This is the circuit diagram from the LM1815 datasheet. I understand everything in this circuit except for the two diodes that I have marked with red. They are in a peak-detector circuit, but all of the implementations of peak detectors circuits that I am aware of only use a single diode. I tried adding them to a LT spice simulation but their effect on the output was insignificant.
You’re looking at a block diagram: it is a somewhat abstract depiction of what is going on. The diodes on the block diagram are not literal components. They are blocks, just as the rest of the diagram. They represent a circuit that has two diode drops and conducts in one direction only. The implementation doesn’t have to use diodes.
The op-amps on the block diagram may not be literal op-amps. They can be application-specific gain stages with negative feedback. And so on.
To understand from the block diagram the purpose of the diodes for the zero crossing and peak detector, consider this:
- The peak detector has a decay resistor referenced to 0V (gnd)
- the differential inputs will then be 1 diode drop from 0V.
- this also means the signal input will be active in the region above 1 diode drop below ground
- elsewhere the signal input is clamped above and below ground to limit the range further but the transition of importance is observed as "just" above -0.5V for a low diode current voltage.
Now the "Theory of Operation" in the datasheet ought to make more sense.
How it is done is less important unless you like the nitty-gritty of how to design all these blocks in the diagram with the fewest number of transistors and be reliable.