Find RPM of a fan by recording a video

[SentinelAeon]

I have a table fan and i need to compare it's rpm in 2 scenarios. Its a big fan with 3 blades. Since i don't have any fancy tools or specs, i thought i could get an approximation by using mobile phone and a stop watch.

My plan would be painting a black dot on 1 of the blades. Next to a fan, i would put a stop watch and start it. I would point my mobile phone at fan and stop watch and record a video with my phone at 60 frames per second. Then i would playback the video at a slower playback speed. I would skip frame until i see the blade with black dot pointing upwards. I would note the time on stop watch. Then i would slowly play forward and count how many times the dotted blade is pointing upwards. I would stop when 5 seconds have passed on a stop watch. Then i would multiply the number i got by 12.

The reason i included a stop watch in this plan is to make sure that my mobile phone really is recording at 60 frames per second and not doing something wrong. If 5 seconds passed on stop watch also means 5 seconds passed in my video, i can be sure its all ok. If i wanted a better result i would measure for more than 5 seconds.

Would this be a good way to get an approximation without special tools ?

 

[jrmichler]

Normal video speed is 30 frames per second. You can check this by taking video of a digital clock that shows seconds, and counting frames in the video for one second. A typical table fan will run at about 1500 RPM, or 25 revolutions per second. A fan running at that speed will make 25/30 (0.83) revolutions per video frame, or 5 revolutions in 6 frames.

This requires software that allows you to step through the video one frame at a time. Better software will count the frames for you. Many mobile phones have "sport mode", which is (I think) 120 frames per second.

 

[berkeman]

Would this be a good way to get an approximation without special tools ?

You might be able to download a free audio spectrum analyzer app for your phone that would show you the frequency spectrum of the sound from the blades. That should be able to tell you what the fundamental frequency of the blades passing the phone microphone is (hold your phone near the edge of the fan close to where the blade tips are going by).

 

[Rive]

Maybe you could get some data from the blur by trying different shutter settings too (if your camera supports manual shutter settings).

 

[sophiecentaur]

You might be able to download a free audio spectrum analyzer app for your phone that would show you the frequency spectrum of the sound from the blades.

I was thinking along those lines. There are phone apps that measure frequency. You may be able to turn the rough LF sound of the blades into a series of clicks, using a small piece of card with the blades striking it (A bit like the way we used to modify our bicycles to make motorcycle noises, using a folded up piece of cardboard jammed into the brake bracket) . You could find if the drag of this clicker has an effect by varying the pressure and then finding the least pressure that will give you detectable clicks.

There are several apps available and they're all try before you buy so the best one can be picked.

 

[SentinelAeon]

Thank you for all the replies. I didnt get a chance to do a full test yet, but i did a quick one just to see if this is even possible.

I used a mobile phone that can record at 240fps (Samsung S7 edge), pointed it at the fan and the stop watch next to it. I only did a 10 second recording and actualy only counted the frames for 1 second. I counted 6 turns in 1 second, which would indicate 360 rpm. Which would make sense since i am only supplying about 10W into a fan that has 48W motor. I cropped the image so you cant actualy see the stop watch running next to fan. I converted it into animated gif, i hope this forum supports it. What you see in a gif are steps frame by frame, i used vlc for that, i could also use a 0.03% speed in it or just extract all frames from a video and check them, though using vlc is easier.

Now to get a more precise measurement i will obviously have to do a measurement for like 10 seconds instead of 1, but for starters it is ok. And besides, it is not the raw number of frames that i am interested in, but rather comparison between 2 modes A and B and what i want to know is which of them has higher rpm. Unless i am missing something i should be able to do that, maybe even at higher rpm, judging by how many frames i record before fan makes a full turn, i should be able to measure this at 1000rpm or even higher. Again, unless i am missing something big here and the whole process is false.

 

[pbuk]

unless i am missing something big here and the whole process is false.

No, it all looks good to me. The high frame rate of 240 fps obviously helps here.

 

[SentinelAeon]

No, it all looks good to me. The high frame rate of 240 fps obviously helps here.

