Split system AC control loops: are my expectations too strong?

[Whales]

I am experiencing my first home air conditioning (the future!) and its behaviour is confusing me.  It's a 2.6kW Rinnai T-series inverter (HSNRT25B), the smallest in this line, setup in my bedroom.

Summary: The temperature setpoint on the remote only vaguely controls the behaviour of the unit.  The hysteresis seems to be about 5degC, but seems to vary quite a bit.  Having the thing on causes constant comfort fluctuations.

Detail: I have a cheap little digital LCD aquarium thermometer with its probe directly touching the indoor unit's temp probe.  Its absolute accuracy might be in question, but when you see the symmetry of the problem you might realise it doesn't matter:

Mode HEAT target 16degC: turns on at 19degC and turns off at 24degC (just tested now).
Mode COOL target 28degC: turns on at 26degC and turns off at 21degC (approx from memory, been a while since summer).

These trippoints are really far apart and not at all what I'd expect.  I have to intentionally choose uncomfortable looking temperature targets on the remote to get semi-reasonable behaviour.  This makes me think I'm completely misunderstanding something.

Maybe the thermal mass of my room is too small relative to the unit size and its inbuilt minimum cycle timers are causing this?  Or perhaps this is considered normal behaviour?

I have also tried running my ceiling fan at the same time as the aircon, to mix things up and make the aircon think there is more thermal mass in the room, but it doesn't seem to affect the behaviour much.

[ArdWar]

Are you sure it is set up properly?

Heck, are you sure it is actually what you think it is?
"Inverter" AC don't typically have that much hysteresis, if at all. The lack of hysteresis is usually one of the biggest selling point for them.

[Whales]

Are you sure it is set up properly?

No idea   Can you configure/tune hysteresis and control loop params in these consumer units?

I've got the user manual and install manual but they don't mention anything about software setup or tuning.  (EDIT: both available online under "Downloads").  The user manual straight up says that you specify the "desired" temperature with the remote, which is what I would expect but don't see.

Heck, are you sure it is actually what you think it is?

Yes, I've checked the model number on the outside unit and Rinnai's site says it's an inverter unit.  Its current draw also varies quite a bit (I've made readings that roughly equate to around 300W and 600W).

[Haenk]

The on/off temperatures make absolutely no sense. I can only assume the sensor is faulty or they used a wrong component (i.e. the curve is not matching the real temperature).

[Whales]

That wouldn't explain the symmetry between the heat and cool modes.  If the temp sensor is being misread then they'd both be offset the same way in both COOL and HEAT mode, not in opposite directions.

The only causes that makes sense (to me) are this being the fault of the firmware/config or my expectations.

[Whales]

I wonder if they tested and tuned this design in a room with open windows?  The temp probe at the unit would read very differently to the temp probe in someone's hand, so they'd "fix" the problem by programming in big heating & cooling temp offsets.  Maybe.

I might try plugging the USB wifi dongle it came with back in and power cycling it outside.  Perhaps removing the evil dongle has upset the firmware somehow   EDIT: nope, no effect.

[mikerj]

From the T-Series brochure:

"The T Series intelligent sleep mode operates on a two-way temperature control that modulates up and down to mimic outside ambient temperatures, ensuring a comfortable sleep."

Are you sure this mode isn't engaged? Not sure how the temperature varying up and down makes for a more comfortable sleep though.

[Whales]

Good catch thankyou, it's not in sleep mode.  This mode shows "sleep" on the remote and also turns off the (obnoxiously) bright white LED display on the indoor unit that shows the temperature setpoint.

Sleep mode might actually help combat my issue by a couple of degrees, I think it tweaks things in the opposite direction.  Maybe I should try using it all of the time xD

[Whales]

It gets worse.  The "ROOM ifeel" (temperature-sensor-in-the-remote) function makes the problem worse by a few degrees.  This makes sense: the remote is a few degrees colder than the temp probe up on the inside wall unit, so it tells it that it should go even hotter!

It's now: mode HEAT+ROOMifeel target 16degC: turns on at 22degC and off at 25degC

LOL.  Theoretically I could exploit this to fix the offset (by heating the remote and pressing the ROOM ifeel button) but (1) that doesn't fix the 5-6degC hysteresis and (2) I don't know if it forgets this setting or not.  I might just turn this feature off again, this is way too hot.

[Haenk]

Maybe exploit the offset by using HEAT when going to cool and COOL when going to heat? Maybe it's just a displaying bug after all?
(Though not sure if the unit actually switches by temperature or by setting...)

[ArdWar]

I'm inclined to return it, or call your installer, or however it usually works in 'straya idk. At least get them to check it.

The hysteresis is atypical for inverter unit, and the up to 8 degree offset is unacceptable regardless of everything.
I can't imagine that's anything resembling comfortable.

