How to alert to sudden temperature drop

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
I would love to build a simple device that alarms when there is a sudden drop in temperature - think walk from one room to another and it’s a couple of degrees colder, and an alarm sounds.

Does anyone here know of such a device or perhaps point me in the direction of where I would start?

Many thanks,

Andy
 

Reloadron

Joined Jan 15, 2015
5,998
Depends on how we actually define a sudden drop. Using your walking room to room with a change of two degrees isn't much change be it C or F. That said it will come down to whatever sensor is used response time. Temperature sensors are chosen based on things like temperature range, environment and in the mix is their response time. Temperature sensors become a world unto themselves. Comparison of Temperature Sensors is a basic start read on the subject. Once you find a sensor with the accuracy / uncertainty and response time you want then you move along to a compatible circuit to use with it. When comparing a few degrees of change things can get a little more complex and expensive. Anyway, if it were my project I would start by choosing a sensor which met my needs.

Ron
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
Depends on how we actually define a sudden drop. Using your walking room to room with a change of two degrees isn't much change be it C or F. That said it will come down to whatever sensor is used response time. Temperature sensors are chosen based on things like temperature range, environment and in the mix is their response time. Temperature sensors become a world unto themselves. Comparison of Temperature Sensors is a basic start read on the subject. Once you find a sensor with the accuracy / uncertainty and response time you want then you move along to a compatible circuit to use with it. When comparing a few degrees of change things can get a little more complex and expensive. Anyway, if it were my project I would start by choosing a sensor which met my needs.

Ron
Thank you Ron. I didn’t think about response times so that’s something else for me to factor in.

I will have a read of that resource now,

Andy
 

ronsimpson

Joined Oct 7, 2019
1,342
How are you going to wire all this up?
Wire or wireless?
I am measuring temperature using "Sonoff TH10 WiFi Smart Switch Remote Control Temperature Humidity Monitor T3Y1". I get them from China for less money but it takes a month. (banggood.com) It uses my WiFi network. I don't have much experience with them yet but seems good so far. I can see them from my phone from anywhere in the world.
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
It will be all wired.

And I’m nothing more than a hobbyist at this. I will be seeking someone to help make me something when I know the best way to do this :)

It would be nice if there were kits that did what I was looking for, but I think that’s a remote chance :)
 

MrChips

Joined Oct 2, 2009
23,958
Define what you mean by a sudden drop.
Does the absolute temperature matter?
Or just the fall in temperature?
Give us some examples of your definition of a sudden drop, for example, falling from 12°C to 7°C, or are you only interested in the difference of -5°C?

Then sudden has to have a time frame, for example, 5 seconds.

Hence, we can define sudden drop as a gradient, for example, -1°C per second.
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
Define what you mean by a sudden drop.
Does the absolute temperature matter?
Or just the fall in temperature?
Give us some examples of your definition of a sudden drop, for example, falling from 12°C to 7°C, or are you only interested in the difference of -5°C?

Then sudden has to have a time frame, for example, 5 seconds.

Hence, we can define sudden drop as a gradient, for example, -1°C per second.
Just the fall in temperature is what is important to me - the 5 degree is just an example and I think it would need to alarm over the course of 3-6 seconds.

I think that 5 Celcius drop is too high and in reality, a drop of 2-3 Fahrenheit would be enough to make it alarm.
 

MrChips

Joined Oct 2, 2009
23,958
Analog or digital solution?

You can do this in a microprocessor. Take temperature readings every second. Average over 5 seconds. Then examine temperature differences.

For an analog solution, smooth voltages with a 5-second low-pass filter, followed by a high-pass filter into an analog comparator.
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
Analog or digital solution?

You can do this in a microprocessor. Take temperature readings every second. Average over 5 seconds. Then examine temperature differences.

For an analog solution, smooth voltages with a 5-second low-pass filter, followed by a high-pass filter into an analog comparator.
Digital sounds like it makes sense but this is way beyond me so I will have to pass this onto someone else to look at now and see where it takes us :)

Thank you for the help.

Andy
 

Reloadron

Joined Jan 15, 2015
5,998
In addition to what I mentioned regarding response time of temperature sensors you may just wish to consider using a small micro-controller as also was mentioned. The use of a micro-controller would likely simplify the project. While a small amount of programming would be required the skill level of what is required is not all that complex or difficult. I see as I type Mr. Chips has added what would be the general idea. Take several samples at an interval and store the result. Then repeat the process and store the result. If Result1 is greater than Result2 then do something is about how the scheme works. There are countless micro-controllers out there which are easily interfaced to a temp sensor, Arduino comes to mind, PICAXE is another but there are also smaller modular package solutions. The goal being keeping things modular so you don't start with a pile of loose parts.

The digital analog thing is just a matter of how your sensor communicates with a device like a micro-controller. While not real simple it is not as complicated as it may sound. You and whoever assist you may want to give this a read. Later you can worry about how the sensor is setup as you do not want, for example, radiant body heat effecting your readings.

Ron
 

michael8

Joined Jan 11, 2015
156
Simple? How about two temperature sensors and a comparator. Have one sensor be slower than the other.

I played with something like this once using diodes (1n4148) as the sensors and a diff opamp circuit (lm324)
and moving a finger (heat source) near one of the diodes was enough to show a change...
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
Hi Michael,

That sounds like a really interesting idea - and probably something I could tackle.

I don't suppose you would know of a circuit diagram of something like this that I could copy? And what is it that indicates one sensor slower than the other?
Thanks,
Andy
 

Alec_t

Joined Sep 17, 2013
12,160
using diodes (1n4148) as the sensors
Using the smallest possible surface-mount diodes would minimise the thermal mass and hence speed the response to temperature change.
Just curious; are you planning to go ghost hunting with this gizmo?
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
Using the smallest possible surface-mount diodes would minimise the thermal mass and hence speed the response to temperature change.
Just curious; are you planning to go ghost hunting with this gizmo?
I am indeed - Bit of a guilty please of mine at the weekends :)

I love tinkering and used to be a bench engineer many many years ago, but forgotten so much in the last 30 years.
 

OBW0549

Joined Mar 2, 2015
3,566
And what is it that indicates one sensor slower than the other?
You can make one sensor slower than the other by enclosing it in some sort of insulating material (foam, cloth, etc.) and/or attaching it to something with a bit of thermal mass, such as a chunk of metal. Either one will slow down the response of that sensor relative to the one hanging free in mid-air.

That said, my choice would be to use two thermistors rather than two diodes, as the thermistors would give more output voltage to play with.
 

Thread Starter

AndyD-UK

Joined Nov 1, 2019
24
You can make one sensor slower than the other by enclosing it in some sort of insulating material (foam, cloth, etc.) and/or attaching it to something with a bit of thermal mass, such as a chunk of metal. Either one will slow down the response of that sensor relative to the one hanging free in mid-air.

That said, my choice would be to use two thermistors rather than two diodes, as the thermistors would give more output voltage to play with.
Thank you - I will probably try both anyway :)

I do need to track down a circuit diagram of how all of this would go together. Any ideas what I might search for?
Thanks,
Andy
 
Top