Yeah, its great. Though it kept bugging me to try to use 60 fps and see how high i can get and compare the result with 240 fps. Because at 240 fps, i count 40 separate frames per rotation of the rotor. But i dont need 40 separate frames. I actualy need only 1 per full rotation - to see whether the dotted blade crossed the top position. If i do a long enough measurement, like 30 or even 60 seconds, i will be off by less than 1 rotation, which is a small fraction. So if a fan speed is 1000rpm, that means it turns 16.6 times a second. So i need to record it with at least 16.6 frames per second, obviously, to record each rotation. I think ?

 

[pbuk]

So i need to record it with at least 16.6 frames per second, obviously, to record each rotation. I think ?

Not really; you could record at 10 fps and then you would expect it to complete exactly $1 \frac 2 3$ revolutions each frame. But note that you could easily mistake 400 rpm for 1000 rpm like this (why?). And the faster you are, the more precise and accurate you can be. But how precise do you want to be?

 

[SentinelAeon]

Not really; you could record at 10 fps and then you would expect it to complete exactly $1 \frac 2 3$ revolutions each frame. But note that you could easily mistake 400 rpm for 1000 rpm like this (why?). And the faster you are, the more precise and accurate you can be. But how precise do you want to be?

Yep, i didnt account that when fan is faster, it moves more each frame therefor completes 1 circle in fewer frames than it does at lower rpm. It was theoretical anyway, i have 240fps mode so i will record at that. I was just wondering how low i could go. As for precision, knowing the actual value of rpm is nice, but what i actualy need is comparison between A and B, for instance, A is more rpm than B. Knowing how many rpm or % more is a bonus.

If anyone is wondering why i am doing this .. my fan has 3 speed settings. Difference in speed between setting 1 and 3 is drastic. But the actual power draw difference between 1 and 3 is not much. 35W vs 48W if i remember correctly, while rpm feels like 500 vs 1500 rpm. So my theory is that it is using some kind of resistor to lower rpm and part of power usage actualy gets wasted as heat somewhere. Also i have an AC step-down since even setting 1 on fan is to fast for what i sometimes want. So now instead of 35-48W range i have 0-48W range. But its also known that step down by lowering voltage is not optimal for such AC fan, you can even hear strange hum. So what i am trying to figure out is this: A -> pressing setting 1 vs B -> pressing setting 3 and using step down until fan uses 35W and then compare if both settings give me same rpm or which one gives me more rpm at given power. It doesn't serve any practical thing since i will continue to use step-down no matter what. But i really was wondering how efficient is step-down since i cant detect almost any heat coming from it. Also i would like to know how power usage in W compares to rpm. I am sure it cant be linear, like 5W = x rpm, 10W = 2x rpm, etc. I want to plot a graph W vs rpm.

 

[sophiecentaur]

Though it kept bugging me to try to use 60 fps

Why? Here's some theory: Under sampling can produce 'alias' components (see figure 2 on the link), such as the wagon wheels appearing to move backwards in old cowboy films.

If you want to sample a signal that changes at a rate of f then you need to sample at 2f to avoid any aliasing. 1500rpm with three blades will produce a rate of 44500 or 75 Hz. With one marked blade, the frequency drops to 25 Hz and that is just marginal. Clearly, sampling at a higher rate would tidy things up and make your counting method more reliable. Do you want to do many measurements?

PS if you can vary the speed over a continuous range (slowing up the spindle bu pressing a pencil against it) if your sample rate is ok then the recorded speed will drop. If you sample too slow then the blade may appear to speed up.

 

[SentinelAeon]

Why? Here's some theory: Under sampling can produce 'alias' components (see figure 2 on the link), such as the wagon wheels appearing to move backwards in old cowboy films.

If you want to sample a signal that changes at a rate of f then you need to sample at 2f to avoid any aliasing. 1500rpm with three blades will produce a rate of 44500 or 75 Hz. With one marked blade, the frequency drops to 25 Hz and that is just marginal. Clearly, sampling at a higher rate would tidy things up and make your counting method more reliable. Do you want to do many measurements?

PS if you can vary the speed over a continuous range (slowing up the spindle bu pressing a pencil against it) if your sample rate is ok then the recorded speed will drop. If you sample too slow then the blade may appear to speed up.

