remote temp sensors instead of programmable stats in each room

[SimonHobson]

I did a bit of reading after going over the posts above and my original choice of standard electro-thermal actuators on the manifold seems to have one big disadvantage. Although the power consumption is "only" 2-3w per actuator, they take about 5 minutes to open! You can get more powerful actuators but even these take about 3 minutes and since I have made the house to be on the low side of thermal mass with a low mass UFH system to have speedy reaction times that is not helpful at all. They will obviously also take 3-5 mins to close which means more chance of a yo-yoing temperature.

FWIW from when I had responsibility for heating and A/C at my last job, if you have a very low thermostat hysteresis in the stats then I don't think it should be a problem.

From cold they do take a few minutes before they start to open, and from saturated hot they take a while to close. But in between they are actually quite responsive and if your system has a low enough mass, and you run with little or no hysteresis, then they will in fact almost modulate. The biggest problem we had was the original installation had mechanical 2-point thermostats without compensating resistors - so we did indeed get the problem you describe (stat clicks on, A/C fires up, stat stays on until room has cooled several degrees, stat clicks off, valve has now been open a while and so takes a while to shut, room undershoots even more, stat clicks again - this time to bring on heating, room warms up and cycle repeats )

When we replaced the mechanical stats with electronic units, things were a lot better - well as "better" as it possible to be while trying to please an office full of (mostly) women who all want individual temperatures which vary during the day and by time of month, and all without making any noise or moving any air whatsoever

So if you are using electronic stats, I'd be tempted to set them with zero (or very little) hysteresis and I think you may find they work reasonably well. Certainly for the Honeywell valves we were using, the manufacturer said there were no cycling restrictions, so it wouldn't really matter if the stat went on and off with every little draught - unless it has an annoying click each time.

It may be worth setting up a trial installation in one room - you can get the electro-thermo-hydraulic heads for some types of TRV, so it should be fairly easy to do a temporary setup and see how it works. Where I'd see it being a real problem is if your heating isn't well matched to the room. Eg, if the heating needs to be on for long periods (undersized rad), then the valve head will saturate and take a long time to shut. Conversely, if the heating is too large (oversized rad), then even a small opening of the valve will drastically increase the temperature of the rad and hence the heat input - meaning the valve will stay closed a lot and take a long time to open.

[biff]

oi simon,
        ave a looksee on my thread and see if i have come up with a miraculous method in mixing voltages and combining battery banks.
           you are after all a self aclaimed pc geek,??
        its in "observations and queries on combined.........."

[mistadave]

When we replaced the mechanical stats with electronic units, things were a lot better - well as "better" as it possible to be while trying to please an office full of (mostly) women who all want individual temperatures which vary during the day and by time of month, and all without making any noise or moving any air whatsoever

Hilarious!

Thanks for the great insight into how they work though, I was beginning to worry about how i would squeeze all these MOMO valves and bits of manifolds in - its going to be a lot easier with just one huge manifold and thermo actuators.

So to make sure I have understood correctly, by setting the thermostat with little hysteresis is basically setting what heatmiser call in their manual the switching differential down to say, 0.5 degrees. Then the stat would call for heat on and off loads (which isn't an issue with a thermal store instead of a gas boiler) and the actuators would react quickly because they are kept in the temperature zone where they move straight away. As opposed to being on for a long period and off for a long period where you would have to wait to reach the wax's melting point before it even started moving?

[SimonHobson]

Yes, what I call hysteresis is what Heatmiser call switching differential. Set it as low as possible, and of course, since you aren't firing a boiler in response it doesn't really matter how often it goes on and off. Just set any pump overrun to a decent time so that the pump doesn't keep starting and stopping - I assume you'll be using a modulating pump so it will be happy running against a variable, and sometimes nil, flow rate.

Where I could see there being a problem is :

1) The room is cold and the heating turns on. It needs to be on for quite a while to get the room warm, and so the actuator is saturated. Potentially this means the room temp will over-shoot when the stat turns off.
My feeling is that for this won't be much of an issue. If the heating is on for quite a while (eg 20mins and upwards), then there's either a limited excess of heating capacity so the overshoot won't be too bad, or there's a large thermal mass so the overshoot won't be too bad.

2) You turn the setpoint down in a step. The heating will need to turn all the way off and it may take a while for the room to cool. Again, I think the valve would need to be completely off for 20 mins or more before it would have a long start-up time. So there's a potential for an undershoot followed by a settling period.

I think the biggest risk is if you have some large radiators in a small (low heat load) room - ie the heating ability is disproportionate to the load. When the valves open, the rads can fill with hot water quite quickly, and even if the valve shuts again, you then have a significant excess heat input. You could well see them cycle.

[SimonHa]

You could use a one wire system  and a computer program  to operate the relays for the separate circuits.

Joining this debate late but it's certainly do-able and is pretty much what I've got: a 7 zone UFH manifold in the centre of the house, each circuit has about 2 or 3 rads plumbed in 15mm copper, 28mm from the manifold to the 27kW boiler. Control is by 1-wire: DS18B20 sensors in a few rooms, a HobbyBoards 8 relay board to control the 24V AC 4-wire actuators, the actuator "open" switches connect 240V to the demand signal for the Viessmann boiler. Software control is currently by some simple shell scripts on a Buffalo Linkstation, crontab for the scheduling, rrdtool for the temp graphs plus PHP so I can switch zones on and off manually from a web page (e.g. from an O2 Joggler in the kitchen). I also have a switchable "eco mode" which adds a resistor into the weather compensation circuit to make the flow/return lower to improve efficiency (often the boiler produces no noticable steam from the flue). You do want an override that bypasses the electronics during development though!

The software is pretty crude and could be much better. Admittedly I've also still 3 zones to plumb(!) first but so far I've been very pleased with the gas savings, and silly things like being able to control the heating from my bedside clock radio.

For retrofit, rather than re-plumb of CH, now I'd probably go for room TRVs wired back to a central point or HR20s though. This also kind of assumes you want to spend quite a lot of time setting it up and will enjoy that; otherwise there's more off the shelf stuff available now, e.g. Heatmiser, Honeywell CM20, though getting a system with many zones gets quite pricey.

HTH if anyone is thinking of going down the 1-wire route...

[TTP]

Hi
I read about the energy consumption problems of the actuators. If I well understand,  you must change the battery pack every 2 or 3 year. Have you some information about energy indipendent wireless radiators valve (no-battery)? I had found something here http://goo.gl/oEDzv. The document describe a sperimental application. Do you know any product whit this features.

[wookey]

I fondled an enocean peltier-powered device at Ecobuild, which was indeed interesting. I'm not sure it's made it to a product you can buy yet though.