Simple thermal air flow sensor

Hello all,
I may sound stupid but does anyone have a simple schematic or know how to build a thermal air flow sensor/meter using two pt100, resistors, variable resistor, multimeter & of course a voltage source? I want to see and build the working principle behind the thermal flow meter. I don't really get how it works. If anyone can help explain in simple terms that'll be great. I do know that you have to use a wheatstone bridge. One sensor is for reference temperature and the other one is heated to maintain the temp difference when the air flows. I'm not even sure if the circuit I did below is in the correct direction. Someone told me instead of using self heating circuit I can rub it with my fingers. I've no electronic background thus my knowledge is limited, making me turn for help here.

amber3 said:

I may sound stupid but does anyone have a simple schematic or know how to build a thermal air flow sensor/meter using two pt100, resistors, variable resistor, multimeter & of course a voltage source?

Click to expand...

No. I could come up with a circuit for such a device if I wanted to devote the time, but it wouldn't be simple; and it wouldn't necessarily use the components you cited.

I want to see and build the working principle behind the thermal flow meter. I don't really get how it works. If anyone can help explain in simple terms that'll be great.

Click to expand...

If you Google the phrase, "hot wire anemometer" you'll get plenty of good explanations. But you touched on the basic principle right here:

One sensor is for reference temperature and the other one is heated to maintain the temp difference when the air flows.

Click to expand...

Air flow removes heat from the heated sensor proportional to the air velocity, requiring the circuit to deliver more power to the sensor to maintain its temperature differential above ambient; by monitoring the power delivered to the sensor, we can infer the air velocity.

I do know that you have to use a wheatstone bridge.

Click to expand...

Not necessarily. That's one way of doing it, but not the only way.

I'm not even sure if the circuit I did below is in the correct direction.

Click to expand...

It doesn't matter. If the circuit indicates in the wrong direction, simply swap the meter leads.

But the circuit you posted won't do anything at all, because it is incomplete: it lacks any mechanism for heating one of the sensors and controlling its temperature to a constant value above ambient.

Someone told me instead of using self heating circuit I can rub it with my fingers.

Click to expand...

I've no idea what that person could possibly have been thinking. None at all.

I've no electronic background thus my knowledge is limited, making me turn for help here.

Click to expand...

If you have no electronic background you are going to have EXTREME difficulty designing and building any kind of airflow meter. This is NOT a trivial task.

For whatever it may be worth, the following schematic shows a design for a very simple, crude hot-wire anemometer that operates on the principle you describe:


Its operation relies on the fact that tungsten, like platinum, has a positive temperature coefficient of resistance, and LMP1 serves as both the temperature sensor and the source of its own heating. With the component values shown, the circuit adjusts its output to always maintain LMP1's resistance at about 300Ω, which corresponds to a temperature of about 100 °C. It lacks a sensor for ambient temperature or the circuitry for maintaining a constant temperature differential, so it cannot be accurately calibrated; but it shows the basic principle quite nicely.

I built this circuit one day just for fun, and found it was quite sensitive to even slight drafts in the room. Very entertaining.

Take care when removing the glass envelope from LMP1; cutting it off or smashing it without damaging the lamp filament (and without getting cut by glass shards) is tricky.

Yeah airflow meters are real tricky. I looked into it a while back and even if you get it built you have to calibrate it somehow which is a lot of math and physics or a lot of testing.

In principle, it works by passing a known current or voltage through a temperature-sensitive resistance element which heats it up above room temperature by a certain, known amount. Then as air blows onto the element, it removes heat faster from the element which changes the temperature of the element which changes it's resistance. Due to the resistance change, if you are keeping current constant then the voltage must change and vice versa. You can use this to back-calculate the new resistance of the sensor which gives you it's actual temperature of the element. You then compare the actual temperature to the known temperature in still air and from that you can figure out the amount of heat loss which lets you know how fast the air is moving to remove that heat.

The hotter you run the element, the more accurate and easier the measurement is since ambient room temperature becomes a smaller contributor to the overall temperature of the element. It also increases sensitivity since more heat is transferred when the temperature gradient is larger so at higher self-heating temperatures, you will get a larger reading for a given airflow than at a lower temperature. This is useful for very slow airflows that won't remove much heat.

But I think if you just want to figure out how it works then all you have to do is run a constant current through an RTD (which heats it up and blow room temperature air at it at different speeds and measure the voltage across it.

Alternatively, you could put a constant voltage across it and measure the current to see how it changes when room temperature air is blow across it at different speeds.


You might need a really sensitive meter though since RTDs don't change in resistance very much. It might be more reasonable to use an NTC thermistor instead. But that makes it even trickier since thermistors don't change resistance linearly with respect to temperature (and even thermal airflow sensors using linear sensors like RTDs have non-linear outputs with respect to temperature). It's not worth it to build a bridge to figure all this out if you just want to see how it works.

The reason I started with bridge is because I used https://www.farnell.com/datasheets/318594.pdf and https://www.farnell.com/datasheets/1674304.pdf electronic circuit recommendation as my reference.

OBW0549 said:

It doesn't matter. If the circuit indicates in the wrong direction, simply swap the meter leads.

