Last edited 01/04/2003.
Direct Conversion Receivers
A direct conversion receiver is the simplest of all receiver types. In essence it is a mixer an oscillator and an audio amplifier. The oscillator runs at the frequency that you want to receive and the mixer mixes this to straight down to the audio band.
As an example, lets assume that there is a cw signal on 7030KHz. We set our oscillator to 7031KHz and the difference is 1KHz (which comes out of the mixer after mixing the oscillator with the signal). This is the tone that comes out of the audio stage simple! But heres the rub. If we set the oscillator to 7029KHz, the difference is still 1KHz and so a 1KHz tone appears. So whats the problem? Well, the problem is that on a crowded band, all the signals appear twice (either side of their centre frequency). This has two effects; the first is that the noise figure of the receiver is degraded by 3dB, however on HF that wont really bother us. The second is that the band will be twice as full as it should be that is to say the likelihood of suffering from interference is worse than would happen with a standard superhet receiver.
In the superhet receiver, the signals are mixed to a fixed intermediate frequency, before being mixed down to audio. At the intermediate frequency (perhaps 9MHz, 10.7MHz or 455KHz) there is a filter. This filters out one of the sets of tones so each signal appears just once as you tune around the band hey presto! Its a better receiver.
Well yes, in some ways a superhet is a better receiver, but there are some downsides to this. One penalty is cost. A decent 9MHz filter could end up costing quite a bit (perhaps $100). This method is also rather complex as more stages and more components are usually required to implement the design.
So direct conversion receivers have survived. But they havent stood still and there are ways of getting that elusive single signal reception with DC receivers. The most common way is well its in all of the textbooks.
Now, as you can see, to build this sort of receiver, you need an audio phase shift network that has a pretty constant phase shift over the entire audio band - for an SSB receiver thats 300-3000Hz, which is a complete decade. Remarkably, there are such networks. The one that is generally used is called a polyphase network. It relies on using a whole load of accurate resistors and capacitors to achieve the constant phase shift. This has its own problems. Firstly the network is physically large and secondly, it is not possible to adjust it. Once you have built it, its performance is determined and thats it.
I started thinking about DC receivers and single signal reception. Perhaps there is an easier way? Its often said of DC receivers that they have a zero frequency IF i.e. the IF frequency is the result of the mixing process. This sounds logical. However, I wondered what might be achieved at the signal frequency making that the IF instead. Sounds daft? Well, it all depends upon what you want to do. I was thinking about building a crystal-controlled rig for my Adventure Radio activities. The idea was that is should be simple and that it would be used to keep a sked. Hence the advantages of crystal control. Now if you are on a single frequency, there is the possibility of using a filter at the signal frequency. So here is the plan. Buy say 5 x 7030KHz crystals, use one for the oscillator and used the rest in a signal frequency filter. Bingo! Its a simple single signal receiver.
OK so youre not impressed. You want to tune over the entire CW band dont you? I thought so. Heres my second solution. Use two stages of mixing but I put in two RF phase shifters to get your single signal performance. Heres how it works.
The signal comes in from the aerial and is fed into a phase shift network that gives outputs of 0 degrees and 90 degrees. These outputs are fed into two mixers. The local oscillator is also fed into a phase shift network giving 0 and 90 degrees shifts. These outputs feed the oscillators. The output from the oscillators is fed to a balancing potentiometer and then through to the AF stage. By switching the phases correctly you can get USB or LSB. I would have drawn this but I have been unable to find a free schematic package on the web. Can you help me?
Lets assume that the RF phase shifters are only accurate at a single frequency this is not a bad assumption. How good will the sideband rejection be? If I assume that they will vary linearly with frequency its easy to work out.
If they are set up for a 90 degree shift at 7015KHz, moving to 7000 and 7035 represents a phase shift of 0.8 and 1.8 degrees.
The chart below shows the resulting sideband rejection. At worst it is 30dB - not too bad. Now if I was rather smarter, I think that it would be possible to do the mixing so that the phase errors in one network cancelled out the errors in the other (by mixing in reverse). This is one to think about while going to sleep.
Building the Concept Model
All talk and no action is no good so I'm going to try the concept out. I suspect that getting two 90 degree phase shift networks adjusted by ear will be hard so during alingnment I will replace one with a quarter-wavelength of co-axial cable and adjust the other. I will use NE612 mixers and a bipolar oscillator.