Re: Taylor detector
Hi, It's not clear to me why you would want to do this. As far as I can see, the purpose of a Tayloe detector is to mix a real-valued RF signal with a complex-valued local oscillator to arrive at an intermediate frequency at or near zero frequency that can be sampled using a sound card. If you're using GR the implication is that the signal has already been sampled. An ideal hardware IQ mixer would be a pair of analogue multipliers with sine local oscillators. However, this is not practical if good linearity is required (linearity is desirable to avoid intermodulation distortion and other nasty things). This is why switches are used. The switches used to be diodes, and still are at higher frequencies, but at HF CMOS analogue switches work better. With a switch based mixer you are effectively multiplying by a square wave, which comes with the disadvantage that odd harmonics of the square local oscillator also convert signals and noise into the IF. This necessitates a filter in front of the mixer to suppress these spurious responses. When IQ mixing is required there are the added challenges of phase and amplitude balance. Sometimes things ain't pretty in the analogue world. As has already been pointed out, in GR a complex multiplier is the usual way to go. An accurate multiplier can be fed with near perfect numerically generated sine waves with perfect amplitude and phase balance. Besides, even if you wanted to simulate a square wave local oscillator in DSP you'd have to approximate by only including the harmonics up to the Nyquist frequency. On 13/8/22 03:08, david vanhorn wrote: Ive been wrestling with this for a while, and im not even seeing how to get started implementing a Taylor detector in gr. Is it even possible?
Re: Taylor detector
Because in GR I can generate impaired signals approximating real world conditions, with fine control and repeatability. I can then run that through the Tayloe while experimenting with it's variables, again with fine control and repeatability. Finally I can work with DSP on the output of the Tayloe, observing the total system response. On Sat, Aug 13, 2022 at 4:38 AM Steve Hubbard wrote: > Hi, > > It's not clear to me why you would want to do this. As far as I can see, > the purpose of a Tayloe detector is to mix a real-valued RF signal with > a complex-valued local oscillator to arrive at an intermediate frequency > at or near zero frequency that can be sampled using a sound card. If > you're using GR the implication is that the signal has already been > sampled. > > An ideal hardware IQ mixer would be a pair of analogue multipliers with > sine local oscillators. However, this is not practical if good linearity > is required (linearity is desirable to avoid intermodulation distortion > and other nasty things). This is why switches are used. The switches > used to be diodes, and still are at higher frequencies, but at HF CMOS > analogue switches work better. With a switch based mixer you are > effectively multiplying by a square wave, which comes with the > disadvantage that odd harmonics of the square local oscillator also > convert signals and noise into the IF. This necessitates a filter in > front of the mixer to suppress these spurious responses. When IQ mixing > is required there are the added challenges of phase and amplitude > balance. Sometimes things ain't pretty in the analogue world. > > As has already been pointed out, in GR a complex multiplier is the usual > way to go. An accurate multiplier can be fed with near perfect > numerically generated sine waves with perfect amplitude and phase > balance. Besides, even if you wanted to simulate a square wave local > oscillator in DSP you'd have to approximate by only including the > harmonics up to the Nyquist frequency. > > On 13/8/22 03:08, david vanhorn wrote: > > Ive been wrestling with this for a while, and im not even seeing how > > to get started implementing a Taylor detector in gr. > > > > Is it even possible? > > -- K1FZY (WA4TPW) SK 9/29/37-4/13/15
Re: Taylor detector
It's possible to simulate a Tayloe detector in gnuradio. Probably this would involve stream demux block, filters, negate and sum. Ie. there's no "capacitor" block. You will need to think in terms of functions. On Sat, Aug 13, 2022, 20:57 david vanhorn wrote: > Because in GR I can generate impaired signals approximating real world > conditions, with fine control and repeatability. > I can then run that through the Tayloe while experimenting with it's > variables, again with fine control and repeatability. > Finally I can work with DSP on the output of the Tayloe, observing the > total system response. > > > > On Sat, Aug 13, 2022 at 4:38 AM Steve Hubbard > wrote: > >> Hi, >> >> It's not clear to me why you would want to do this. As far as I can see, >> the purpose of a Tayloe detector is to mix a real-valued RF signal with >> a complex-valued local oscillator to arrive at an intermediate frequency >> at or near zero frequency that can be sampled using a sound card. If >> you're using GR the implication is that the signal has already been >> sampled. >> >> An ideal hardware IQ mixer would be a pair of analogue multipliers with >> sine local oscillators. However, this is not practical if good linearity >> is required (linearity is desirable to avoid intermodulation distortion >> and other nasty things). This is why switches are used. The switches >> used to be diodes, and still are at higher frequencies, but at HF CMOS >> analogue switches work better. With a switch based mixer you are >> effectively multiplying by a square wave, which comes with the >> disadvantage that odd harmonics of the square local oscillator also >> convert signals and noise into the IF. This necessitates a filter in >> front of the mixer to suppress these spurious responses. When IQ mixing >> is required there are the added challenges of phase and amplitude >> balance. Sometimes things ain't pretty in the analogue world. >> >> As has already been pointed out, in GR a complex multiplier is the usual >> way to go. An accurate multiplier can be fed with near perfect >> numerically generated sine waves with perfect amplitude and phase >> balance. Besides, even if you wanted to simulate a square wave local >> oscillator in DSP you'd have to approximate by only including the >> harmonics up to the Nyquist frequency. >> >> On 13/8/22 03:08, david vanhorn wrote: >> > Ive been wrestling with this for a while, and im not even seeing how >> > to get started implementing a Taylor detector in gr. >> > >> > Is it even possible? >> >> > > -- > K1FZY (WA4TPW) SK 9/29/37-4/13/15 >
