I enjoy listening to the classic rock format on WRMI radio, broadcast on 5050 KHz mid evenings, and 9455 KHz late night. With the DC receiver project winding down, I have enough left over parts to ( I think ) build a simple radio for listening to WRMI under good conditions.
Having inventoried what I have on hand, I think the "design" goals are going to be:
Simple, beginner friendly circuitry, with no complicated circuits, or unobtainium parts;
Self contained except for power supply, suitable for daily listening, convenient to operate, hopefully not a toy that one becomes disinterested in after a week or so of use;
Sounds good
I've always liked the oddball look of this case that someone pretty good at 3D design designed for the bitx transceivers:
BitX40 Ergonomic Enclosure by fire5ign - Thingiverse.
So, I printed it up, in a really bright color to go with the fun tropical / Caribbean Sea motif that WRMI is using. That takes care of the case.
I have the Si5351 based raduino built and debugged - it just needs some minor tweaks to the code, and the 1602 LCD with an I2C drive board fits without difficulty in the case as is shown. That takes care of the VFO.
Two out of three already done, that just leaves the radio part. I've been thinking about the best way to go about it, and I think a single conversion superhet is the best solution to meet the design goals,
For the front end, I'm choosing a single tuned circuit, because I think the chances of correctly tuning up a double tuned circuit are nil, without some type of spectrum analyzer. Tentatively, I am going with a 50 ohm input. An active antenna, or maybe even a regenerated ferrite rod, might be attempted at some point.
I pulled a previously wound T-37-6 coil, and gave up on counting the turns wound on it, and just used an LC meter to measure its inductance at 3.54 uH, then calculated the capacitance necessary to resonate at 5050 Khz as 280 pF. I picked a 220 pF cap that measured out at 210 pF, and a 75 pF cap that measured 69 pF, to get the needed 280 pF. I used a 5 turn link on the toroid to go from a 50 ohm antenna to the 1500 ohm input resistance of an NE602, and assembled and swept the filter with the red pitaya.
And it peaked low, by about 30 pF of stray capacitance, when I back calculated the peak. I substituted a random 47 pF cap for the 75 pF, reswept the filter, and it looked close enough. An alternate, easier, approach is to substitute a trimmer cap for the smaller cap, and just peak it for maximum atmospheric noise or signal strength.
I made a simple mixer with an NE602, and tested it with an RF input of 0.01 volts at 5000 KHz, and an LO input of 0.1 volts at 5500 Khz and looked at it with the red pitaya to verify it worked. The 500 KHz IF, 5000Khz LO, and 10.5 MHz IF are plainly visible. The smaller peak at 11 MHz is probably an intermod artifact of the two IF's. The 0.01 volt RF input is very strong compared to an over the air signal.
All of this is being built on the board that I abandoned when I went a different direction with the DC receiver. Only the regulators were salvaged. The NE602 is using the 8 volt regulator. That's it for now. This is a low priority project, and updates / progress may be infrequent.
73.
Win W5JAG
Here is the LO leakage at the antenna port. Same test parameters as previous.
This is not the same, obviously, as LO leakage at the mixer RF port, but is important for legal compliance and to other users of the spectrum. It looks real good and is obviously legal. If an RF amp is needed, it won't be to suppress spurious emission at the antenna port:
Plugged into the 455 KHz input of the general coverage receiver.
Works. Listening to WRMI right now.
Pictured is an HP IAM81008, another variety of the gilbert cell double balanced mixer. This is a GHz class, low voltage, 50 ohm part, that requires very few external parts, a quality I always find appealing. It has an eye watering 15 dB noise figure, although that may be entirely adequate for 5 MHz operation.
It is an obsolete part, but still readily available at inexpensive prices on eBay and AliExpress, and, once in a while, from reputable surplus dealers at considerably higher prices. I have a handful that were purchased as surplus, that are genuine HP parts, one of which is installed as the RX mixer in my 14 MHz SSB transceiver. That mixer follows s a +28 dB RF preamp based on a BF992 dual gate MOSFET, is driven by a DDS LO, and has given excellent service.