But the circuit you posted won't do anything at all, because it is incomplete: it lacks any mechanism for heating one of the sensors and controlling its temperature to a constant value above ambient.

Click to expand...

OBW0549 said:

If you Google the phrase, "hot wire anemometer" you'll get plenty of good explanations. But you touched on the basic principle right here:

Click to expand...

ah my bad I did not meant the current direction but the project as a whole as I'm not sure what to start with.
so it uses only one sensing element instead of two hmmm. most of the examples only gave bridge circuits. Are these incomplete also? (attached files)

OBW0549 said:

If you have no electronic background you are going to have EXTREME difficulty designing and building any kind of airflow meter. This is NOT a trivial task.

Click to expand...

yes I've been struggling for a month to do the proof of concept. T___T

OBW0549 said:

For whatever it may be worth, the following schematic shows a design for a very simple, crude hot-wire anemometer that operates on the principle you describe:

Click to expand...

Thank you for this! the LMP1 is a normal led I assume? How do I know what op amp to use? for this or for any other circuits

amber3 said:

Thank you for this! the LMP1 is a normal led I assume?

Click to expand...

No!!! It is a tungsten lamp-- an incandescent light bulb. Tungsten has a positive temperature coefficient of resistivity (see **broken link removed**) so if we measure its resistance we can infer the temperature. An LED is a diode, and would be completely useless.

How do I know what op amp to use?

Click to expand...

If I recall correctly, I think I used one section of an LM324.

Understood. Thank you for bearing with me, I'll read more and will try it out soon!

dknguyen said:

But I think if you just want to figure out how it works then all you have to do is run a constant current through an RTD (which heats it up and blow room temperature air at it at different speeds and measure the voltage across it.

Alternatively, you could put a constant voltage across it and measure the current to see how it changes when room temperature air is blow across it at different speeds.

Click to expand...

yes yes this is what I'm actually planning to do with the two rtds but have no idea how to do it cuz in principle, they compare with each other to keep a constant difference.
1) is <1mA okay for a constant current? 2) what constant range of voltage must I apply?

amber3 said:

yes yes this is what I'm actually planning to do with the two rtds but have no idea how to do it cuz in principle, they compare with each other to keep a constant difference.
1) is <1mA okay for a constant current? 2) what constant range of voltage must I apply?

Click to expand...

Both those things are completely up to you. They are determined by how hot you want to run the sensor. However, it is advisable to run the RTD has hot as you can to make things more accurate and easier on yourself so 1mA will probably will not be enough. Most RTD are not designed to be run in self-heated mode so you have to watch out for this (it just means they burn out easily if you drive too much current through them).

One way to get around this problem is to not run the RTD in self-heating mode. Run the RTD like normal (only minimal current just enough to get a reading) but wrap the RTD around a separate heater circuit that provides the heat.

About constnat current or constant voltage, you cannot have both, ever. You must decide which you want to be constant. If you decide you want constant current than the circuit automatically adjusts the voltage as necessary to keep the current the same. If you decide you want constant voltage then the circuit automatically adjusts the voltage to keep the voltage constant and the current ends up being whatever it will be. (The reason you dont adjust current to keep voltage constant is due to it being a lot easier with real-world components to directly adjust voltage than it is to directly adjust current).

dknguyen said:

You then compare the actual temperature to the known temperature in still air and from that you can figure out the amount of heat loss which lets you know how fast the air is moving to remove that heat.

Click to expand...

I'm planning to use constant voltage for now. Another question, ^ For example, if the known temperature in air is 27 and the actual temperature measured after conversion is 25 degrees Celsius, the heat loss is 2 degrees. How do you equate that to the air velocity or power used?

2C above ambient will be very difficult to work with (and by difficult, I mean impossible for all intents and purposes unless). I'd be aiming for 100C or higher just to make it easier on myself.

But the part about equating temperature drop to airspeed is far and away the most difficult part. There no simple equation or method. It also depends on the geometry of the heating and sensing element. You are best off calibrating the sensor empirically by blowing air at room temperature at the sensor at a known speed and recording the temperature drop and recording it all in a look-up table.

There are mathematical models that have been derived but only for specific geometries and they still require knowing parameters that are not directly measurable whose values are unique to the construction of your overall sensor head.

Here are some materials:
http://ocw.metu.edu.tr/pluginfile.php/1871/mod_resource/content/0/AE547/AE547_9_Hotwire-son.pdf
http://www-g.eng.cam.ac.uk/whittle/current-research/hph/hot-wire/hot-wire.html

There was one other website that I am having trouble finding again where a guy made a bunch of different sensors and tested them. It will show you what you are getting into, but I can't seem to find my way back to it.

amber3 said:

For example, if the known temperature in air is 27 and the actual temperature measured after conversion is 25 degrees Celsius, the heat loss is 2 degrees. How do you equate that to the air velocity or power used?

Click to expand...

You can't ignore the ambient temperature. This must be either be accounted for, or you must run so hot that the influence of ambient temperature is negligible. In your example, if ambient was 25C then airspeed would have to be infinite.

Found it:
http://www.fonema.se/anemom/anemom.html

He doesnt bother with equations (King's Law) at all since its not feasible for DIY. Straight empirical calculation.