While researching homebrew DC receivers for my “Polyakov” style receiver board, I stumbled across Dave AA7EE‘s various projects which combined the NE602 Gilbert-Cell mixer IC and some form of IC audio amplification to make simple, direct-conversion receviers. In particular, the Rugster receiver (named after a dear and recently departed pet), which itself seems to take inspiration from a number of other receivers, including the GQRP Sudden. VK2AWC has also published some ideas along similar lines, as I’m sure have others.
As of yesterday, 1/14/16, I’ve heard the very first (faint!) signals coming out of the headphones, so I think the project deserves the start of a write-up. I think it’ll make the most sense to show the schematic as it currently exists and discuss it afterward. So, here’s the circuit, as currently built:
Mixer/demodulator: The heart of this receiver is an NE602 doubly-balanced mixer. It has a an input impedance of ~1.5kΩ, but for a receiver I think this will be less troublesome than it would be for trying to interface with other circuits. (But see Front end for other thoughts.) The mixer is accepting a single-ended, untuned input on pin 1, while the other RF input is simply AC coupled to ground through a couple caps.
Oscillator: Currently, the receiver is rock-bound to whatever frequency of crystal is connected to pin 6 of the NE602. I installed a bit of 0.1″ header in this position to make it easy to swap crystals in and out for experimentation. Down the road, I think I’ll need to swap this for a bit of machine-pin header for more mechanical security.
The crystal and a couple of supporting capacitors are the entirety of the external oscillator parts. I’ll admit I grabbed whatever capacitors were available in my kit – the NE602 data sheet shows a 10pF cap between pins 6 and 7 and a 22pF cap in parallel with the crystal, so I just got close enough. The crystal seems to resonate just fine, but I’ll admit I don’t know what I’m doing I this area. Likely, I’m pulling the crystals off frequency; I’ve got a frequency counter on its way from eBay, I’ll have to check this out when the counter arrives,
I mean to experiment with some deliberate frequency-pulling once I get the rest of the receiver working a little better.
Front-End: A simple 40m pre-selector filter sits right behind the antenna jack – the filter values are taken straight from the GQRP Sudden 40 and the Rugster. I formed my 5.3μH inductors as 33 turns on a T50-2 toroid.
AA7EE has an impedance transformer in his front end, with a 1:9 turns ratio, presumably to help a 50-ohm antenna to the ~1.5kΩ input impedance on the NE602.? Or perhaps just to present a higher voltage level to the mixer. VK2AWC has a 4t:38t transformer in the same spot. I may need to try this as well.
Audio Amp: The audio amplification section is a simple LM386 run in a medium-to-high gain configuration. The 10μF cap across pins 1 and 8 should lead to a gain of around 200, according to the datasheet. To this point, I’ve neglected to bypass pin 7 to ground (the datasheet suggests 0.1μF), and I think that must be necessary for a high gain-figure, since the circuit turned into a squealer when I first assembled it!
Replacing the 10μF cap with a 4.7μF cap and a 1kΩ resistor helped this problem, but I think if I can improve the stability and get that lost gain back, so much the better.
The LM386 datasheet shows a 250μF cap between the IC output and a speaker; AA7EE and the GQRP use a 100μF cap here. I’m using a 220μF electrolytic since I had one on hand.
Power: For testing, the rig is being powered by a little 9V battery, but I think may be a factor in limiting the audio output (see Known Issues below). Given that the NE602 is being powered by a 7808 regulator and the LM386-3 will take any supply voltage between 5V and 12V, I should have some flexibility in choosing a power supply later on.
I didn’t originally have a reverse-current protection diode (1n4001) nor a10μF decoupling cap in place across the power input, but in future, I think they’re a must.
Here’s an image of the circuit as currently built:
It’s not pretty, but it sure is small! I know copper clad would have provided a better ground plane, but I got impatient waiting for my shipment of copper clad to arrive from eBay.
So, with power applied and a pair of basic Apple earbuds in place, I can pick out some faint (to me) signals around 7.023! Very cool! Well, at least I could when I turned off the LED desk lamp, fluorescent work lights, and cordless-drill battery charger in my office/shack, which were all QRM-firehoses. I’ve now run these all to a single power strip, for easy on-off operation.
Nothing’s perfect, here’s my short list of known issues:
- With a 10μF cap between pins 1 and 8 of the LM386, the unit squeals like a banshee whenever the 10k pot is advanced past about 15%. This was improved by better bypassing right at the IC’s Vs and Gnd pins, and made better again by replacing the 10μF cap with a 4.7μF cap and a 1kΩ resistor. Still not quite sure what was happening, or why this oscillation depended on the position of the input pot.
- Update 1/25: see my later post for a solution to the squealing. Turned out, used 9V batteries are not a great power source.
- For some reason, when I advance the 10k audio level pot past about 25%, the audio cuts out entirely – no squealing, no AC hum, nothing.
- There is no front-end tuning at all a little more verification of the filter characteristics would probably help.
Some possible future improvements, besides addressing some on the issues above:
- Better output overall – the audio output is quiet at best.
- VK2AWC had some interesting ideas about the combination of NE602 and LM386 – notably feeding back a little of the output to pin 7 of the LM386 to provide higher gain, and some passive RC networks between the NE602 and the LM386 to provide some shaping to the audio, that I think will be worth checking out.
- Mounting the project in a nice, metal enclosure.
- Providing an interface for an external VFO. (Possibly the Si5351-driven VFO from an earlier project.
Problems are just un-implemented features.