Design Ideas: January 19, 1995
Fig 1a shows input filters and Schmitt triggers wired to square up tachometer signals. However, given normal var-iations in IC lots and operating environments, Fig 1a's fixed-bias scheme (comprising RV and RG) stands little chance of holding the dc level at the Schmitt triggers' inputs anywhere near halfway between their thresholds. Because the strength of tachometer signals is proportional to speed, below a certain speed, this circuit does not respond properly to the resulting low-level signals.
Fig 1b positions the Schmitt triggers' dc bias exactly halfway between their thresholds. In Fig 1b, the top inverter is a relaxation oscillator that produces an asymmetrical triangle-wave output. The triangle wave pingpongs between the two Schmitt-trigger thresholds. Thus, the wave's average voltage is the desired bias voltage.
However, you cannot simply connect this average voltage to the remaining signal-conditioning circuits. Loading the oscillator with bias resistors would stop the oscillator and perhaps cause interchannel crosstalk.
However, CMOS input and leakage resistances are so high that no dc current flows through the feedback resistor, RF. For this reason, although the signal at point 1 is a square wave, its dc value is the same as at point 2. Hence, the signal's dc component is the needed value, and circuit loading does not bother it. If you make the oscillator's frequency sufficiently high, the signal conditioner's input circuits filter the signal at point 2.
For tachometer generators, an RI of about 100 kOhms is necessary to avoid large signals' wiping out the inverters' input-protection diodes. CA, even for high-frequency tachometers, places the input filter's cutoff frequency at 10 kHz. In contrast, the oscillator runs at 1 MHz. Therefore, the input filters and the enormous filtering action of RBCI reduce the ac component from the oscillator to innocuous status, far below the level needed for false triggering. (DI #1599)