Yeah, i wanted to do it as a challenge to see how high a rpm i can "count" with only 60fps. But as a reliability of course i would use 240fps. I would use more if i had handy, though im kind of impressed that phone really is making 240 pictures a second, not just recording at like 60 and adding "in the middle" fake frames. I will come back to report if i get same/less/more rpm Watt for Watt by using step-down. Curiosity is what got me started with this in the first place :)

 

[sophiecentaur]

Yeah, i wanted to do it as a challenge to see how high a rpm i can "count" with only 60fps.

"Challenge" the Nyquist criterion? That's a big ask lol. There are two aspects to this. You can over sample greatly and be sure you are seeing what's there or you can use low sample rate and look for anomalies in your results. If you want to look at as short a burst as possible then over sampling would be the way to go. I'm wondering how you would plan to find out what you can 'get away with'. You would need more low sample rate samples as your measurements will probably be moire blurry(?).

Under sampling can give you speed and direction errors (wagon wheels again). Sampling at 60 Hz is OK for your lowest speeds but, for higher speeds, I'd expect some artefacts. Incidentally, the best filming conditions would be against a uniform (say black) background and you may identify the edges of the blades better. It looks as though your fan vents to the outside. Maybe night time would be a good time (with good illumination from your side. Probably could be with the local mains frequency, producing beat patterns, though.

Looking at your short video, I can see some odd artefacts (spots) on the blades as they go through the 1 o'clock position. I think they could be sampling artefacts Try with a high and a low sample rate and spot any differences. Have fun.

 

[SentinelAeon]

Again a quick test. This was indeed harder than at slower rpm. When moving frame by frame by clicking, it was actualy very hard because the movement wasn't the same in everyframe.. every few frames it seemed to jump, or maybe at times it even looked like it skipped back, could be the ghosting, you were talking about this. So actualy playing at 0.03% speed (lowest vlc allows) it was a lot easier. Though i have to say that the recording was really bad. Timer was barely visible due to wierd angle of recording and lightning conditions. And at this speed, i think i will need to paint 1 of the blades black or white. Right now i only detect which blade it is because it has some hot glue and yellow duct tape on it and i see 1 of the blades has darker spots because of it. Nevertheless, i got what i wanted and i am happy to report that i was right.

1) If i manualy set speed to 1/3 with button on my fan, it will use 37W and do about 720rpm.
2) if i manualy set speed to 3/3 with button on my fan but then use step-down to actualy lower the voltage i am supplying to fan, i get 37W and fan will do about 900rpm.

This also makes sense since you can actualy hear the fan is spinning faster, but not because of the sound of the motor but sound of the airflow.

So clearly some power is getting wasted and those buttons are a bit like using a resistor to limit current - yes, fan will spin slower, but also the resistor will heat up. Also the fact alone that 1/3 setting uses 37W and 3/3 setting uses 48W, but the difference in speed between the 2 is like 1:2 at least, so you would expect something like setting 1/3 using 24W and setting 3/3 using 48W.

Before i go ploting Watt vs rpm in 5W increments i have 1 final question. Is there any data about relation between rpm and airflow. I am guessing its not linear due to losses at higher rpm due to friction forces ? For instance, if 500rpm gives you airflow of X and static pressure of Y, then for 1000rpm airflow is less than 2X and static pressure is less than 2Y. Like with cars, double the horse power gives you a lot less than double the speed/acceleration.

 

[jrmichler]

Is there any data about relation between rpm and airflow.

Yes, there is. Search fan laws to learn about this. The relation is as follows:

Air flow is proportional to RPM. Twice the RPM equals twice the air flow.
Torque to spin the fan is proportional to RPM$^2$. Twice the RPM equals four times the torque.
Power to drive the fan is proportional to RPM$^3$. Twice the RPM equals eight times the torque power.

You will not see this in the tests on your fan because the motor and speed control has extremely low efficiency. Various electrical and friction losses are much larger than the actual fan power. You might find it interesting to do a test with the fan blades removed to find just how much power is used to spin the motor alone.

Edited to fix an oops.

