar to the idea of the German DipIt meter: https://www.qrpproject.de/UK/DipItUk.html . Here's a video of my dip meter in action, measuring the resonant frequency of a non-energized LC tank. https://www.youtube.com/watch?v=nVi3L0BwaXg&t=0s . I have a write-up on my blog, but the images seem to have disappeared from my blog. I'll update the blog sometime (soon?).
Also, a PC audio-frequency spectrum analyzer (using the PC sound card's microphone input) is also useful and can be used for checking bandwidth of narrow filters like crystal filters or a regenerative receiver just below critical threshold. Here's an example: https://www.qsl.net/pa2ohh/04audio.htm .
I also sometimes use my RTL-SDR (with homebrew upconverter) as a makeshift spectrum analyser, albeit with the presence of images. Connected to my smartphone with a USB OTG cable, the RTL-SDR (plus upconverter) is quite compact, and I just place it near an oscillator circuit to scan for the oscillation frequency.
For example, to measure the frequency of my superhet's LO, I place the GDO near the LO and turn off the GDO frequency modulation, using the GDO as a DC receiver with headphones attached. I tune the GDO until I hear the LO (making sure I am tuned to the fundamental and not a harmonic), and zero-beat the GDO with the LO. Then I hold the GDO near the RTL-SDR and check the frequency on the RTL-SDR (making sure I am viewing the real GDO signal and not some image or unwanted mixing product). A similar procedure is used to measure the resonant frequency of the (non-oscillating) RF LC tank. Somewhat tedious -- but it works.
Odds and Ends
Since I built this receiver, Mouser has begun carrying a somewhat wider selection of the 45 MHz ECS filters. I use a two pole +- 10 KHz filter here; +- 7.5 KHz two and four pole filters are now carried by them. If I were starting from scratch, I would probably go with the +- 7.5 KHz four pole filter.
The receiver might benefit from a crystal controlled second conversion oscillator. I have crystals to convert from 45 MHz to 455 KHz, but I don’t have any suitable 455 KHz second IF filters. I can’t find crystals to convert to 450 Khz. The radio performs satisfactorily for my purposes as is, so I have not bothered to “fix” this. For now, the second IF stays at 450 KHz.
I used a +- 4.5 KHz ceramic filter at the second IF because in the evenings I like to listen to WTWW 5085 KHz, and their transmitted signal is excellent. The wide filter gives excellent fidelity for broadcast stations. A narrower filter might be desirable in some circumstances, but I want the fidelity and am willing to accept the other shortcomings of a wide filter.
As noted previously, for DX-ing above 19 or 20 MHz, an RF amplifier would improve the sensitivity. I have some Mini Circuits MMIC amps that would work nicely for this purpose, but this is not a high priority for me at the present time.
The meter is a Calectro part. I got a whole sackful of these meters in various movements, cheap, at a hamfest. They look nice, and mount easy in a single large hole with a lock nut from behind. The meter used here is 0-1 ma movement. Microamp meters could be used with a suitable meter shunt.
The small locking connectors are the XH 2.54 mm type. They are a poor match for, and not very convenient to use, on these prototype boards. They are cheap, however, and do make it very convenient to remove the boards for modification without having to unsolder a bunch of connections. I feel they are worthwhile. If there is a better solution, I would like to know of it.
I used store bought mixers, but they can also be homebrewed. I used a store bought low pass filter, but that, too, can be homebrewed. Designs for homemade Si5351 frequency synthesizers are also common. I just found it convenient to use store bought parts to save time. I have serious doubts that I could make two identical mixers, but Mini Circuits can make millions of them and they will all perform the same.
