I've simulated the Tayloe integrator in LTSpice - sort of. I simulated the circuit in "Part-I of SDR for the Masses" using Helmut's 74HC library for the logic chips with the voltage controlled switch devices forming the heart of the FET multiplexer. The switching stuff works ok, but the simulation is so slow, and has to run so long in order to see something at baseband; it's just not practical.
I'm going to give it another try, this time I'll eliminate the 74HC devices as they may be causing a lot of the slow-down. I'll replace them with voltage pulse sources at the switch control inputs. I might also do with some deeper optimization as well. I'm simulating in WinXP SP2+ on a 3.2GHz Prescott with 2GB of RAM. Regards, David --- In [email protected], "Daniel Jackson" <[EMAIL PROTECTED]> wrote: > > I was thinking today of the Taylor circuit also. As Cecil says, > sometimes the simulations can run slow. But if you got a 2.4 GHz > processor it might not take too long. My processor is not that fast. > > I am seeing that simulation times from 4.5u to 200u appear to show me > most of what I need to see. > > Also, I can see that the idea of a switching mode mixer seems to be > better in broadband HF use than the analog biased circuit is. When > there is no agc control voltage to control the gain. > > The switching mixer seems to have a set level of input that does not > vary in input amplitude over the range of the spectrum and hence does > not result in a high level input on low frequencies and low levels on > higher. Which imbalance analog mixers if there is no agc used. > > So circuit ideas such as I am trying have to be moved from the analog > to switcher versions for completing the picture I think. > > We all should try to simulate the Tayloe circuit as an educational thing. > > Dan > > --- In [email protected], k5nwa <k5nwa@> wrote: > > > > At 08:24 AM 2/28/2008, you wrote: > > > > >Hi Dan, > > > > > >I would like to simulate a Tayloe mixer using a quad bi-lateral > > >switch, e.g., 74XX4066 and/or 74CBT3253 etc. I have a working > > >schematic in LT, but as I suspected, the simulation would take years > > >to provide reasonable data at the I-Q output, it's just too slow given > > >the input vs. output frequencies. > > > > > >Anyway, I'd like to discuss this with you off-Forum, May I? > > > > > >Thanks, David > > > > > > > I use LTspice and it just takes a few seconds to simulate through > > several cycles of output. > > > > On the simulation is the clock and Quadrature generating circuitry, > > the switches, and the output op-amps, basically an equivalent to a > > SoftRock and it just takes a few seconds on my PC to run through a > > simulation of 200 uS. > > > > For the switches I use a sub-circuit to emulate the properties of the > > FT switches which are shown below, so it doesn't look like the actual > > switch but it behaves like the actual switch; > > > > 1. Has a finite resistance > > 2. Make before break > > 3. The switch turns 90% on in the first ns, and fully on in a > > exponential rate for the turn on time. > > 4. The switch turns 90% off in the first ns and its exponential for > > the turn off time until it hits 100% > > > > It's main element is a voltage controlled resistor, with RC circuits > > used to shape the waveform fed to the resistor so it has the proper > > turn on rates. That is a generation 2 switch, the latest is > > generation 4 but I seem to have lost it along with version 3, it > > physically implements a real switch with all it's warts, somewhere I > > have to have a copy of it but I have no clue where. > > > > > > Cecil > > K5NWA > > www.softrockradio.org www.qrpradio.com > > > > "Blessed are the cracked, for they shall let in the light." > > >