 

[sophiecentaur]

If i manualy set speed to 1/3 with button on my fan, it will use 37W

What sort of meter are you using to measure the AC Power input? Without a true AC Power meter, you cannot be sure about the power in / speed curve. AC motor characteristics are very complicated; the only successful / accurate form of speed control is with a variable supply frequency (which is basically how a brushless motor works. I don't think ordinary desk fans use those because they (and the power supply) are expensive.

You have two hard problems here. Despite the fact that you want to use a video camera to measure the rpm, I seriously suggest that you consider doing the speed measurement with audio. Put a paper 'clicker on each blade and record the sound. Find an App to give you the sound spectrum and you should find a peak at the repetition rate of the fan blade clicks. No sampling problems and the software does all the processing for you.
No reason not to use your camera as you could compare the results of the two methods to get some corroboration.

Paint the fan blades black and hang a white sheet outside to mask the 'interesting' background cos that's another unknown quantity. AS I said earlier, lighting from inside the room could be a problem due to lamp flicker. Do you have a good bright torch? That could do it.

 

[SentinelAeon]

I will check fan laws, thank you. Though it's good to hear that double the rpm means double the airflow. This at least gives me some feeling of what i am getting for the "buck" (Watt).

I am measuring the fan power usage with killawatt thats plugged into a socket and fan plugged into killawatt. In my case it's not about how accurate the actual reading is, rather how accurate comparison between 2 measurements is, in my case 37W with 1/3 setting vs 37W with step-down. If this 37W is in reality 30W doesn't really matter, as long as it's then 30W both both settings. Is sampling rate of killawatt to slow to give accurate measurement with table fan ? Killawatt offers a few options, like measure current (A), measure power (W), measure Hz, measure kWh over long period and a few others.

As for fan rpm i agree - changing voltage is not optimal, which i read online, and you can even hear strange hum coming from fan when you lower it to like 10-20W. But it seems to be working alright considering there is almost no heat lost in step-down circuit. And for my usage, 37W is far to much for most of the days, where i just want very light airflow but keep the option to run fan at 100% when i need to.

 

[jrmichler]

. And for my usage, 37W is far to much for most of the days, where i just want very light airflow

Take a look at a low speed muffin fan, such as this one: https://www.digikey.com/en/products/detail/gelid-solutions-llc/FN-PX14-11/16714422.

These style fans are used for cooling electronics. This particular one is rated 21 dB(A), which is so quiet that few people could hear it in a normal room. It's not the only possibility. It's rated to draw 0.09 amps at 12 volts, or 1.1 watts. The square fan housing allows you to set it on a table without a special mount or stand.

You can find a selection by searching the web site for fans, desired voltage, and RPM. The low RPM fans are the quiet ones, the larger ones move more air. It's also available from Amazon: https://www.amazon.com/dp/B09QT8MMDW/?tag=pfamazon01-20.

 

[Tom.G]

You can not necessarily believe a Kill A Watt device at low currents and strange waveforms.

I once tried to measure a battery charging power by measuring the input power to the charger.

The following was a few years ago when a neighbor bought a battery powered vacuum cleaner that kept dying, even after a couple trips to an "authorized service shop." (numbers below are approximate due to elapsed time!)

The charger was rated about 3A at 18V, but was defective and putting out around 12V, not able to charge the battery at all! (the battery voltage was around 13V)

But the Kill A Watt showed a power difference from no load to full load of around 30W. I even tested with a resistive load instead of the battery with similiar results.

So, yes, the Kill A Watt is useful -- just don't bet anything valuable on it's accuracy under strange conditions. (maybe it is waveform dependent)

Cheers,
Tom

Addendum: The charger open-circuit output voltage was very close to a common reference voltage used in voltage-regulator ICs, which led me to believe that the voltage divider in the voltage feedback circuit had an open or missing low-side divider resistor. The charger was returned-to-manufacturer and a replacement received which solved the problem.

 

[sophiecentaur]

take a look at a low speed muffin fan,

Would the airflow be anything like enough for a room? Surely everyone would use them if it was enough for venting rooms. Also, extractor fans need to cope with varying pressure differentials as conditions change.

 

[sophiecentaur]

Though it's good to hear that double the rpm means double the airflow.

