Where and how best to install temperature sensor in UFH screed

[readiescards]

I want to be able to log the relative day by day temperature of my ufh.

I propose to simply secure a DS18B20 temp probe to the ufh insulation approx half way between two ufh pipes before the screed pour.

 

But are there any better suggestions? Does the cable need protecting from the screed material? from the screed installer boots?

 

Thanks 

[Onoff]

I ran mine in just another length of Pex-Al-Pex UFH pipe capped at the end with tape. 

 

 

Goes to the "dead zone" but I can pull the cable back to put it amongst the pipes if necessary. 

 

Think it was @Nickfromwales said about running in pipe and @PeterW who donated the sensor?

 

EDIT: Belt and braces if the sensor fails!

Edited March 2, 2017 by Onoff

[readiescards]

Good old @Nickfromwales - such a simple idea. Thanks! 

[Triassic]

I ran mine through small bore plastic corrugated conduit.  The wall of the pipe is thinner and its more flexible than UFH pipe.

Edited March 2, 2017 by Triassic

[Onoff]

The walls of UFH pipe are silky smooth though!

[Nickfromwales]

The corrugations are a bit of a concern if you are trying to push a flimsy wire past more than one bend. Fwiw, try long radius bends, or even better, as near to straight as possible. Fine if your getting the sensor in on the day into loose conduit, but a pita when your trying to change one out.  

Smooth walled pipe for me every time. 

Edited March 2, 2017 by Nickfromwales
Crossed with onoff

[Triassic]

Should have said the corrugated pipe I used had a smooth lining. 

[SteamyTea]

3 hours ago, readiescards said:

I want to be able to log the relative day by day temperature of my ufh.

I propose to simply secure a DS18B20 temp probe to the ufh insulation approx half way between two ufh pipes before the screed pour.

When collecting data you need to know what you are hoping to achieve.

With just one sensor, in one fixed location, all you can collect is he temperature at that spot, at the time you took the reading and only within the accuracy and precision of the sensor.

There are a few statistical methods that will allow you to reduce the inaccuracy, but with single point measurement it is impossible to infer what is happening over the whole slab.

You say you want to log the 'relative' day by day temperature.  This is known as an anomaly i.e. the deviation from the central tendency. This relies on historic data, so as time goes by, it gets more accurate.

By only sampling one point, and that point is not that important as you are looking at change, the sampling rate becomes important.

So if you only sample once a day, say midnight, you will get a different mean from two samples a day, say noon and midnight.

There are two ways around this:

More samples, say every 10 minutes

Random sample times, say between 120 minutes down to 1 minute

Fixed sampling is generally the easiest to understand and gives you plenty of data points to play with later, you can easily 'bin' the data into both time and temperature bins.

Random sampling will generally give you a better overall picture of what is going on, but takes longer to collect data for every time bin.

As data is easy and cheap to collect, store and process, fixed sampling is probably the easiest.

The location of the sensor is a bit harder.

Too close to a pipe and it will over read, too far from a pipe and it will under read.  This is why we generally use multiple sensors at different locations.  Then use a weighted averaging technique (just an arithmetic method to make the sensors show the same temperature regardless of position) to show the temperatures.

If I was using single point sampling, then I would place the sensor either 1/3^rd or 2/3^rd from the pipework.

Ideally I would use 3 sensors in one location and a 4^th sensor outside of the slab.  The 4^th sensor would be the 'true' sensor and the one that the others are calibrated against.

You can keep this sensor separate an use it to calibrate any other sensors.

By using 3 sensors in one location, if one fails, then the other two are still useful, and the sensors are really cheap, I think the lot that turned up yesterday were under £2 each.  So for an extra £4, it saves having to pull a wire out, replace and calibrate a replacement, and then feed it back in.  And they will fail.

One mistake I made when writing a bit of Python script to check and read that sensor was that if the sensor failed, the script stopped.  I am not a programmer and have little interest in it, but if it is part of a larger program, it can crash the whole thing, so worth thinking about how to loop around this.

One thing I have found out about the DS18B20 is that they do not like high humidity, so either get some ones that are already encapsulated or put them in after the slab had fully dried.

Edited March 2, 2017 by SteamyTea

[billt]

Put mine in 10mm cu tube.

