2008-01-27
30 Metre Autodyne Receiver
As mentioned in the Australia Day article, building this receiver was the result of QRP-L discussions of WWV receivers. The design comes from Nick Kennedy's page, and itself seems to be the result of QRP-Tech discussions. I only discovered QRP-Tech last week, and just subscribed, looks better than QRP-L which has a somewhat worse signal to noise ratio if you interests are more technical.
The receiver worked fairly well breadboarded, but did suffer from tunable-hum. This isn't surprising, as unshielded it is essentially a direct conversion receiver and the LO was illuminating all kinds of mains powered devices in close proximity. In particular my soldering iron which seems very efficient and hum modulating the near-field of coherent detectors. The breadboarded prototype wasn't (easily) tunable, but would lock my noisy BK (im)Precision generator/sweeper and hold it indefinitely with sufficient signal.
What does this look like in the time domain? When locked the LO perfectly phase-tracks the RF signal. When unlocked you see hetrodyne from the frequency difference of the LO and RF. Triggered from the RF signal you see something like this:
And unlocked, this:
I had a much better dual signal trace with the LO and RF triggered from the RF in several configurations, but I didn't take pictures, sorry.
Akshay Parelkar VA7AAX shot me an email about the bread boarded version shortly after I posted it, and inspired me to build it properly.
With the toroids waxed to the board and a trimmer to bandset the receiver more practical tests could be made. The varicap (1N4007 rectifier) gives a good range for fine tuning and it is quite easy to lock moderately strong shortwave signals with just an alligator clip lead for the antenna!
WWVH is far too weak here to lock with such a primitive antenna, or even really hear above the garbage coming from my computers. However the 31 metre broadcast band gave me plenty of strong signals to experiment with. Radio Australia gave three closely spaced, rock solid signals. I could also hear a few others, but I didn't catch their call signs. Quite surprisingly I could briefly lock them too as their signals came up.
Apparently my counter's frequency reference disagrees with whatever Radio Australia is using. I assumed they have a precision clock. All three Radio Australia signals seemed to disagree by the same amount, as did the others I could briefly lock, so I took the leap and tweaked the calibration of my counter to reduce the error.
The lash-up held this to within 200 mHz for over an hour. Pretty good considering my counter has only a fairly cheap reference. Now and then the signal would fade and the receiver would break lock. The hetrodyne was immediately obvious and a tweak of the tuning pot would relock it, or the signal strength coming back up would also drag the LO back into phase lock.
Videos
There was something about the Australian Open on Radio Australia at the time I was experimenting with the circuit. Here is a video of a rocking the receiver across lock with the strongest copy. I also stick my finger near the tuned circuit which pulls it off lock temporarily. You'll note how it captures the carrier of the station, a normal direct conversion receiver would have all kind of beating and phasing effects as you approached zero-beat.
Radio Australia was transmitting three signals separated by only 10 kHz, here we have a video of tuning across all three. It displays the heterodyne heard when the receiver is unlocked, and the locking effect once you are close enough to the carrier. Observe the frequency counter settling into each of the frequencies once lock is achieved.
Problems
One minor problem with the receiver is VHF interference. The 270 pF capacitor that feeds back the LO tank to the transistor base appears to be resonant in the FM broadcast band. At one particular setting of the bandset trimmer 2MIX FM 106.5 MHz can be heard! Shielding the unit inside a box would fix this (and any final traces of direct-conversion LO radiation hell). It seems to be largely direct pick-up by the circuit, although putting a lashed-up low-pass filter into the antenna line did reduce the problem a bit.
I mentioned to Akshay that I thought the receiver needed a real front-end. A 30 metre bandpass is advised. I can't imagine what would happen with a real antenna rather than just a foot-long clip lead!
Better layout might be advised, reducing the length of all the HF circuit paths and pushing any possible spur resonances above the transistor bandwidth. I tried beads on the transistor base, but it killed the oscillation as well. Experimentation with a small amount of additional Miller capacitance may tame any VHF oscillations.
The VHF interference does suggest a similar circuit topology might work quite well as an FM radio autodyne front-end. Even fairly modest modern transistors (I used 2N3904s) have huge bandwidth in the differential configuration. I'd imagine it would be possible to have one side a tuned LO resonator and the other an IF one, with the cascoded transistor in the tail as the input buffer.
Notes
This autodyne receiver implements a coherent AM demodulator, which has some advantages to conventional envelope detection. It lacks AGC, which is annoying for general shortwave reception, but otherwise it makes a pretty good general purpose receiver. It can receive AM, CW and SSB with one only adjustment (tuning), unlike a regenerative set. Strong CW signals can grab the LO though, and pull it, right into lock in some cases, eliminating the beat note! Also, strong adjacent signals will capture the LO pulling it into around, which can be extremely annoying. If there is sufficient noise margin just attenuating the signal will help.
I shunted the AF amplifier feedback resistor with a 1 nF capacitor to roll off the HF response. The audio was otherwise a bit sharp. You might like to add a pair of cascaded Sallen-Key filters to roll it off more sharply and improve the selectivity. I wouldn't recommend the receiver for CW, but you might also like to add a 700 Hz bandpass for that purpose if you intend to use it. Not too tight though, its a bit chirpy with strong adjacent signals.
I used the same turns ratio at the input transformer, but wound 3:12 instead. This isn't too critical really, and I am yet to measure the input impedance or attempt to predict it, I'd imagine it will be moderately high though.
Shield your counter interconnection, LO leakage here will cause the all too familiar direct conversion tunable hum problems. The buffering is generally sufficient, but I can hear the count windowing of one of my counters back through the AF output which suggests you might like to improve it. A common-base isolator or padding will probably help. I picked off the LO at the tap point on the tank rather than its hot end, this reduced the effect of the load compared to the original design.
8 comments.
Updates
- 2008-02-04: Autodyne Receiver Now In a Box
- I add a front-end and a box to the 30 Metre Autodyne Receiver