However, that assumes the same pressure differential. That may change over time so you'd probably need some control for different conditions.
I still ask why a method using the sounds of the fan doesn't appeal to you. It would be a very quick job, once the App is installed and you have learned to use it.

 

[Averagesupernova]

You should be able to get away with a lower sample rate if each fan blade has a different mark and it is a guarantee the fan always spins in the same direction. However this is not ideal. Just doing this as a challenge is one thing. However as others have said doing it through video isn't the most practical way.

 

[SentinelAeon]

I might do the sound variant in the future.

As for pc fans, that would obviously be fantastic, i have probably about 100 fans with sizes between 4 and 14cm. Sadly there are a lot of nights when i enjoy running the fan at max over night and due to configuration on my apartment, it really makes a huge difference in temperature when running a big table fan, i think its like 40cm. I could stack a few 140mm fans together but due to multiple rotors that would certanly lose any energy efficiency advantage over big AC fan. Since fan is blowing out of the window in my storage room, i can just slightly open windows in all other rooms, including bedroom and i get huge airflow there with zero noise. I am finishing my mini esp8266 device that will measure inside and outside temperature/humidity to automaticaly start and stop the fan when it makes most sense to run it, taking lowering temperature and humidity in apartment as first goal. Apart from temperature/humidity i use this fan to improve odor. This fan obviously is just a temp solution since its a table fan not meant to be mounted on a wall hole and its energy efficiency is probably not very good. I have a few ideas of what i want and how to accomplish, though this setup right now is doing just good enough job to not make me wanna do anything new in a hurry. I really like fans, have a bunch of them around apartment doing various tasks 24/7 :)

 

[sophiecentaur]

I really like fans, have a bunch of them around apartment doing various tasks 24/7 :)

Maybe, if you turned all of them off, they'd stop supplying heat to your apartment.lol

Have you tied up all the other loose ends of heat getting into your apartment? When it's hotter outdoors then it's often worth while closing all windows and drawing all curtains (especially on the sunny side). Once the sun goes away and the outside temperature starts to drop, get some ventilation going and encourage the heat inside to leave through open windows. Note which way the wind is blowing and provide a through path for air until next sunup. These measures are all free, too.

 

[SentinelAeon]

Only 1 fan is for heat purposes and it only works when its cold outside, during the day like you said, closed windows and curtains down. All other fans together normaly use about 10W, but it's totaly worth it. Im lucky in this regard, this fan is enough to keep things nice and cool without AC. And in case of global warming i always keep a 50W diy mister in closet that will get you pneumonia faster than AC for backup I do this things mostly for hobby :)

 

[sophiecentaur]

this fan is enough to keep things nice and cool without AC.

That's good to know. I'm sure there are millions of other people who could do very well without AC. You clearly have the right idea about this.
Enjoy the experiments.

 

[jrmichler]

i have probably about 100 fans with sizes between 4 and 14cm.

Since you like fans, here's an article reviewing a few: https://www.treehugger.com/best-fans-5092513.

Also, this is a pair of 250 mm high speed fans that I had laying around. They were too noisy for use as room air circulators, so I wired them in series. That slowed them down to a speed where they move air with almost no noise. I have not used them in years, so I think they will be taken to the dump soon.

 

[SentinelAeon]

Those look like something i would put on a shelf as a decoration. Real nice :)

I have another question and i decided to first try ask it here before opening a new thread, hopefuly someone will read it. It's real simple.
From time to time i will read some article about new computer fans, mostly interested in how physics is used to further improve them. One of the things i read was that some news speakers are made from some new/better plastic material that is harder and more expensive. So from what i read, what this is supposed to do is that at high rpm, the blades of the fan will change their shape less than they do with cheap plastic. And because of that, they can put blades closer to the fan case and therefor slightly improve airflow.

Now how much this is noticable and how much its market material i dont know. But it got me thinking about my fan i use on a wall. It's a table fan that is obviously not made for what i use it for but its doing an ok job. But when i installed it i noticed 1 wierd thing. If i use a piece of paper or smoke to figure out where air is going, if i put it behind the spinning blades, air and smoke will obviously be drawn out from the apartment. But as you can see on a picture, there is a big gap between blade's edge and the wall case. And there, the air is actualy blowing into the apartment. At first i thought that was just some turbulence or something but now that i think about it, it actualy makes sense. As the blades push air out, this gap is the easiest way to draw air in. That is actualy not what i want - i want as much air as possible to be drawn through the window in the other room.

