This means lesser charging current means lesser voltage needed to charge and discharge giving more flatter ramping output voltage. This will also give smaller peak to peak output as the result. In this demo, R7, C4 output node yields +/- 240mv of ripple voltage (triangular wave) riding on ~6 volts DC. See fig. C with overlaid linear signal combined.
Some of you may think that 240 millivolts of triangular wave is too small for LM319 to amplify? Rest assured fellas, most comparators are designed to amplify differential voltages smaller than 5 millivolts of overdrive. So, 240mv of this sort is way too easy for LM319.
Now we are done! Biasing the comparator using this reference is the next easiest thing to do.
Let`s analyze the drawing of Fig. 1. Notice the DC bias resistor R10 connected to output pin 12 of U2? This node produce more than 11 volts of continues pulses at 50% duty factor, but I simply tap R10 on that output node because this is the best possible source of DC bias to pin 10 of U1 – meaning it will always seat at the crossover voltage of the triangular wave appearing on pin 9 to obtain the required 50% duty factor. Another possible DC bias to pin 10 is to use trimmer pot connected to VCC and ground then the center wiper goes to R10 but adjusting it to half of VCC to sample the crossover point of that triangular wave appearing on pin 9 would be very tricky to achieve. It works ok in sim, and also in real world (sort of) but when switched off for a while (coffee brake) and then on again, it needs recalibration again ouch!. Temperature drift and VCC fluctuation affect that trimpot DC bias. With the circuit I made above, node B always follows the crossover point of node A at all cost – it is immune to parametric changes mentioned above.
Fig. D shows simulation of what’s going on at the output of pin 12 of U2.
Fig. E. Output waveform at node A and B.