[Whales]

I called Rinnai and described the symptoms.  I was told the hysteresis should be about 2degC (not sure if +/- or total range).  Got passed around 3 times.  Eventually got to an overseas call centre that couldn't hear me and hung up.  I'll try emailing them.

The on/off temperatures make absolutely no sense. I can only assume the sensor is faulty or they used a wrong component (i.e. the curve is not matching the real temperature).

The sticker on the panel says it's supposed to be a 15Kohm@25degC thermistor, but it doesn't mention the curve.  I tried to measure it: 20.2Kohm@18.4degC, which would vaguely indicate B=3920 if correct.  It was a pain to get to and I couldn't remove it entirely (I think they designed the units to require snapping some plastic off to remove the PCB properly, probably as some kind of tamper evidence) so I didn't fully characterise it.

As a test I set the aircon to mode HEAT target 25degC (which I think should give the unit the right thermistor resistance regardless of the curve).  The indoor unit only started ramping down its fan when its intake was around 29degC    I didn't push it any further to see when it actually turned off.   It also had a panic attack halfway through -- the indoor unit turned off and started flashing its display, whilst the outdoor unit kept running its fan but not the compressor.  I was only able to get it up to 29degC by holding a big sheet of cardboard in front of the indoor unit (for a good 10 minutes) so output air was partly redirected towards input. 

 :'(

[Siwastaja]

Sounds like it's broken. Or too difficult to use for a normal human being, which, to be fair, IMHO, counts as being broken. Contact the seller/installer and demand a fix.

[tooki]

Detail: I have a cheap little digital LCD aquarium thermometer with its probe directly touching the indoor unit's temp probe.  Its absolute accuracy might be in question, but when you see the symmetry of the problem you might realise it doesn't matter:

Mode HEAT target 16degC: turns on at 19degC and turns off at 24degC (just tested now).
Mode COOL target 28degC: turns on at 26degC and turns off at 21degC (approx from memory, been a while since summer).

These trippoints are really far apart and not at all what I'd expect.  I have to intentionally choose uncomfortable looking temperature targets on the remote to get semi-reasonable behaviour.  This makes me think I'm completely misunderstanding something.

One thing I would do is get a thermometer intended for air/gas.

An aquarium thermometer is designed to measure the temperature of a large mass of water with a slowly changing temperature, and the water’s high thermal mass and slowly changing temperature means that there is plenty of time for the probe to acclimate, which the water can do quite effectively.

In contrast, gases need temperature sensors with very low mass, because gases are terrible conductors of heat. A sensor intended for liquids can respond very slowly in air, depending on its design.

And if your aquarium thermometer’s possibly large mass is touching the small mass of the AC unit’s sensor, it might be completely slowing the AC’s sensor response by coupling a much larger mass to it. This could totally throw off whatever PID control loop the AC is using.

Now, I don’t actually think the aquarium thermometer sensor is likely to be the source of your problem, but it’s just something I would rule out first.

[Whales]

Contact the seller/installer and demand a fix.

Working on it.  I'll update here if it gets fixed and/or I find out what's wrong.

An aquarium thermometer is designed to measure the temperature of a large mass of water with a slowly changing temperature, and the water’s high thermal mass and slowly changing temperature means that there is plenty of time for the probe to acclimate, which the water can do quite effectively.

In contrast, gases need temperature sensors with very low mass, because gases are terrible conductors of heat. A sensor intended for liquids can respond very slowly in air, depending on its design.

Good points.  I'll throw some other thermometers of different styles and families at it.

And if your aquarium thermometer’s possibly large mass is touching the small mass of the AC unit’s sensor, it might be completely slowing the AC’s sensor response by coupling a much larger mass to it. This could totally throw off whatever PID control loop the AC is using.

I was originally hanging them touching or almost touching, but then I discovered that the indoor unit keeps its fan going (very slowly & silently) even after hitting its max trip point.  Now I just sit the thermometer on the plastic inlet grill atop the unit (a few cm from the internal thermometer); with the forced airflow they should be very close in temperature.

[Whales]

BUMP

Going through the warranty process, no resolution yet.

Rinnai told me over the phone that it should be regulating to within a couple of degrees C.

I have been visited 3 times by Rinnai's warranty subcontractors (they're a normal aircon business).  They have replaced the inside unit thermistors and now the whole inside unit control board, but the problem persists.  They don't know what else to do, they are going to see if they can get someone else to call me.

Not happy.  The unit still doesn't let the room get anywhere near its set point and overheats the room by a good 8 degC or so.  The oscillations are bad enough I have to change clothes constantly (or constantly use the remote to turn the unit on & off). 

> I'll throw some other thermometers of different styles and families at it.

I have tested and compared with another brand of thermistor and a type K thermocouple, all agree within 1degC.  The subcontractor's temp probe agrees too.