With regard to the input resistor and capacitor R1 and C1 in your post #22, it is more of an issue there than just protecting the input.
The gain of the circuit (output code / input voltage) is proportional to the feedback resistor divided by the input impedance, R2/Z1. When the input impedance is just a resistor, that is simply R2/R1. However it is different when there is a capacitor or RC circuit for Z1. With a capacitor instead of a resistor at the input, driven by a low impedance source such as an op-amp, the gain increases with frequency. The impedance of a capacitor is Z = 1/(2*pi*f *C). For example, the impedance of a 1µF capacitor at 1000 Hz is 160Ω, so the gain of the Sigma loop is proportional to R2/Z1 = 100kΩ/160Ω = 625. And it will be ten times that at 10kHz.
By putting a resistor R1 in series with the capacitor (or by driving the circuit from a source that has significant intrinsic resistance) the gain is lowered and made less frequency dependent. For example, R1=1kΩ in series with C1=1µF, driven by an op-amp with essentially zero output resistance, the gain at 1kHz becomes R2/Z1 = 100kΩ/(1000 + 160) = 86.
With a capacitor input, the basic equation that comes from solving KCL at the summing node is,
H is the proportion of time the feedback pin is high, from 0 to 1, and with no input signal that is nominally H=0.5. The symbol V1 with the dot on top is the rate of change of voltage at the input to capacitor Ci. This analysis assumes no additional resistance in series with the input capacitor. A step of voltage at the input will peg the output, that is, it will drive the Prop output code to the full count or to zero count, and that can happen when the rate of change of voltage exceeds ±1.65/(R2*Ci). Again, that equation is for a capacitor alone (without R1 in series with it).
3.474 v P-P is the magic number for +4dBu which would be ideal. 0.894 v P-P for -10dBV. I guess I could be happy with -10dBv, but the rest of the audio path is designed for over +6 without distortion.
Lardom, if the resistor R1 in series with the capacitor is 33kΩ, the impedance of the capacitor becomes insignificant at audio frequencies and the gain is R2/R1 = 100kΩ/33kΩ = 3. If you look at the effect of a -15 volt step at the input, and assume that the capacitor is initially discharged, the instantaneous current into the Prop's substrate protection diode is (-15V + 0.6V)/33kΩ = 0.44mA. That current drops off fast as the the capacitor charges up to -15V on the input side and about -0.6V on the other side. The data sheet suggests that current through those substrate diodes be limited to less than 0.5mA. That is a conservative figure, good to follow in general. But the Prop pins are capable of standing off a lot more current, especially a pulse of very short duration.
If you make C1 big enough, its effect on the gain vs. frequency curve will be minimal, since its impedance, compared to that of R1 will be negligible. Just pick your minimum frequency and make sure that C1's impedance at that frequency is no more than, say, 1/10 that of R1 -- or even less.
BTW, to accommodate a 3.474V P-P range, R1 should be at least 160K, assuming a 100K feedback resistor and the usual 50% margin for threshold voltage variation.
Tracy, thanks so much for your posts. There's a lot of information for me to soak up. I'm just guessing at this point, but I take it 160k/100k with a 1uf would put me in the ballpark??
Comments
The gain of the circuit (output code / input voltage) is proportional to the feedback resistor divided by the input impedance, R2/Z1. When the input impedance is just a resistor, that is simply R2/R1. However it is different when there is a capacitor or RC circuit for Z1. With a capacitor instead of a resistor at the input, driven by a low impedance source such as an op-amp, the gain increases with frequency. The impedance of a capacitor is Z = 1/(2*pi*f *C). For example, the impedance of a 1µF capacitor at 1000 Hz is 160Ω, so the gain of the Sigma loop is proportional to R2/Z1 = 100kΩ/160Ω = 625. And it will be ten times that at 10kHz.
By putting a resistor R1 in series with the capacitor (or by driving the circuit from a source that has significant intrinsic resistance) the gain is lowered and made less frequency dependent. For example, R1=1kΩ in series with C1=1µF, driven by an op-amp with essentially zero output resistance, the gain at 1kHz becomes R2/Z1 = 100kΩ/(1000 + 160) = 86.
With a capacitor input, the basic equation that comes from solving KCL at the summing node is,
H is the proportion of time the feedback pin is high, from 0 to 1, and with no input signal that is nominally H=0.5. The symbol V1 with the dot on top is the rate of change of voltage at the input to capacitor Ci. This analysis assumes no additional resistance in series with the input capacitor. A step of voltage at the input will peg the output, that is, it will drive the Prop output code to the full count or to zero count, and that can happen when the rate of change of voltage exceeds ±1.65/(R2*Ci). Again, that equation is for a capacitor alone (without R1 in series with it).
BTW, to accommodate a 3.474V P-P range, R1 should be at least 160K, assuming a 100K feedback resistor and the usual 50% margin for threshold voltage variation.
-Phil