Question about air conditioner regulator implementation

[Just_another_Dave]

Recently we bough a new air conditioner which unluckily had some weird behaviors (temperature measurements that magically disappear, modes that behave differently to the described behavior in the user’s manual). After calling both the installer and the official technical service (neither of them had tested that model and both of them said that they didn’t know what was happening), the manufacturer sent a representative that showed us some confidential internal documents in order to prove that the device works as intended, but that they made a mistake writing the user’s manual (which is the only document that they officially provide to everyone else).

However, those confidential documents showed some interesting behaviors that I wouldn’t have imagined. In particular, I was surprised to find that it uses an on-off controller that turns it on if the temperature surpasses that the user sets in the control panel (for example 26ºC). Then, the controller turns it off if the temperature decreases more than 2 degrees (24ºC). I don’t understand this decision, as it means that setting a temperature of 26ºC causes the room to have an average temperature of 25ºC.
Do you know if this is a common practice in air conditioner controllers? Is there a reason for not using a PID controller?

On the other hand, the manufacturer says that their devices just work with the provided control panel, but it seems that some third-party manufacturers like tado sell vendor-neutral control panels compatible with them.
Do you know if air conditioner control panels follow any communication standard (like the ones used for heaters)? In that case, is the on-off controller implemented in the control panel (where it would be possible to replace it)?

[BrokenYugo]

I'm thinking if it cut off right at 26 the building and stuff inside won't have had time to cool off, so it will just creep back up and short cycle if you don't program in some sort of correction overshoot.

Furnaces do similar stuff, cut off a bit early to account for the energy left in the hot heat exchanger after cutting the fire.

[Just_another_Dave]

I'm thinking if it cut off right at 26 the building and stuff inside won't have had time to cool off, so it will just creep back up and short cycle if you don't program in some sort of correction overshoot.

Furnaces do similar stuff, cut off a bit early to account for the energy left in the hot heat exchanger after cutting the fire.

That would make sense, but I would have expected it to be centered (-25.5ºC to 26.5ºC) as it usually happens in heater regulators.

It can use the temperature measured by the control panel or in the return conduct (according to the preferences of the user). The last option leads to higher deviations, as more time is needed to sense temperature changes in the rooms

[soldar]

However, those confidential documents showed some interesting behaviors that I wouldn’t have imagined. In particular, I was surprised to find that it uses an on-off controller that turns it on if the temperature surpasses that the user sets in the control panel (for example 26ºC). Then, the controller turns it off if the temperature decreases more than 2 degrees (24ºC). I don’t understand this decision, as it means that setting a temperature of 26ºC causes the room to have an average temperature of 25ºC.
Do you know if this is a common practice in air conditioner controllers? Is there a reason for not using a PID controller?

A PID (inverter) machine and an on-off machine are completely different machines. Completely different.

I would not buy an on-off machine because inverter machines are much more efficient.

But, yes, the temperature differential between on-off is called hysteresis and it is necessary and designed like that.

[EPAIII]

There may be some modern ones that work with fancier controls, but traditional AC control is a simple On/Off temperature switch, same as for a heater. Both heater and AC is either ON or OFF: normally there are no between states or proportional control. Usually the temperature sensing element is the same for both AC and heat. In both cases there is a temperature spread between the on and off points to prevent the unit from oscillating between on and off too frequently with only a small fraction of a degree of change. So the 2 degree spread sounds normal.

In the thermostats that I have seen that temperature interval is adjustable with an internal control that the installer should set for best performance with your particular heater and AC equipment.

The other thing that a furnace has is a timer that keeps the fan running for a minute or so after the heat source has turned off. That distributes the residual heat in it into the house so it is not wasted.

[soldar]

There may be some modern ones that work with fancier controls, but traditional AC control is a simple On/Off temperature switch, same as for a heater. Both heater and AC is either ON or OFF: normally there are no between states or proportional control.

Maybe in America, the land of cheap energy and of waste. Talking of residential AC, I would be very, very surprised if on/off units are sold in Europe any more in any significant numbers. Inverter units are much more efficient and that is what I see everywhere.

A bit over 20 years ago I installed an AC and it was on/off type. At that time inverter units were still new and quite a bit more expensive. That unit is still working well but the efficiency is dismal when compared to more modern units.

About ten years ago I installed another one and it was inverter technology and there was never any question about using anything else.

