Message	Date	From	Subject
6465	2011-07-19 17:44:13	KK7B	Alternative Receiver Architectures
The Single-Signal discussion has generated a lot of interesting points, but no one has yet referred to figure 8.23 in EMRFD, or the accompanying text. There is more than one way to obtain selectivity--it can be in your head (for example IQ binaural technique), electronically at the IF using filters, with a phasing architecture that accomplishes the same mathematical operations, electromagnetically in space (shown in the left half of figure 8.23), through clever use of digital signal processing techniques (right half of figure 8.23), or even selectivity in time--operating when the band is wide open but not many stations are on. Here's another way you could build a single signal receiver. It is relatively easy with baseband DSP to throw a notch at signals you don't want to hear. You can tell which side of zero-beat a CW signal in on by the change in pitch as you tune--are you tuning toward it or away from it? A DSP algorithm could keep track of the rotation direction of the tuning knob and the pitch changes of all the signals in the passband, and null all the ones that tune the wrong way. That would sound like single-signal selectivity, but the actual receiver would have a simple DSB mixer. It wouldn't sound like a microR2, but I hope that it might sound better than my Kenwood TS-570 with "Noise Reduction 2" enabled. My friend Bill Carver mentioned the 3dB noise penalty of a DSB system. 3 dB is an engineering number that we can predict mathematically and measure in the lab. But as soon as we connect an antenna, all bets are off. The external noise floor on all of the amateur bands up through at least 50 MHz is "lumpy" on a scale of around 1 kHz. A narrow receiver can park in a quiet spot, while a wider receiver response includes more of the noisy peaks. Before we call CQ, we tune around for a clear spot. Such spots are rare between 7.03 and 7.04 MHz--twice as rare with a receiver that receives both sidebands. So in practice the penalty is often much worse than 3dB. But simple microwave gear may have more noise contributions downstream from the mixer than the external noise floor, and experience much less than 3 dB degradation. When the noise floor is the sum of a large number of impulses (that may include distant thunderstorms), wider bandwidths might be better. Note my careful use of the terms "may" and "might." Generations of hams have made use of single diode ring mixers with no RF gain in simple 40m CW rigs--Roy Lewallen's Optimized QRP Transceiver and the Haywards's Ugly Weekender transmitter and receiver are two good examples. Those receivers work exceptionally well on quiet bands out here in the West with low ionospheric noise and low population density. I've tried AB listening tests to prove that my phasing receivers are better--but I have yet to encounter a signal where the theoretical 3 dB improvement makes a difference in copy. So I expect something else is going on, perhaps the different way the two receivers react to broadband impulse noise. But when I've included a front-panel switch to select either both sidebands or just one--after the comparison tests are done, the switch always ends up in the single-signal position. Bill also mentioned his 100 dB opposite sideband suppression quirk. I know that one well, but I got over it, hi. I prefer about 50 dB opposite sideband suppression in the receiver with another 30 dB from a smart antenna system. But that's just me. It's our personal quirks that make these problems worth pursuing. My personal quirks keep moving me in the direction of systems that do only one thing, exceptionally well. That includes choosing not just the band and mode, but the operating location, antenna I'm going to use, battery drain, weight, durability, and how uncomfortable I'm going to be while trying to make the contact. Those little rigs are more like formula one cars (or perhaps go-karts) than something big, comfortable, and air conditioned that will tow a 4 ton boat at 70 miles per hour up any interstate highway in the US--and deliver the kids to kindergarten too. Once you start thinking about the specific receive problem you want to solve, there are a number of interesting ways to go about it. The mix of electrical, mechanical, RF, analog, DSP, vintage, hardware, software, brainware...will be different for each of us. If you can have only one vehicle and you have to drive kids around and tow a big boat, your options are limited. But we can have more than one receiver. Keep sharing those ideas. This stuff is cool. Best Regards to the group, Rick KK7B
6466	2011-07-19 20:19:31	qrp.gaijin	Re: Alternative Receiver Architectures

6468	2011-07-20 05:51:11	Tim	Re: Alternative Receiver Architectures
You mention the pattern recogniti

The Single-Signal discussion has generated a lot of interesting points, but no one has yet referred to figure 8.23 in EMRFD, or the accompanying text.

