so i've been messing around with a direct conversion receiver for 40m over the past couple months, nothing fancy just a NE602 based design i cobbled together from a few different schematics i found online plus some stuff from EMRFD. got it mostly working now and figured id share some observations since i didnt find a lot of practical info when i was troubleshooting.

the audio chain was the biggest headache honestly. i was getting this nasty hum that i couldnt track down for like two weeks, tried everything - different bypass caps, ferrite beads on the power leads, even built a separate regulated supply thinking maybe the bench supply was the culprit. turns out it was my audio amp board sitting too close to the NE602 oscillator and picking up interference through the ground plane. moved it about 3 inches and put a little tin shield between them and it cleaned right up.

sensitivity is pretty decent for a simple design, im hearing stations that my commercial rig picks up no problem. the main limitation is selectivity obviously since its just a lowpass filter after the mixer, SSB sounds kind of wide and mushy but its totally usable. CW is actually really nice with a simple audio filter i tacked on. VFO stability took a while to sort out too - i ended up using an MPSW01 for the oscillator transistor and a good NPO cap for the main tuning cap and that helped a lot compared to my first attempt which wandered all over the place when the room temperature changed.

anyway if anyone else is building one of these and running into issues happy to compare notes. im thinking about doing a 20m version next but i know the image rejection gets more challenging up there with a direct conversion design.

the NE602 is such a great little chip for this kind of thing, i built basically the same setup years ago from a QST article and ran into almost identical problems with the audio stage. one thing that helped me was putting the whole audio section in its own shielded box, like just a small tin enclosure, made a huge difference. also if you havent already try winding a common mode choke on the audio leads going between the rf and audio boards, just a few turns through a small ferrite, that can help too.

on the 20m idea - yeah image rejection is the issue. you could add a bandpass filter on the front end, doesnt have to be elaborate, even a simple double tuned circuit helps a lot. or just accept that direct conversion at 20m means you'll hear some images and work around it. depends what you want out of the project really.

the VFO stability thing is real, i spent way too long chasing that on a similar build. what helped me was just thermally insulating the whole oscillator section with a chunk of foam, sounds dumb but it keeps the temperature from swinging as fast and the drift becomes way more manageable. also are you using any AGC or is it just straight audio gain? i found without AGC strong local signals kind of saturate everything and it gets messy on a busy band like 40.