The part pictured here is one of the dubious variety purchased from eBay, as I didn't want to waste a known good one on this project, and I figure anyone duplicating this is most likely to wind up with one of the questionable parts anyway, as they are easiest to come by. And quite cheap. Its markings do match my known good parts and the data sheet.
As installed here, applied voltage is 6 volts. In the following spectrum plots, the Si5351 LO drive is - 10 dBm at 5455 KHz. Drive at the RF port is about - 50 dBm (1000 uV) at 5000 Khz, a lot higher than it is likely to see in the real world, but as low as the signal generator .in the Red Pitaya will output, and I did not want to fool with putting an attenuator in line. I used 1 ohm resistors for the de - Q of the RF and LO low frequency capacitors, which are 10 nF. DC blocking capacitors are 10 nF.
This picture shows the Si5351 drive at the LO port:
And this picture shows the IF output port from 0 - 63 MHz ( the limits of the Red Pitaya):
This picture is the LO port and IF port overlaid, IF in yellow, LO in green:
This picture is the IF output from 0 - 12 MHz, where it is a bit easier to visualize the stuff of interest, note that the mixing products at 455 KHz and 10455 KHz are almost identical amplitude:
The predominant component is the LO leakage. The data sheet shows a typical LO leakage of -25 dB at the IF port. I'm not seeing that here - more like - 15 dB. I might experiment with the LO low frequency capacitor and see if I can improve this.
On the other hand, the data sheet gives a typical RF feedthrough at the IF port of - 25 dB, and this chip is showing better than - 30 dB.
I did not look at the LO leakage at the RF port. I guess I forgot.
IF output at 455 KHz looks to be about - 41 - 42 dBm, so based on an RF input of -50 dBm, there is about 8 - 9 dB conversion gain. The data sheet allows 6 - 10 dB, with 8.5 dB being typical. Conversion gain looks okay.
So, except for the LO leakage (a big except) this chip looks okay. It is my recollection that the known genuine part in my SSB transceiver beat the data sheet typical values at all ports.
Win W5JAG
I removed the BF1105, and turned what was left into a double tuned bandpass filter. I wanted to avoid a double tuned filter because of the tuning difficulty, but I think if both tuned circuits are peaked to maximum background noise on the same frequency, it will probably be okay.
The transformers are not changed: 34-35 turns (about one foot of magnet wire) on an Amidon T-37-6 (yellow) powdered iron core, with a 3 turn link for input / output. The inductance is approximately 3.5 uH. A 60 pF ceramic trimmer, in parallel with a 220 pF 0805 C0G smd capacitor completes the tuned circuit for 60 meters. For a 31 meter receiver (9455 Khz), 33 pF ish would probably be about right for the fixed capacitor.
I used a 10 pF ceramic disc to couple the transformers together. This gives a -3 dB bandwidth of about 225-250 KHz centered at 5000 KHz. The transformers look to be just a slight bit overcoupled. I left enough room between the transformers to put in a bipolar amp, but the liklihood is that if I decide to add an RF amp back in, I will probably place it between the filter output and mixer input.
The skirts are dramatically better than the previous single tuned filter. I don't know if that will make any difference in the real world.
Win W5JAG
I played around some with the BF1105, and have to conclude that in this particular application, it's just a dumpster fire.
There is no question the unwanted AM detection was coming from the BF1105 because, while it takes a couple of seconds for the microcontroller to boot and program the Si5351 for LO output, AM detection of WTWW or WBCQ - both huge signals at my QTH when they are on the air - occurs immediately upon power being applied to the BF1105 if they are on the air.
I tried providing an adjustable voltage to Gate 2 to vary ( turn down ) the gain, and this was of little benefit in practice. Any gain at all seemed to invite the unwanted detection, although at lower gain levels, WRMI sounded fine until QSB would set in, then the unwanted AM detection would just overwhelm the intended signal.
Unusable.
Which is regrettable, as it was quite a hot performer when the big signals were off the air. As above noted, in the morning just after local dawn, WWVH was plainly audible underneath WWV.
My only previous experience with BF1105 was in the development of the general coverage broadcast receiver, where, in a very early variant, it was used as a 45 MHz IF amplifier following a 2 pole crystal filter for initial selectivity. It worked well in that application, exhibiting none of the ill effects it demonstrated here, although I did not use it in the final variant of the receiver.