In my latest trips to China all I see is inverter units everywhere. Any on/off units are very old.

[Just_another_Dave]

However, those confidential documents showed some interesting behaviors that I wouldn’t have imagined. In particular, I was surprised to find that it uses an on-off controller that turns it on if the temperature surpasses that the user sets in the control panel (for example 26ºC). Then, the controller turns it off if the temperature decreases more than 2 degrees (24ºC). I don’t understand this decision, as it means that setting a temperature of 26ºC causes the room to have an average temperature of 25ºC.
Do you know if this is a common practice in air conditioner controllers? Is there a reason for not using a PID controller?

A PID (inverter) machine and an on-off machine are completely different machines. Completely different.

I would not buy an on-off machine because inverter machines are much more efficient.

But, yes, the temperature differential between on-off is called hysteresis and it is necessary and designed like that.

I understand that hysteresis is necessary (I’ve designed on-off controllers before), but what surprises me is that it isn’t centered around the temperature set by the user. For example, if you set a temperature of 27ºC, the minimum and maximum temperatures would be 25ºC and 27ºC (centered around 26ºC) instead of 26ºC and 28ºC. Furnace control panels normally use the second option.

According to the user’s manual it is an inverter machine, so it seems that some inverters are on-off machines

[soldar]

I understand that hysteresis is necessary (I’ve designed on-off controllers before), but what surprises me is that it isn’t centered around the temperature set by the user. For example, if you set a temperature of 27ºC, the minimum and maximum temperatures would be 25ºC and 27ºC (centered around 26ºC) instead of 26ºC and 28ºC. Furnace control panels normally use the second option.

According to the user’s manual it is an inverter machine, so it seems that some inverters are on-off machines

It makes no sense that an inverter machine would be only on-off because the inverter would have no use. It just makes no sense. But, of course, inverter machines turn off when they are not needed. Say a machine can modulate between 100% and 30%, then if less than 30% power is required of course it will shut down and start up as needed.

And the insistence on a very precise temperature also makes no sense. In any air conditioned room there are not two places with the same exact temperature as the air is moving around. The air temperature is measured at the return entry and nobody is placed there. The numeric indication is merely a reference and will vary all over the room and will vary between units because the sensors will have a deviation of their own. Take several digital thermometers and they will all differ by a degree or two ... or three.

The unit may be defective or not but I get the feeling that you are quite despistated regarding all this.

[Siwastaja]

As others have said, inverter heatpumps also need on/off because they can't go down close to 0%, but rather like 30-40% minimum power. When less heating/cooling than this is needed, then it's under on/off control.

Control feedback strategy is again a separate issue. I don't think PID control is that common; it's too sophisticated for the accuracy needs, hard to understand for users/installers/maintenance and requires correct parameters. Instead, simple P controllers have a long tradition in HVAC (classic on/off controllers are actually P controllers, too, by mechanisms many overlook and which I won't go into here). For example, my air-to-water inverter heatpump uses a simple combination of hysteretic on/off + P controller; actual parameters are adjustable but as I have it set now, when T > Tset + 2, it turns off, when T < Tset - 1, it turns on, and power is modulated 100% @ T = Tset - 2 and minimum (40%-ish) at T = Tset + 2. This is rather easy to understand.

[soldar]

As others have said, inverter heatpumps also need on/off because they can't go down close to 0%, but rather like 30-40% minimum power. When less heating/cooling than this is needed, then it's under on/off control.

Control feedback strategy is again a separate issue. I don't think PID control is that common; it's too sophisticated for the accuracy needs, hard to understand for users/installers/maintenance and requires correct parameters. Instead, simple P controllers have a long tradition in HVAC (classic on/off controllers are actually P controllers, too, by mechanisms many overlook and which I won't go into here).

I totally agree. When I say "PID" in this case I and D are zero but I still use the general concept of PID as taught and used in engineering.

In any case, the service person is not going to get into that. If the electronic control board is faulty they will just replace it.

[T3sl4co1l]

Well... manufacturers not selling the products they advertise/document is one thing...

As for actual operation, it may simply be an offset.  For example, my in-wall AC unit has a "economy" mode, which I'm pretty sure just turns off the fan and AC after a while (normal, the fan runs all the time); but to make it worth while, it stays on and off longer, probably drifting to a higher mean setpoint in the process.  (I never use that mode, as the normal mode is adequate for the room I use it in.)