There is more than one way to obtain selectivity--it can be in your head (for example IQ binaural technique), electronically at the IF using filters, with a phasing architecture that accomplishes the same mathematical operations, electromagnetically in space (shown in the left half of figure 8.23), through clever use of digital signal processing techniques (right half of figure 8.23), or even selectivity in time--operating when the band is wide open but not many stations are on.

Here's another way you could build a single signal receiver. It is relatively easy with baseband DSP to throw a notch at signals you don't want to hear. You can tell which side of zero-beat a CW signal in on by the change in pitch as you tune--are you tuning toward it or away from it? A DSP algorithm could keep track of the rotation direction of the tuning knob and the pitch changes of all the signals in the passband, and null all the ones that tune the wrong way. That would sound like single-signal selectivity, but the actual receiver would have a simple DSB mixer. It wouldn't sound like a microR2, but I hope that it might sound better than my Kenwood TS-570 with "Noise Reduction 2" enabled.

My friend Bill Carver mentioned the 3dB noise penalty of a DSB system. 3 dB is an engineering number that we can predict mathematically and measure in the lab. But as soon as we connect an antenna, all bets are off. The external noise floor on all of the amateur bands up through at least 50 MHz is "lumpy" on a scale of around 1 kHz. A narrow receiver can park in a quiet spot, while a wider receiver response includes more of the noisy peaks. Before we call CQ, we tune around for a clear spot. Such spots are rare between 7.03 and 7.04 MHz--twice as rare with a receiver that receives both sidebands. So in practice the penalty is often much worse than 3dB. But simple microwave gear may have more noise contributions downstream from the mixer than the external noise floor, and experience much less than 3 dB degradation.

When the noise floor is the sum of a large number of impulses (that may include distant thunderstorms), wider bandwidths might be better. Note my careful use of the terms "may" and "might." Generations of hams have made use of single diode ring mixers with no RF gain in simple 40m CW rigs--Roy Lewallen's Optimized QRP Transceiver and the Haywards's Ugly Weekender transmitter and receiver are two good examples. Those receivers work exceptionally well on quiet bands out here in the West with low ionospheric noise and low population density. I've tried AB listening tests to prove that my phasing receivers are better--but I have yet to encounter a signal where the theoretical 3 dB improvement makes a difference in copy. So I expect something else is going on, perhaps the different way the two receivers react to broadband impulse noise. But when I've included a front-panel switch to select either both sidebands or just one--after the comparison tests are done, the switch always ends up in the single-signal position.

Bill also mentioned his 100 dB opposite sideband suppression quirk. I know that one well, but I got over it, hi. I prefer about 50 dB opposite sideband suppression in the receiver with another 30 dB from a smart antenna system. But that's just me. It's our personal quirks that make these problems worth pursuing.

My personal quirks keep moving me in the direction of systems that do only one thing, exceptionally well. That includes choosing not just the band and mode, but the operating location, antenna I'm going to use, battery drain, weight, durability, and how uncomfortable I'm going to be while trying to make the contact. Those little rigs are more like formula one cars (or perhaps go-karts) than something big, comfortable, and air conditioned that will tow a 4 ton boat at 70 miles per hour up any interstate highway in the US--and deliver the kids to kindergarten too.

Once you start thinking about the specific receive problem you want to solve, there are a number of interesting ways to go about it. The mix of electrical, mechanical, RF, analog, DSP, vintage, hardware, software, brainware...will be different for each of us. If you can have only one vehicle and you have to drive kids around and tow a big boat, your options are limited. But we can have more than one receiver. Keep sharing those ideas. This stuff is cool.

Best Regards to the group,

Rick KK7B