#garfield Logs

Jul 29 2022

01:11 PM rue_mohr: why does the calandar on the default debian install I did take up 1G of memory?
08:03 PM rue_mohr: so
08:03 PM rue_mohr: apparently nothing depends on it...
08:03 PM rue_mohr: I can just remove it
08:03 PM rue_mohr: and have 1G more free ran
08:24 PM aandrew: yay, got the flow sensor working
08:24 PM aandrew: it's actually push-pull, but the input on the ESP32 didn't like it, so I put a little buffer between it (74LVC1G97, I have thousands here) and that fixed it right up
08:24 PM aandrew: now I have to calibrate it
08:27 PM aandrew: 11 pulses per second and with this 1.5" flow sensor they say F = 0.5Q, where Q is in LPM, so 11Hz should be 5.5LPM which seems *awfully* low. I should turn on an "empty" zone because I did fill a 5gal bucket in about 20 seconds so that is about 16gpm which is 16*3.78 or roughly 60LPM, not 5.5
08:27 PM aandrew: the pressure sensor is also out of whack but that's because it's a 0.5V-4.5V sensor feeding a 3.3V total shit ADC on the ESP32
08:35 PM aandrew: 0.5V = 0psi, 4.5V = 100psi, I'm reading 2600/4096 counts which is a little over half full scale, so 3.3*2600/4096 or 2.09V. 100psi * (2.09V - 0.5V) / 4V should be 40psi but that's way low, it should be up around 70psi I think
08:37 PM aandrew: measured_psi = fullscale_psi * (sensor_volts - sensor offset) / (sensor_fullscale_mv - sensor_min_mv) sounds right for the math
08:51 PM aandrew: hm they might have shipped me a 150psi model not a 100psi model, the same numbers with 150psi fullscale give me 60psi which is a lot closer to reality
08:52 PM aandrew: if I turn on two zones the flow does damn near double (5lpm -> 8.5lpm) so maybe it is being restricted by the rotors
09:47 PM rue_mohr: oh I wonder if I have the numbers on mine
10:01 PM aandrew: the other zone has more on it and the flow's different by a bit so I guess it's in the right ballpark
10:01 PM aandrew: not expecting super accuracy
10:01 PM aandrew: just a means to identify either a zone not working when it should or a zone running when it shouldn't
10:01 PM aandrew: all in it turned out pretty slick
10:01 PM aandrew: gotta play with graphite or grafana or whatever the cool kids use to graph mqtt data these days
11:25 PM rue_mohr: I get surprising repeatability from the sensor I'm using for kitchen water