"Greg Hines" wrote in news:OtjHe.492$ snipped-for-privacy@nnrp1.ozemail.com.au:
I'll have a quick go at explaining. U3B and U5A form an oscillator: note the feedback from the output of U5A to U3B's non-inverting input ie +ve feedback. This forms a sawtooth oscillator. U3A is simply a buffer of half the supply voltage to set the DC operating point. Probably a single supply or battery supply. U5B is a comparator without hysteresis. Now, the Delta P signal is first amplified by U5C, a DC stage. And no, the 10k feedback resistor does NOT add hysteresis but simply puts a single pole in the response to roll off high frequency response, probably pulsations in the flow or pressure sensor.
The next stage is U5D forming an integrator. Of Course this has open loop gain at DC which means that the PWM is capable of going to 100% and 0%. I am assuming here though that the sawtooth amplitude is less than the supply rails, you can verify with calcs or simulation. When the system is in balance, the set point voltage of R28 will equal the value of Delta P average voltage and there will be a constant duty cycle value. Q2 is obviously an enable function.
The two 200k resistors in series in the integrator would have been chosen by the designer presumably because that was an available value, and/or to minimise the number of different values for a SMT design to minimise the number of different reels used.