SMD / SMT components are used whenever possible here. There seems to be a prevailing attitude among home builders that SMD components are more difficult to use than leaded through hole components, and I also held that perception. When I was working on a simple SSB transceiver, I was encouraged to make the switch to SMD by more experienced builders, and, I have to say, it was really good advice and I have not looked back from that decision to switch. The parts are much cheaper, the circuits work better, and assembly is faster and easier. I have come to prefer 0805 size parts. 1206 parts seem a little big to me. When necessary I use 0603, and sometimes 0402, although the latter is difficult to use on these prototype boards. Two tools are essential: a hot air gun makes removing smd components a snap. A good temperature controlled soldering iron is a necessity. I just use a regular tip with SMD components. Some people say to use a fine point tip. I don’t think it is necessary. I have found that I like solder with a little bit of silver in it. I despise lead free solder and won’t waste my time with it.
Test equipment is really an unfortunate necessity when fooling with RF. I have a lot of VOM’s, all cheapies. A set of tweezer type test probes are helpful with SMD parts. For everything else I use a Red Pitaya StemLab 125-14, the 14 being the bit depth of the device. This is a great device - it can do two oscilloscope channels to 60 MHz, two signal generators to 50 MHz, two spectrum analyzer channels to 60 MHz, with frequency sweep and all sorts of other stuff in a package about the size of a pack of cigarettes. It’s just a terrific device - I can do as much with the Red Pitaya on the kitchen table at my second home as I can in my ham shack. This receiver is literally a kitchen table receiver - it has never left the kitchen table at my second house.
It works very well. The attached video is Radio New Zealand Pacific on 9700 KHz received in the south central USA about forty five minutes after my local sunrise.
Chassis - part two May 17, 2022
The controller for the Si5351 frequency synthesizer is a compact unit that is sized identically to the 1602 LCD PCB and piggy backs unto it; the actual Si5351 frequency synthesizer piggy backs unto the controller / LCD assembly, and then the whole smack mounts to the back of the front panel of the breadboard. It requires 5 volts to operate.
Only four wires are required to connect the rotary encoder / switch assembly to the controller. The momentary contact push button switches on the front panel are a convenience feature as they duplicate the two momentary contact switches that are already present on the back of the controller board. These two switches and the encoder control all the functions of the VFO assembly. The pushbutton switches have pins that are a convenient fit to a female two pin XH connector, making it easy to remove or install the VFO assembly from the panel without having to unsolder connections or remove the switches. The two momentary contact switches are threaded into the front panel by making their mounting hole about a mm smaller than would ordinarily be used.
Because volume is controlled by the electronic attenuator within the TDA7052A, and there is no need for a precise mechanical alignment of the main tuning element with the VFO, and finally broadband circuitry eliminates the need for front end peaking, wiring of the front panel is clean and uncomplicated. The panel is rigid, but it doesn’t matter - tuning or volume won’t change even if it were to flex.
Wiring in between and to the boards is non critical, because all of the wiring, with the exception of the high level audio output from the TDA7052A to the speaker jack, carries only DC. They are twisted and routed for my convenience. The wire tie wraps from old bread bags are used to hold the bundles in place.
When printing the breadboard, support has to be enabled to make sure the LCD and meter opening come out correct. The pictures show a breadboard after it comes off the printer and the supports are removed. I make identical sized models of the boards to be used in a project, put square standoffs at each corner of the board to support a circuit board, and then merge this completed model(s) with the chassis which results in correctly spaced standoffs to assemble the circuit boards unto. Pilot holes can be modeled into the standoffs, but I find it easier to simply use an empty board as a template, and touch a pilot hole into the plastic a mm or two deep. Coarse sheet metal screws will then tap and self thread into the stand off and securely hold the board in place. These can be removed and reinstalled many times without damaging the tapped threads so created. I use washers, also 3D printed, to spread the force of the screw unto the circuit board.
I use a couple of stick on rubber bumpers attached underneath the breadboard to slightly elevate the front of the radio relative to the rear. At some point I’ll figure out how to make a tilt bail, but I’m not there yet.
The knobs, other than the main tuning knob, are just cheap colored knobs from AliExpress that will push over a splined shaft. The thickness of the front panel is about 3 mm. Encoders, displays, meters, and potentiometers don’t really care about the thickness of the panel, but small jacks like the 3.5 mm speaker jack and the small pushbutton switches do, so these need to be looked at carefully when laying out a panel.