Win W5JAG
I tried the new front end out last night, BF1105 RF amp and HP IAM 81008 active mixer, and, wow, it was a hot mess of unwanted AM detection.
But, really sensitive - at dawn this morning, it was easily hearing both WWVH and WWV on 5000 KHz.
All of the unwanted detection is surely in the BF1105 - dual gate MOSFET's are notorious for it.
Back to the drawing board.
I've built a couple of prototypes, all of which were unsatisfactory to me, so I've decided to start over. Again.
Lack of overall gain was a motivating factor in abandoning the previous prototypes, so the next variant is starting out with an RF amplifier.
The active device is an NXP BF1105R, a UHF low voltage, high gain, low noise figure device especially made for RF amplifier use. All of the biasing resistors are in the die, making it a low parts count device and convenient to use. It is unusual in that it is an enhancement mode device, not that it matters much as a practical matter.
The circuit is essentially the data sheet figure 16, with single tuned circuits on each end. Vds is +6 vdc.
The tuned circuits are a couple of Amidon T-37-6 toroids, about 35 turns more or less, and approximately 3.5 uH. The parallel capacitance is about 300 pf, 60 pF of which is in the trimmer. Coupling in and out is by 3 turn links.
Edit: when peaked at 5000 KHz, gain is approximately 45 dB. In the picture channel 2 (green) is the amp input; channel 1 (yellow) is the amp output. This gain number looks really high compared to old school dual gate MOSFET's like 40673, but the data sheet says 38dB typical at 200 Mhz, and there is a step up in the input transformer, so the gain number for the middle of the 60 meter shortwave band looks plausible. I'm testing it with 0.001 volts. 0.01 volts shows clear overload artifacts. I can decrease the voltage on Gate 2 if I need to turn the gain down.
A new Raduino. Functionally, it's exactly the same as the old Raduino, just more visually acceptable to me.
In this variant, I used right angle pins on both sides of the Si5351 board and Pro Mini board to mount them vertically. This cuts the footprint in half in the horizontal dimension, and also creates a very rigid structure.
This arrangement would be well suited to rough duty portable or mobile service.
I swapped the LM317 for a 7805, but that's not of much consequence - just cut the parts count down some.
TDA7052A undergoing the Golden Earring "Moontan" stress test.
This chip is from a lot of 10 I bought from AliExpress priced at 10 / $1 USD.
I would ordinarily expect these to be counterfeit or sub standard in some way to the genuine article, of which I have examples for comparison.
This chip sounds to be functioning about like I would expect. It's running at 8 volts. I might give it a go at 12. The input attenuator seems to be working correctly. I don't hear any hiss or distortion.
Tentatively, it looks usable.
I'm trying to avoid using an LM386, but the really high gain of the '386 is useful for simple receivers.
Win W5JAG
This would interest me. Regenerated antennas seem to me to be the best, because when brought to critical threshold, the regenerative process can make a poor antenna with high losses into a low-loss structure that is very sensitive to EM field disturbances at the regenerated resonant frequency. It's of course possible to build a non-resonant, broadband, active loop antenna, but that requires a bit more care in constructing a low-noise, broadband loop amplifier. A regenerative loop antenna is less critical in construction, but more critical in adjustment.
I'm having a lot of fun with a small ferrite rod antenna in my shortwave superhet, and recently started regenerating it. I'm now thinking that with regeneration in the front-end RF tank coil, even the ferrite rod might not be necessary. A simple air-core coil, even with a small diameter like 1 cm or 1.5 cm, should still be able to pick up signals even without the assistance of the ferrite rod. I recently saw this YouTube video that seems to show a regen receiving signals strongly when using only a quite small (about 1.5 cm diameter) air-core coil, which is around the same size as my ferrite rod antenna. As far as I can see, the regen in that video uses no external antenna.
Hmm, on closer inspection, the red alligator clip in the videos might be an external wire antenna connected to the coil. But the principle still stands: applying regeneration to the antenna circuit (whether it's a small air-core coil, an ferrite-cored solenoidal coil, or an LC tank with a wire antenna connected to it) improves the performance of the antenna.