So i am wondering, is there any way to approximate how much airflow/pressure is lost due to this gap, if i measure the fan blades and this gap ? I am not looking for exact number but it would be good to know whether 5% is lost (which is ok) or 30% is lost (in which case it would make sense to make a wooden frame that will move closer to fan blades)

 

[sophiecentaur]

how much airflow/pressure is lost due to this gap,

I think it's more complicated than that. A fan running in a duct can produce more air flow than the fan alone Think about all the various designs of aero engines depending on the speed you want the aircraft to travel and what efficiency you require (and the acceptable noise level), they use different widths of duct and different lengths. Then the shapes of the blades has an effect. So I think 'it all depends'. Aerodynamics is a very hard subject, too.

 

[jrmichler]

There is no simple easy way to get a good measurement of the percentage of air back flowing. But you can get a good idea by generating a thin stream of smoke that gets sucked into the fan. Just be aware that the percentage of back flow depends on the back pressure. If the fan is pushing into the wind, or trying to suck air out of a closed room, there will be more back flow. The back flow could range from near zero to 100% depending on the back pressure.

 

[SentinelAeon]

There is no simple easy way to get a good measurement of the percentage of air back flowing. But you can get a good idea by generating a thin stream of smoke that gets sucked into the fan. Just be aware that the percentage of back flow depends on the back pressure. If the fan is pushing into the wind, or trying to suck air out of a closed room, there will be more back flow. The back flow could range from near zero to 100% depending on the back pressure.

But since this back flow is always occuring since there is such a wide gap, it would probably make sense to make the gap smaller since it takes mere minutes to make a wooden frame. It's hard to judge by hand since air blown into hand is felt a lot more than air being pulled away from hand. If i put a hand behind the fan, i barely notice anything but if i stand besides a window in the other room, its almost like a fan was blowing towards me. Do you have any idea what might be a good way to make such smoke to see ? A single cigarette isnt enough to see at this high speed and mist maker module is not good for the same reason - airflow is so high that it disapears into thin air as soon as you put it behind or infront of a fan.

 

[Tom.G]

For smoke generation, see this for ideas:
https://www.physicsforums.com/posts/5863664/

Or actually the whole thread addresses a similiar situation:
https://www.physicsforums.com/posts/5863327/

And of course:
https://www.google.com/search?hl=en&q=smoke+bombs

Cheers,
Tom

 

[Algr]

Your principle is sound.

I'm not aware that any camera would "Fake" 240 frames per second by shooting at 60 and interpolating in the camera. That would completely moot the point of having a high frame rate! I think photographers would make a fuss if some manufacturer tried that. Interpolation in post makes more sense, since the user could judge when it was appropriate and when it wasn't. I've seen some GoPro demos that had interpolated frames, but I figured they were taking the advertised 120fps and playing it back at 15fps. (1/8 speed)

The timing from your camera should be exactly what it tells you. The only variation might be that 60 is often 59.97, and 30/120 might follow that pattern. 50 fps should be exactly 50. If you are seeing uneven progressions of frames, that might be on the video player.

I disagree about shooting at night. The more light you have, the better, as this will cause a shorter shutter angle and less motion blur. Your photos are backlit. That would make your clock hard to read but in your case that shouldn't be a big deal otherwise. (I'm assuming artistic issues are relevant in this case.)

 

[sophiecentaur]

I'm not aware that any camera would "Fake" 240 frames per second by shooting at 60 and interpolating in the camera.

Dunno if it's available in smart home cameras but a posh TV receiver does Motion Smoothing when it upscales the frame rate. However, the Nyquist limit still applies when there are long lasting bursts of high temporal picture frequencies ( repeating fan blade positions etc.) The aim, in all this processing is to produce an overall best result and sometimes things have to be impaired or blurred in order to make a pretty picture. Jerky motion is annoying so that's dealt with when possible. (The smoothing can be switched off, afaiaa)

Modern image processing has to be very clever when you want to get many HD programmes down a limited bandwidth.