Yes, I think gear reduction drive, pulleys, drums, etc., are all quite feasible on 3D printing, probably would be quite easy. A rack and pinion type system to drive a pointer might also be possible.
If you look at some of the designs in the TinkerCad library, there are concentric knob designs where two pots or encoders are coupled by gears.
I've printed it and it works fine - I just have not had an occasion to use it.
That said, the advantages of digital tuning are enormous, it's cheap and easy, there are no alignment issues with any of the involved parts, no backlash, selectable tuning steps, and did I mention it's cheap and easy.
Write your receiver up and post it!
Given your apparent preference for digital tuning, I suppose that my question may fall outside of the scope of your interests, but: Have you ever tried printing geared reduction drives for variable capacitors? I'm thinking it should be quite doable, though the gears might need to be somewhat large due to lack of precision in the 3D printing process.
Or, instead of gears, perhaps a dial-cord type mechanism might also be feasible to create with 3D printing. Dial-cords are really quite interesting mechanisms, as a quick web search will reveal. Here's an example from https://onetuberadio.com/2021/02/11/stringing-the-dial-cord/ :
That looks quite feasible to recreate with a 3D printer.
With all the talk about your superhet, I dusted off my all-transistor superhet (2.8 MHz IF) and confirmed it's still working. One of these days I want to rebuild it and package it into a pocket-sized case. A real dial-cord with a working analog pointer display would be icing on the cake!
Chassis May 14, 2022
The breadboard “chassis” for the receiver was printed on an inexpensive 3D printer ( Tronxy P802M ) that I purchased for about $110 USD and was delivered as a knockdown box of parts, requiring assembly. The heated bed is 220 mm x 220 mm and it can print vertically to 240 mm. I also have a Creality CR-10 S5 which is a much larger 500 mm x 500 mm x 500 mm printer, and considerably more refined in appearance and operation, but I actually prefer to print on the cheap Tronxy, provided the object can be fitted into the available print space.
The breadboard and tuning knob were created in TinkerCad, which is pretty easy to work with. TinkerCad is web based and runs well on my simple two core Celeron laptop. I am a complete beginner, hence the very simple and straightforward “design” of the breadboard. I like TinkerCad better than working metal, although that is not really saying much as I don’t enjoy metalworking at all, so I will be 3D printing as much as possible for the foreseeable future.
For this radio, the breadboard size is 190 mm x 190 mm. The vertical height of the panel escapes me at the moment, but it is tall enough to comfortably accommodate the 1602 LCD and a large fluted tuning knob. It, and the tuning knob, were printed with ordinary PLA plastic. One of the nice things about 3D printing is the variety of colors available - don’t like your present color scheme, just print up another. I’m still tinkering with color schemes. Also, it’s very cheap compared to metal.
The rotary encoder for tuning uses a 15 mm D flat shaft. In the old days, knobs that fit on a D flat shaft used a setscrew or had a slightly arched metal piece in the knob that functioned as a spring to hold the knob tight to the shaft. Here, instead of drilling and tapping the knob for a set screw, I used a “spring” which is just a narrow gauge piece of solid wire placed length wise over the D flat part of the shaft, and the knob is carefully pushed over it, compressing the wire in between the flat of the shaft and corresponding part of the knob, for a very solid and rigid fit of knob to shaft. This obviously requires the knob to have the corresponding and inverse D flat hole shaft, somewhat complicating its design and manufacture. A splined encoder shaft might be preferable, and I will probably order some in the future.
Encoders can be free spinning or detented, and available in a wide variety of pulses per shaft rotation. Since this is a broadcast receiver, and broadcasting is channelized, this radio uses a detent style of encoder. I usually set the VFO to tune in 10 KHz steps, and select 1 KHz steps if I need to fine tune. Tuning steps down to 1 Hz can be selected. Even though detented, the large knob with finger indent, allows for easy, rapid, one finger tuning.
