Setting a Honeywell CM900 to control Mitsi ASHP

[TiredGeek]

Guys, I'm having problems with my ASHP's icing up, and I'm wondering if I've set the clever room stat properly to work with ASHP.
The room stat is a Honeywell CM900 Programmable Room Thermostat

There are a few settings I'm not sure about: Proportional band width, minimum on time, cycle rate, optimisation.

I have set the thing to the recommended settings for Heat Pump, which is minimum on time = 4 mins & cycles per hour = 3

A "consultant" has suggested the Honeywell maybe causing the ASHP to cycle on and off and leading to issues with icing, or even turning them off part way through the defrost cycle.
I wonder if it's because the thing is so accurate it's turning the units on and off too often before they've had much of a chance to warm the water enough for them to use it to do a defrost. I'm wondering if there's a way to "desensitise" it from it's current 0.5'c accuracy.
The problem only happens when the house is unoccupied and the stat is set to 10'c.

I don't understand what "Proportional Band Width" does as there is no explanation in the manual. Maybe that's what I'm looking for?

"Optimisation" turns the heating on early if it thinks the room temp won't make it to the set target in time. So say it's set for 12'c overnight 'till 6am then 18'c at 6.10am, it will turn the heating on at 5.45 to start to build up the heat to meet the target at 6.10.... I think.

Frankly everyone is scratching their heads at the moment over the cause of the icing so I wonder if there is a better setting for the room stat and if that will help.

[Brian_UK]

Brief description here..

Proportional Band Width
http://en.wikipedia.org/wiki/PID_con...portional_term

[TiredGeek]

Cool, thanks for that link......

Now, if someone could just explain the explanation to me as I still don't get it

[Brian_UK]

In simple terms it means that the control will allow the switching point, around the set point, to float slightly while the control monitors the result.

The temperature swings around the set point will reduce in size until the control is as close as possible to set point.

If you look at the graphs shown in the link you will see the sine wave reduce in amplitude.

[TiredGeek]

Thanks, I think I get it now

In effect then, if I alter the setting it has the effect of "deadening" the response of the stat a bit?
I have the option of 1.5'c to 3'c, what I'm trying to achieve is a stat that behaves more like the old bi-metalic ones, in other words if it's set to 18'c it switches heating on at about 16'c and off at about 19'c.
The Honeywell is too accurate causing my heat pumps to cycle too frequently....