PT1000 4 in parallel and series to average temps of two tanks

I have two hot water tanks, one is heated by heating elements, and both are heated by a heat pump. The one with the electric ellemtns flows into the other as the under floor heating takes heat from one and returns cold to the other.
Basically, I want to measure the average temperature of both tanks as one measurement, in order to r=trigger the heat pump. I want to avoid constantly cycling the water through them.
So I came up with the idea of putting 2 pt1000 sensors in parallel at the top outlets and 2 in parallel at the bottom inlets, and then wiring the pairs in series in order to average the temperatures.

Would this work?

Welcome to the forum.

For any project, trying to cut corners or trying to outsmart yourself is never a good idea, certainly not at the very start of the data path.
The physical sensors are the start of the path that results in calculating the temperature and, for example, showing it on a display. Treat the sensors as if they are the most important part of the project.

Suppose that a wire of one of the sensors is broken or if there is a short circuit. You can detect that if each individual sensor can be read by the Arduino, but not if you have combined multiple sensors into a single signal.

Try to get the data from each individual sensor as good as possible into your Arduino board. After that you can do in software whatever you want.

Short answer: No

Putting two RTD’s in parallel is the same effect as putting any other two identical resistors in parallel. You get half the resistance. So your 1000 ohm RTD’s would show 500 ohms. A very very cold temperature.

But then the parallel pairs are put in series to get back to 1000. The result is rather meaningless since there is no idea what temperatures are contributing to the result.

1 Like

No, it would not work to average the temperatures.

Even if you assume that the temperature response of a PT1000 resistor is linear (it isn't):
Rt = R0*(1 + aT)

the parallel combination of two PT1000 resistors is nonlinear:
Rp = Rt1*Rt2/(Rt1 + Rt2)

1 Like

Hello greenblood2

Welcome to the worldbest Arduino forum ever.

Do you have to use PTxxxx sensors?

A nice and clever alternative is to use 1wire Dallas Sensor DS18S20.

Have a nice day and enjoy coding in C++.

2 Likes

Yes I have to use a pt1000. It’s an existing heat pump with one pt1000 in a thermo well but it only takes the temperature of one tank, and now I have two. The second tank has no thermowells, and is also fitted with heating ellemnts that heat just that tank.
I was thinking that if I put a sensor at the top of both and at the bottom, and put in series the 2 pairs, they’d average out roughly.
Ie tank one top 70c, bottom, 20c & tank two top 50c and bottom 40c.

70c = 1270

20c= 1077

50c=1194

40c =1155

Maybe if the pairs were grouped in a certain way, ie the most likely hotter pair (tops of tank) and the most likely colder pair (the bottoms).

Anyone good with maths? Lol

Think think it will work ok either way.

70c = 1270
20c= 1077
50c=1194
40c =1155
615.41 tops parallel
+
555.44 bots parallel
= 1170 = 44c

Or -

Tank 1 top and bot parallel 580.73
Tank 2 top and bot parallel 587.09

= 1168 = 43c

Hi,
Can you please post a diagram of your tank and plumbing arrangement and where you have the temperature sensors.

Knowing the temperature of each tank separately has the advantage of fault detection and you can do the "averaging" in software.

A picture is worth a thousand words.

Thanks.. Tom.. :grinning: :+1: :coffee: :australia:

@greenblood2
Resistors in parallel:
R1*R2/(R1+R2) is only equal to (R1+R2)/2 when R1=R2, so it won't work.

1 Like

Where the red crosses are are where I propose the pt1000s. There is no software for averaging the sensors in this situation.

Which is which tank, and why two tanks?
Picture of a hand drawn diagram would be better and include more info.

Thanks.. Tom.. :grinning: :+1: :coffee: :australia:

Doesn't matter the parallel combination won't work

You can strap ds18b20’s to the outside of the pipe and then add insulation on top, it will be pretty accurate !

And don’t forget water sits in layers and does not mix unless there is a flow , so the average of a top and bottom temperature is not the average temperature of the water , it can be virtually full of hot water and the bottom will still be cold .( which why the water remains hot until the tank has been emptied of hot water and all replaced with cold).
The layout of the tanks is important - do they thermal syphon between them ? Where is the hot water taken from? . Where is the cold inlet ?
How does the pump circuit work ?

Normally I would expect a three way valve to direct the heat pump output to either tank, controlled by thermostats near the base of the tank .
Your existing PRT maybe to control the outlet temperature of the heat pump.

I can go into detail but it’s complicated. I have just discovered it’s NTC not pt1000. I’m not sure if these can be used in the manner I said?

Four NPT (not pt1000 as I first thought).
Two pairs in parallel, and the pairs in series.

They can be used that way, but they won't give the average temperature. They will give some other number.

You haven’t said so explicitly, but is the reason you want to do this so you can just plug the parallel arrangement into an existing input of the current control system? Your goal is to not create a different system using an Arduino or other microcontroller. Correct?

1 Like

Yes that’s it. The heat pump has a NTC as a temperature sensor input. I need to have a rough average of the temperature of both tanks to feed the heat pumps control system.
I am not using an Arduino.