I initially thought 1602 LCD’s were a commodity item and, except for color and voltage, were all the same, but that is not true. After purchasing a whole bunch of them to play with, it appears all of them have the same size PCB, but the dimensions of the actual bezeled display can vary. So far my experience has been that blue and greenish yellow background displays have identical display dimensions. When you start getting into other colors, or a black background, the latter displays are about 3 mm larger in the vertical dimension. This extra height is displaced downward to the lower mounting holes, making the display bezel edge virtually adjacent to the mounting holes. This would not be a problem when mounting from the rear, but is problematic with front mounting. The attached picture clearly shows the difference in size between a black background, red character display, and a blue background, white character display.
May 4, 2022
The Everything Else board.
The Everything Else board does, well, everything else.
It is fair to describe it as a stand alone, fixed frequency, radio receiver. The detector and audio board receives the second IF output, filters it, and applies the filtered RF to a TDA1072 integrated circuit. The TDA1072, by itself, is a complete single conversion AM radio receiver system containing an RF amplifier, an oscillator, a double balanced mixer, an IF strip, an envelope detector, an AGC system, and a meter driver. Maybe other stuff.
Here, I use only the IF amplifier system, the AM envelope detector, the AGC system, and the meter driver. The R. L. Drake SW-1, a similar entry level up converter, uses the TDA1072 in its entirety, using its on board balanced mixer to down convert from 45 MHz to 455 KHz, and its onboard oscillator as a crystal controlled conversion oscillator. I’ve owned an SW-1 since they were released, and after careful consideration, believe the RF performance of the homebrew set described here to be as good as, or better. That anecdotal comment is not to denigrate the entry level Drake which works well, but to affirm that the set shown here can perform to the level of well designed, commercially produced, entry level receivers, and considerably better than one would expect given its extreme simplicity - two RF transistors and two integrated circuits essentially comprise the whole receiver.
As constructed here, the Everything Else board works at 450 KHz. The second IF output at 450 KHz feeds into an impedance matching network, then into a 4 pole Murata ceramic filter, with a very wide bandwidth of +- 4.5 KHz. This wide bandwidth was deliberately chosen for “high” fidelity reception, but allows more noise into the receiver on weak signals. This is a trade off I was willing to make for fidelity. The skirts of the filter are surprisingly steep for a cheap filter, so that when the receiver is set to tune in 10 Khz steps, stations, even very strong ones, cleanly snap in and out of the broadcast channels without bleedover into adjacent channels.
That part of the TDA1072 that is used here, is used as per it’s data sheet, except for the meter driver circuit, and the input to the IF strip. The data sheet depicts a balanced input to the IF strip, however, a single ended input is used here, with the other input being put to AC ground through a 10 nF capacitor. The meter driver circuitry is as shown on the block diagram a couple of posts back. Used in this manner, the driver circuitry can directly drive the common 0 - 1 ma meters to provide a signal strength indication, and provide a DC control voltage for the shunt diode attenuator at the output of the 45 MHz crystal filter. This arrangement provides very good AGC action; I have yet to detect any evidence of the receiver overloading. It surprises me that Drake did not take advantage of this free voltage to control an attenuator at the 45 MHz IF amplifier in their SW-1.
Audio output, at a pretty good level, is taken from the TDA1072 and applied to a generic J310 source follower; from there the audio is amplified to speaker level audio by a (T)DA7052A bridge tied load power amplifier.
Both TDA1072 and TDA7052 and TDA7052A are obsolete, but still readily available. The A suffix on the 7052 indicates that it contains an electronic attenuator that can be used to control volume. The attenuator can be controlled by a variable resistance to ground, or by a DC control voltage beginning at 2 volts. I tried it both ways and settled on using the DC control voltage. A standard red LED is biased at about 2.1 volts, so was used as a convenient voltage reference. This has the additional effect of providing a clean mute when the control voltage is below a certain level. Further, simple unshielded wiring to the volume control is acceptable, making for cleaner front panel wiring.