 

Just big enough to take the DS1820 sensor and cat5 cable and smooth enough to be able to push the sensor to the end of the tube.

 

ST is seriously over thinking things, from a theoretical point of view.

 

Although there will be some temperature variation between loops, it's going to be small and unimportant in the scheme of things. One carefully sited sensor (away from the edges) is quite adequate.

 

Mine are used to control the heating as well as for logging. (The kitchen floor is an electric mat with inadequate insulation, so responds more quickly than the water fed and better insulated other rooms. It isn't used much, but the sun was out yesterday, so it didn't cost anything to run.)

[MikeSharp01]

Just planning my sensor locations and after going through all the treads here have come up with my plan. I will have four sensors across / in the slab. The slab has three UFH zones two which get full sun via the windows and one, at the north end, which does not. I plan to monitor the surface temperatures in the three zones, basic layout as attached, and have a fourth sensor in a no UFH area under one of the internal walls which itself is out of the reach of the sun. Another four sensors will monitor the main flow and each return UFH temperature. There are also several air temperature sensors in the scheme to control the UFH and the summer ventilation strategy with the roof vents. 

 

I think this layout will give me enough telemetry to enable me to play with things around detecting solar gain happening at the front of the building while also having control of the heating via the under wall / air temperature sensors as @JSHarris suggests. 

 

Have I got my locations right I wonder? Air temps can go in later with the frame build but the slab pour must go ahead in 10 days time.

 

[Onoff]

Could you not run your sensor ducts to the furthest possible point they can feasibly go but route them via points where they MIGHT be better. Then, knowing the run length, you can just pull them back as required. 

[MikeSharp01]

6 minutes ago, Onoff said:

Could you not run your sensor ducts to the furthest possible point they can feasibly go but route them via points where they MIGHT be better. Then, knowing the run length, you can just pull them back as required. 

Yes I guess so, I was going to use your Pex-Al-Pex idea and go straight (line) to the sensor points, so I can pull them out for replacements etc, I will fix the runs to the upper R193 mesh.

[Onoff]

1 minute ago, MikeSharp01 said:

Yes I guess so, I was going to use your Pex-Al-Pex idea and go straight (line) to the sensor points, so I can pull them out for replacements etc, I will fix the runs to the upper R193 mesh.

 

Not my idea. It was Nick or Peter's. In fact I think it was Peter sent me the sensor?

[Onoff]

Don't forget to tape / cap the ends! Cant remember if I squashed mine flat and duct taped or gun foamed it.

[joe90]

Ah, using off cuts of under floor heating pipe as duct for sensors , brillIant idea (but too late for me)?. Got loads if anyone wants some?

[MikeSharp01]

1 hour ago, joe90 said:

Got loads if anyone wants some?

You definition of loads is? As I may have a use. 

 

[SteamyTea]

As the sensors are only a couple of quid each, why not fit more.  The cabling is easy as they are 1Wire.

[joe90]

5 hours ago, MikeSharp01 said:

You definition of loads is? As I may have a use. 

 

 

I have a full roll left over and about 20 metres of bits. The only thing I was going to use it for was cloche ends for the garden ( bendable pipe that stays where you put it!). How much do you want?

[Onoff]

I was going to make a dragon/gargoyle/Alien Queen skeleton out of the stuff I've left over them foam it, carve it and GRP. To sit atop my gate pillars.....

[MikeSharp01]

On 25/06/2018 at 22:25, joe90 said:

How much do you want?

I don't actually know is the easy answer as the drawing says I need 380m of 16mm and I have brought 400m so I am hopeful that I will have enough for the sensors otherwise I will just use HEP2 (10mm or some such) to place the sensors.

 

On 25/06/2018 at 19:19, SteamyTea said:

As the sensors are only a couple of quid each, why not fit more.  The cabling is easy as they are 1Wire.

I have 10 sensors and they are all calibrating on the desk beside me as I type they are all reading 26.18 +/- 0.12  as the temp of the room, the room thermometer says it 25.4 I guess I need the temperature calibrator out! Once I get my WEMOS board up and running I will do some graphing.

[lizzie]

Our sensors are in the slab - we dont have a screed - we used ufh pipe as ducts and it was very successful and easy to work with.