AC behavior debugging

I’m trying to debug the central AC unit behavior in my apartment, and to that end I setup some home weather monitoring

Consider the following example, where you can see three distinct phases:

1 - AC is turned off, ambient temperature with some windows open. The living room has much better air flow than the office which has the window closed so the 2 degree temp diff makes sense.

2 - AC is turned on at 14:12, set to cool at 26C, duty cycle starts (as can be heard by the compressor), temps start to drop, as does humidity. I believe the BME280 sensor measures relative humidity.

3 - Compressor duty cycle stops, humidity starts to climb back even above previous levels???

Does this behavior make sense? I have a gut feeling that the air here is more humid that it needs to be, but I have failed to prove it thus far.

Any thoughts? cc: @jonathan

on the off chance of sounding like a complete idiot…

is your humidity sensor covered by a metal can ? can it be that some water condensed on it ?

Yes it’s relative humidity.
Though the absolute humidity rises too. This is mostly because the cooling coil is still wet from the previous duty cycle before condensed water has had a chance to completely drip off while the circulation fan is still active.
Some possible remedies:

  1. Cleaning - There may be some blockage (dust, fungus) between the coil fins or also common in the drip tray drainage that may be making things worse.
    Try to push some air via the drain pipe to unclog it. Also consider using some cleaning materials on the fins.
    If the fin density is high then that on its own is enough to trap condensation for longer via capillary forces, but dust and fungi surely don’t help.

  2. Also noting the ON cycle is a bit on the short side. There may be a temp error & offset compared to the return air temp sensor of the A/C
    The longer the cycle, despite some overcooling of the room, the smaller the ratio between what remains in the coil and what drips.
    This could be overcome by increasing the deadband (maybe also possible via dipswitches), another option is to set it in the “i feel” mode where the temperature feedback is from the remote.

  3. As opposed to heating cycles, the fan is usually kept on even when the compressor is off to maintain some cooling sensation for occupants.
    It may be possible to interlock the fan and compressor by adjusting dip-switches at the back of the A/C remote controller (check the manual 1st).

Some notes:

  • The effect of re-evaporating condensed water doesn’t lose overall energy (it’s a zero-sum in energy between sensible heat (temperature) and latent heat (humidity), however it’s not particularly healthy especially if the cause is from microorganisms.
  • Absolute humidity in the room is still a bit lower than what you started (you can calculate “humidity ratio” using %RH and T here http://www.sugartech.co.za/psychro/index.php), but indeed it doesn’t seem great.
  • Another possibility is freezing of the coil, usually due to lack of refrigerant. This can also cause the compressor to trip on low pressure.
1 Like

I know this behavior (empirically) from personal experience as well. When the AC reaches the target temperature it continues to blow air while the compressor is off and therefore the radiator inside isn’t cold. I think the reason for the increase in humidity is that the radiator is still wet, and possibly also the collecting tray below it has water, and the AC is just collecting all this water (which vaporizes and increases the humidity) and blows it back into the room. That’s my theory at least, it’s not tested.

Since you arranged this nice measurement setup, perhaps you want to test this - once you hear the compressor turns off, manually turn the whole AC off and see if it takes longer for the humidity to rise. Do a control experiment afterwards where you let the AC keep going (like the result in your mail) to have a control experiment with more or less the same conditions.

@kfir - nice try but it’s not that :slight_smile: the sensor is just hanging from the jumper cables

@jonathan - so many great ideas! The AC is rather old and I suspect cleaning is the immediate culprit since there definitely is some mold-ish smell as well. Will probably handle that first.

@guyovadia1 - interesting test to run with all the neats graphs I have

Thanks for all the advice everyone, will report back with progress!

For the fins this is what was used for the tami A/C.

There are more domestic solutions like using water diluted with bleach (can cause corrosion but once every-now and then should be okay).
But 1st start with the tray drainage…

Updating here for posterity’s sake -

I brought a pro technician to do the cleaning (on the landlords dime). There was indeed plenty of dust and fungus to be washed away. The most obvious effect is that the air quality is significantly better. But then the metrics have also improved:

Relative humidity peak has significantly decreased since the cleaning.

@jonathan can you explain how you plugged the values into the calculator you posted, and what absolute values you were looking at to determine how efficient the entire system is running?

Glad to hear it’s sorted, and even better when LL’s paying!

http://www.sugartech.co.za/psychro/index.php
plug in the air temperature (known as dry bulb temp) and %RH highlighted values in yellow
Circled in blue is the humidity ratio ( AKA absolute humidity) for the pair entered - measured in mass H20 per mass Air .
image

I’ve taken some formulae graciously provided by J and written some simple Python functions to calculate absolute humidity values which are added to my Grafana dashboard.

Feel free to use these:

import math

def vapor_pressure(rh, t):
    # rh in percentage as [0,1]
    coeff = 17.27 * t / (t + 237.3)
    return rh * 0.611 * (2.7183 ** coeff)  # kPa

def absolute_humidity(vp):
    return 0.62197 * vp / (101.035 + vp * (0.62197 - 1))  # g / kg (water / dry air)

def enthalpy(t, ah):
    return 1.006 * t + ah * ((1.84 * t + 2501))  # kj / kg

def dew_point(t, rh):
    return 237.7 * (17.27 * t / (237.7 + t) + math.log(rh)) / (17.27 - (17.27 * t / (237.7 + t) + math.log(rh)))

def wet_bulb_temp(t, th):
    return t * math.atan(0.151977 * (rh + 8.313659) ** 0.5) + math.atan(t + rh) - math.atan(rh - 1.676331) + 0.00391838 * rh ** 1.5 * math.atan(0.023101 * rh) - 4.686035
1 Like