Adjust delta for new input sensor

diff --git a/pool.yaml b/pool.yaml
index badd2781a7bc4e65398b285687244aed84711ecc..b43b229f7d01c9b7d46014f115aeb34354d4b159 100644 (file)
--- a/pool.yaml
+++ b/pool.yaml
@@ -74,23 +74,27 @@ script:
         // We consumed this reading pair.
         //id(solar_out_fails) = id(solar_in_fails) = 1;
 
-        // We believe the output sensor is reading slightly lower than the
-        // input sensor. We see a fair amount of hysteresis, and the bypass
-        // kicks in before the output reports lower than the input... but
-        // the roof temperature continues to rise. When it gets high enough,
-        // the bypass turns off again and the warm water is pumped out. But
-        // then the reported output temperature is lower than the input again
-        // and it switches off again. Repeatedly, each 5 minutes over the
-        // course of an hour or so.
+        // The sensors are not perfectly calibrated. We believed the original
+        // input sensor was reading slightly higher than the output sensor,
+        // and we saw a fair amount of hysteresis. The bypass opened when the
+        // output *reported* lower than the input... but the temperature in
+        // the roof continued to rise. When it got high enough, the bypass
+        // closed again and the warm water was pumped out. But then the
+        // reported output temperature was lower than the input again and it
+        // switched off again. Repeatedly, each 5 minutes over the course of
+        // an hour or so.
         //
-        // By adjusting the output sensor up by about 0.1°C we attempt to
-        // reduce this hysteresis. Overcorrection would mean that we keep
-        // pumping heat into the sky for longer than we should, which is
-        // much worse than a little bit of hysteresis when the heat *is*
-        // actually accumulating in the water on the roof even while we
-        // vacillate.
+        // By adjusting the input sensor down by about 0.1°C we attempted to
+        // reduce this hysteresis. Overcorrection would mean that we we keep
+        // pumping heat into the sky for longer than we should, which is much
+        // worse than a little bit of hysteresis when the heat *is* actually
+        // accumulating in the water on the roof even while we vacillate.
         //
-        outtemp += 0.2;
+        // Having replaced the valve assembly in 2024, the new input sensor
+        // seems to be reading *lower* than the output by about 0.2°C, so we
+        // end up pumping heat into the sky, especially in the evenings. So
+        // we now make the correction in the opposite direction.
+        intemp += 0.2;
 
         if (outtemp == intemp)
           return;