Evohome Bug

[mikey8156]

No signal is sent to the boiler to fire until the room temperature has fallen by 0.5C below the set point. This is not normally a problem as other radiators in the system are telling the boiler to fire but if only 1 radiator is demanding heat from the boiler I get temperature oscillation in the room.

[electronicsuk]

How is that a bug? All thermostats have some level of hysteresis, and 0.5C is probably pretty good compared to a mechanical room stat. Without some kind of deadband, the motorised valves and boiler would constantly be cycling on and off, which isn't going to be any good for battery life, or your boiler.

I have my doubts as to whether 0.5C is really even all that perceptible - I imagine it's more a case of the room haven risen above the setpoint, your body adjusting to that temperature, and then the temperature falling below the setpoint, the sum of which is enough to be noticeable.

[G4RHL]

I have noticed a bug in Evohome. If the room temperature has been above the set point and the room slowly cools, Evohome will not demand the boiler to heat when below the set point but waits until the room temperature is 0.5C below the set point. This makes for an uncomfortably cool room.

Does not sound like a bug to me bit more a design feature. Wonder how many notice a 0.5c drop in temperature.

[mikey8156]

It is a bug because normally the thermostat keeps the room temperature to within 0.1C. I have a digital thermometer next to the room thermostat so I can see exactly what is happening.

[erik]

Does yours only cool down to 0.5 below setpoint, or does it also heat up to 0.5 (or more) above setpoint when heating?

[mikey8156]

Only to below the 0.5C., no overshoot once the heat kicks in.

[Rameses]

I have noticed a bug in Evohome. If the room temperature has been above the set point and the room slowly cools, Evohome will not demand the boiler to heat when below the set point but waits until the room temperature is 0.5C below the set point. This makes for an uncomfortably cool room.

No it won’t do this. You may have seen these temperatures but this doesn’t mean that this is how evo control works or what it is trying to do, it would be working hard to keep that the setpoint.

The Evo will start firing the boiler at +0.5°C above Set Point - this is around the control band of Set Point ±0.5°C

At +0.5°C is will be a small demand AND this will be in relation to the amount of heat the room needs at setpoint.

At this mild time of year (plus bodies in room etc) the load at setpoint will itself be very small so at +0.5°C it will be next to nothing and hard to trigger a 1 min demand on the boiler, almost certainly it will be delaying firing cycles to build up to the min on time of (probably) 1 min. (looking at it in 10 min blocks (cycles per hour setting - 6 default, 12 max), checking as it goes, that demand is merited and ensuring we don't get an overshoot also.

Our thermostats display 0.5 increments. (our stats measure greater level of accuracy, but dont necessarily display)

If you do really feel that -0.5 under setpoint is uncomfortably cold, then put your stat up 0.5 degree (or amend the offset within the system settings) - we believe in comfort, so tweak the system to suit - either way it will work optimally.

[mikey8156]

You say, The Evo will start firing the boiler at +0.5°C above Set Point - this is around the control band of Set Point ±0.5°C. I am saying there is no demand sent to the boiler until the temperature the temperature is below 0.5C. of the set point. It is another example where the user observation is different from Honeywell.

[erik]

could it be that the outside temperature has dropped very much lately (in the Netherlands at least) and that Evohome still has to learn that the house is cooling down much faster than it used to up untill recently? And therefor it lags behind in sending a demand signal to the BDR91? And that the algorithm is missing some sort of 'over-rule' feature to over-rule this learned behavior once temperature unexpectedly drops below setpoint?

Kind of like it's saying: I see the temperature is below setpoint, but I'm going to ignore that. The house normally cools down much slower. So I'm not going to fire the boiler yet untill the time where I would normally expect the room to have cooled down as much. And slightly adjust the learned curve up untill then. So it will go a little better next time. Etc. etc. If that's the case, then this problem should go away in a week or 2.

[Rameses]

You say, The Evo will start firing the boiler at +0.5°C above Set Point - this is around the control band of Set Point ±0.5°C. I am saying there is no demand sent to the boiler until the temperature the temperature is below 0.5C. of the set point. It is another example where the user observation is different from Honeywell.

This depends on the thermal curve of the room you are observing. If there is residual heat in the radiator (its not stone cold at 0.0 setpoint I am guessing) and the change/drops are slighter than the sampling cycles, then it wouldn't trigger a fire command so as not to waste energy or overshoot.

Change the cycle rate, this might match your expectations more.

[mikey8156]

No, what ever the weather the temperature of the room has to fall 0.5C below the set point before heating is demanded. This effect is only seen when only 1 radiator is demanding heat.

[mikey8156]

The radiator is cold, the temperature has been above the set point in the afternoon, only begins to warm up 0.5C below the set point. Changing the cycle rate would not change the problem of no demand being sent?

[Rameses]

could it be that the outside temperature has dropped very much lately (in the Netherlands at least) and that Evohome still has to learn that the house is cooling down much faster than it used to up untill recently? And therefor it lags behind in sending a demand signal to the BDR91? And that the algorithm is missing some sort of 'over-rule' feature to over-rule this learned behavior once temperature unexpectedly drops below setpoint?

Kind of like it's saying: I see the temperature is below setpoint, but I'm going to ignore that. The house normally cools down much slower. So I'm not going to fire the boiler yet untill the time where I would normally expect the room to have cooled down as much. And slightly adjust the learned curve up untill then. So it will go a little better next time. Etc. etc. If that's the case, then this problem should go away in a week or 2.

Top be fair to Erik (thank you Erik ) he is observing correctly. A mild time of year challenges all thermostats with thermal changes which are 'fine'.

[erik]

Thanks Rameses. But mikey is saying it happens no matter what the weather is. That would rule out this theory.

Besides, if it WERE true, I think the algorithm should always kick in a heat demand when the temperature is below setpoint (and decreasing). Even if what it has learned so far predicted that the temperature shouldn't be below setpoint yet. The Evohome in that case has to 'admit' that its prediction has 'failed' and needs to take immediate action instead.

Sounds like Evohome might be stubbornly rely'ing too heavily on what it has learned and not enough on what's actually happening.

[mikey8156]

I have just bought a temperature data logger and will be able to show exactly what I mean.