Few circuits are more commonplace than the basic squarewave-output astable multivibrator illustrated in Figure 1. Useful whenever a relatively non-critical clock source is needed, this familiar circuit can be implemented with any rail-to-rail-output comparator or the venerable LMC555. The output period is given by 2log(2)RC = 1.386RC, and the stability is better than ±100 ppm/°C against temperature and ±100 ppm/V against power-supply variation.

But if an output duty cycle different than 50:50 is needed, application of this simple circuit gets complicated. That’s because the usual ways in which the basic RC multivibrator is modified for asymmetrical output timing either allow annoying interaction between output duty cycle and period or degrade the tempco and the PSRR. But a new twist to this old circuit (Fig. 2) adds a few diodes to the traditional recipe to allow independent adjustment of pulse width and period while retaining the good frequency stability of the standard topology.

The new modification starts with an old idea—addition of diodes D1 and D2 so that the positive and negative phases of the output cycle are controlled by separate timing resistors R1 and R2. This trick works well and is a handy way to implement a multivibrator with an arbitrary output on/off ratio. The trouble is that D1 and D2’s temperature-dependent diode voltage drops (V_{ds}) can drastically erode both the multivibrator’s output period temperature coefficient and PSRR. Canceling these effects, accomplished with compensation diodes D3 and D4, is the gimmick featured in Figure 2. If V_{d3} = V_{d1} and V_{d4} = V_{d2}, Figure 3 reveals how the duration of the positive output pulse is now given by:

and the negative by:

Static V_{d} effects thus cancel and leave the output period unchanged from the original Figure 1 circuit, which has no diodes at all. Admittedly, this approximation ignores the fact that due to variation in charging current over the RC timing cycle, the various Vds aren’t perfectly constant.

Even so, canceling diode effects good enough so that matched diode tempcos add no more than about + ppm/°C to the oscillator period temperature dependence. The excellent PSRR of the original circuit is similarly preserved.

The new topology makes various novel timing circuits possible, one which is illustrated in Figure 4. Here, pulse width and frequency are independently adjusted by R1 and R3, respectively. The reason pulse frequency is independent of R1 is that total pulse period is given by:

P_{P} = 0.693(R1 +R2 + R3)C

But (R1 + R2) = R_{POT}, which is constant and independent of position the potentiometer wiper. Meanwhile, the pot wiper position determines and, therefore, pulse width (P_{W} 0.693R1C) independent of R3.

For most combinations of R and values, D1 through D4 can be simple switching diodes, such as the 1N4148. But for higher R values, where junction leakage can become a significant factor, better performance will be obtained if transistor collector-base diodes (e.g., 2N3904) are used.