The concept of representative or related figures comes to mind.  For example pipe sizes, which are nominally rated in "inches" in the US, but bear no relation to the actual unit as such.  In particular, steel pipe sizes are... bonkers pretty much whatever they are, and copper pipe sizes are mostly nominal plus 1/8" (so '1/2" copper pipe' is 5/8" OD).  This... is stupid, but it's accepted industry norm, and everyone working with these materials knows to look up the real dimensions in a table.  (And, I forget if metric pipe/tube sizes have the same disparity?)  Perhaps your manufacturer is going for a similar thing, where their temperature is calibrated to some value (peak, valley, average, etc.; not to mention temperature drop between sensor and actual system outputs), not to any given parameter you might think / expect.

Tim

[Just_another_Dave]

I understand that hysteresis is necessary (I’ve designed on-off controllers before), but what surprises me is that it isn’t centered around the temperature set by the user. For example, if you set a temperature of 27ºC, the minimum and maximum temperatures would be 25ºC and 27ºC (centered around 26ºC) instead of 26ºC and 28ºC. Furnace control panels normally use the second option.

According to the user’s manual it is an inverter machine, so it seems that some inverters are on-off machines

It makes no sense that an inverter machine would be only on-off because the inverter would have no use. It just makes no sense. But, of course, inverter machines turn off when they are not needed. Say a machine can modulate between 100% and 30%, then if less than 30% power is required of course it will shut down and start up as needed.

And the insistence on a very precise temperature also makes no sense. In any air conditioned room there are not two places with the same exact temperature as the air is moving around. The air temperature is measured at the return entry and nobody is placed there. The numeric indication is merely a reference and will vary all over the room and will vary between units because the sensors will have a deviation of their own. Take several digital thermometers and they will all differ by a degree or two ... or three.

The unit may be defective or not but I get the feeling that you are quite despistated regarding all this.

The defect is completely unrelated to temperature control. The reason why I asked about how the control panel is usually implemented is because some of the specifications shown in the repairer’s manual weren’t what I expected. As a consequence, I’m just trying to understand how those systems work.
Just for clarification the unit is a SDH 19-050 IDNI from Saunier Duval.

Taking into account the previous responses, I understand that the inverter is used for modulating the frequency of the refrigerant plumb. Do those systems use a simple control loop (which measures the temperature error as the difference between the one set by the user and the one measured by the ambient temperature sensor and obtains the required frequency using a constant)?

[Someone]

Taking into account the previous responses, I understand that the inverter is used for modulating the frequency of the refrigerant plumb. Do those systems use a simple control loop (which measures the temperature error as the difference between the one set by the user and the one measured by the ambient temperature sensor and obtains the required frequency using a constant)?

Unlikely any significant smarts or technical complexity to it. Compressors will have a fairly narrow range of efficient operational load/power so even with an inverter modulating the power it will typically only be a 3:1 or 4:1 range. There will be on/off cycling in most installs (even fancy commercial installs don't usually have fully proportional controls).

[RJSV]

Seems a bit confused, grabbing a possible state of affairs, I.E. that 'there's a defect'.   Not aware of sensible accuracy and the whole 'thermal flywheel effect' that can slow things down to where an 'ON-OFF' control cycling is perfectly adequate for usual cooling accuracy.
   Similar, to use a baby, as example.  Hearing that 4 pounds is normal for a 2 month old, you wouldn't totally freak out, if baby was 4 pounds, 1 ounce...would you ?
I mean, in cooling situation there are temperatures varying all over the place, but usually not quickly, except for extremes like opening the outside door.
Anything controlling to within 3 degrees (f) or so is not bad accuracy.

   Dissecting 'PID' ? That's not 'control theory',...is it ?
 A proportional control, on a heat pump maybe, but still doubt that any room-sized volume to be cooled is going to be inside a large variation window, ultimately

[RJSV]

Oh and also wanted to comment about the setting, '26'...That might not mean '26', degrees, literally as control point but rather as the setting that will CAUSE control to center at '26' on average.  That might mean that physically, literally, it might be a switching point of 28, to start the on-off compressor.
OR, when the air conditioner is active, might be a switch-off point, at exactly 24.5, for, again, a resultant average, of '26' over the course of the next hour.