Gestalt. By use of low-side injection, the inverting characteristic of the wide-range exponential generator can be used directly to produce the audio frequency span. This circuit configuration removes a second op amp performing offset control voltage inversion from the control voltage input loop. Instead, one inverting summer directly uses the frequency inverting characteristic to map control voltage to exponentially-weighted frequency.
The inversion involved does mean that higher pitches are controlled by relatively lower amounts of output current from the exponential generator. This must be stable for excellent frequency accuracy. Stability should be the case though, because the LO has a compressed frequency range of < 5:1. More particularly, the most musically useful frequencies (< 5 KHz) are used under conditions of higher current from the exponential generator.
In this application of frequency synthesis, a “pure” RF is mixed with square-wave LO to produce clean IF signals: the pitched audio range signals we want. There cannot be an image frequency, because the RF has very low harmonics, and none image to the IF band.
Additionally, plural “channels”of like-oriented low-side injected down-conversion can be done relatively inexpensively, affording a programmable frequency spectrum, esp. where one exponential generator is used to make multiple frequencies, each frequency being a fraction of the output current. In the case of the inverted frequency response, these fractions map to higher harmonics. There is even an option of restricting the base frequency range to ≈︎10 KHz, leaving lowest-current output solely for harmonic generation.
This system is also markedly easier to tune, because full-scale output from the exponential generator, where raw CV=0V represents DC, and LO=RF. So, these two frequencies are tuned to beat very slowly. Then, at full-scale CV input (CV=10V, becoming -10V after summing inversion), the LO is tuned to Fmin, to match the highest pitch out of the IF. The V/OCT scale control does not interact with the first adjustment, and interacts only moderately with the second adjustment.
Frequency Plan using 32.768 KHz RF REF LO
RF LO IF=RF-LO
32.768 KHz 32.768 KHz 0 KHz
32.768 KHz 27.768 KHz 5 KHz
32.768 KHz 22.768 KHz 10 KHz
Limiting LO frequency output for a more musically useful frequency range of 0-10 KHz makes a lot of sense. Neither RF or LO oscillators, nor upper modulation products are in audio range. This limits the effects of mixer feed-through. More substantially, frequency range of LO is now < 1.5:1, dramatically easing V/F linearity performance required, and minimizing other effects, such as tuning for capacitor tolerances.
The inversion involved does mean that higher pitches are controlled by relatively lower amounts of output current from the exponential generator. This must be stable for excellent frequency accuracy. Stability should be the case though, because the LO has a compressed frequency range of < 5:1. More particularly, the most musically useful frequencies (< 5 KHz) are used under conditions of higher current from the exponential generator.
In this application of frequency synthesis, a “pure” RF is mixed with square-wave LO to produce clean IF signals: the pitched audio range signals we want. There cannot be an image frequency, because the RF has very low harmonics, and none image to the IF band.
Additionally, plural “channels”of like-oriented low-side injected down-conversion can be done relatively inexpensively, affording a programmable frequency spectrum, esp. where one exponential generator is used to make multiple frequencies, each frequency being a fraction of the output current. In the case of the inverted frequency response, these fractions map to higher harmonics. There is even an option of restricting the base frequency range to ≈︎10 KHz, leaving lowest-current output solely for harmonic generation.
This system is also markedly easier to tune, because full-scale output from the exponential generator, where raw CV=0V represents DC, and LO=RF. So, these two frequencies are tuned to beat very slowly. Then, at full-scale CV input (CV=10V, becoming -10V after summing inversion), the LO is tuned to Fmin, to match the highest pitch out of the IF. The V/OCT scale control does not interact with the first adjustment, and interacts only moderately with the second adjustment.
Frequency Plan using 32.768 KHz RF REF LO
RF LO IF=RF-LO
32.768 KHz 32.768 KHz 0 KHz
32.768 KHz 27.768 KHz 5 KHz
32.768 KHz 22.768 KHz 10 KHz
Limiting LO frequency output for a more musically useful frequency range of 0-10 KHz makes a lot of sense. Neither RF or LO oscillators, nor upper modulation products are in audio range. This limits the effects of mixer feed-through. More substantially, frequency range of LO is now < 1.5:1, dramatically easing V/F linearity performance required, and minimizing other effects, such as tuning for capacitor tolerances.
Comments
Post a Comment