The data sheet indicates that the 7052 will operate over a very wide voltage range, and this is sort of true - it just doesn’t work very well. I initially used it at 12 volts and while it worked, it got so hot I thought it would unsolder itself, and power output was not really adequate. Reducing voltage to 6 volts gave goof performance. With a 4 inch speaker in a homemade enclosure, it provides ample power output to fill the upper floor of my home with clean volume, and the chip runs cool. I added another LM317 to my power supply board to supply the audio chip. The VFO runs off 5 volts, and there is a separate LM317 regulator for the VFO that could probably power the 7052, but LM317 are so cheap I just gave the chip its own regulator.
Additional circuitry on the everything else board is an 7808L TO-92 regulator that powers the TDA1072, and the source follower. The unused XH connector is wired to plug in a pot to manually control the shunt attenuator at the second IF. Most components are 0805. There may be a few 1206 or 0603, except for the large electrolytic caps which are leaded. SMD electrolytic caps are a bit dodgy to install on these boards, so I don’t use them much.
The Intermediate Frequency Board
The Intermediate Frequency Board contains: the 45 MHz crystal filter and matching networks; a PIN diode shunt attenuator; an untuned first IF amplifier; a Mini Circuits ADE-1+ as second mixer; and a second untuned IF amplifier.
The initial selectivity of the radio is set by the 45 MHz crystal filter at the first intermediate frequency of 45 MHz. In amateur circles, this filter is commonly known as the roofing filter. 45 MHz is a popular IF, and as such 45 MHz crystal filters are widely available as new from suppliers such as Mouser, and on the surplus market. The filter used here, an ECS 45 K20A is still a current production part. Simple LC matching networks match the filter impedance out of, and into, 50 ohms. The impedance and capacitance of filters can vary a great deal, so it is important to get the data with the chosen filter so it can be properly installed in circuit.
After the 45 MHz roofing filter, the signal proceeds to a PIN diode shunt attenuator, controlled by a DC voltage from the TDA1072 meter driver circuit, or, if desired, manually by a potentiometer. Here, the meter driver circuit provides sufficient drive voltage to prevent overloading of the radio, so manual control is not used, although the circuitry is in place on the detector board, needing only a pot to be plugged into the provided connector.
These PIN diodes are TO-92 devices purchased decades ago and pulled from my junkbox. I do not know their specifications, and I suspect they were intended for switching purposes, which makes them less effective as an attenuator. I have purchased some modern SMD PIN diodes, but the installed devices work sufficiently well so they are still in service in the radio. One item to note about PIN diodes - they are frequency sensitive because of carrier lifetime, so this specification needs to be checked if they are to be used at low frequencies like 455 KHz. At 45 MHz, probably about anything will work as shown here. The shunt diode attenuator circuit is provided - this is a generic circuit and can likely be improved upon.
There are many different types of IF amplifiers and chapters in textbooks are devoted to them. I chose 50 ohm broadband amplifiers as my IF amplifiers for several reasons: broadband amplifiers are generally much more stable than tuned amplifiers; I wanted to be able to use different intermediate frequencies without having to redo the circuit; and I wanted to run the amplifiers at a high constant stsnding current so they would not be overloaded by strong signals.
The IF amplifier circuits are a variation of the W7ZOI /W1FB feedback amplifier circuit and I am running them at about 25 mils standing current. I used MMBT5179 transistors, which in older days were known as 2N5179 (TO-72 case), and PN5179 (TO-92 case). Pretty much any NPN transistor will work in this circuit.
The second mixer is a Mini Circuits ADE-1+. The LO port receives the second local oscillator signal directly from the fixed frequency clock_x output of the Si5351 through an intermediate 10nF blocking capacitor. The IF port of the mixer goes directly into the second IF amplifier. A diplexer or resistive attenuator might be desirable here, but on the other hand, the second mixer only sees the slice of spectrum that makes it through the roofing filter. I was running out of space on my board, so that made the decision for me - the IF port feeds directly into the following IF amplifier.
Output is to an SMA connector on the rear of the board.
Voltage is not critical. The board is hooked directly to the DC input to the radio, and seems to function fine from 8 volts to about 14 volts.