Fortunately, the DS2423 has built in debounce circuitry. Here's some relevant notes from the spec sheet:
10) The counter inputs are designed for interfacing to mechanical switches and piezo sensors. If
interfacing to digital circuits, one should use an open drain driver.
11) A lower impedance pullup, e. g., for reed switches, can be achieved by connecting an external resistor
from the counter input to VBAT.
13) Each low-going edge on a counter input resets the channel’s debounce timer. The debounce time
starts as the input voltage rises beyond the trip point. In order for the next pulse to be counted the
debounce time must have expired.
From that homeseer post:
"had the same problem with my homebrew water meter reader. The pulses came too fast for my ocelot to record. I built a divide by 10 counter"
And that's why you should use an external dedicated counter, whether it be a DS2423 or something else. An external counter will monitor the input and count for you. The host only has to retrieve the current count.
Also:
"I actually want to have the fine resolution of 10 pulses per gallon. I want to accurately measure toilet flushes, dish washing, etc."
Sure, you can determine how much a toilet or a dishwasher uses in a controlled situation. You never flushed again before the tank was full to clear the floaters? Your dishwasher and washing machine have only one cycle and one water usage pattern? Even if you collect all the fixed types of water usage, you can't easily differentiate somebody turning on the shower then taking a whizz and flushing while the shower gets hot.
The only time a higher pulse rate might come in useful is for overall minor leak detection. Unless you have something that uses water all the time, like an RO filter, there should be some periods of no water draw. If you get some pulses during those periods, then you've got a leak. That doesn't tell you where the leak is, which could be as simple as a toilet flush valve or a dripping faucet or a bad sprinkler valve. Better to use water sensors in critical areas if you're that paranoid.
1PPG is more than enough to measure water consumption. I would monitor the counter once per minute. 1440 data points per day is more than enough.
10PPG is 12.8oz per pulse, about a glass of water. Too granular for consumption, not granular enough for useful minor leak detection.
100PPG is 1.28oz per pulse. Analysis would show if there's not an obvious water leak. That would be semi-regular pulses during times of expected inactivity. I would monitor the counter once per second, log the results at the 59 second mark and start again for the next minute. Assuming no other water usage, seeing a lone pulse on a regular interval, say every +/- 40 seconds, would indicate some sort of a consistent low grade leak.
