Testbench for fans. Please help me create it.

[mk]

I am planing to test a lot of fans and publish the results on my website (and in this forum).

So the goal of this thread is to get help to create a good fan testbench that can provide all the information that is relevant to silent enthusiasts.

The manufacture is normally providing a lot of useful information. So before we start testing lets see what we already know.

- Physical specifications (size, number of fins, connections, open or closed corners, etc.)
- Voltage and ampere
- RPM
- CFM (can we trust this one?)
- Noise (further testing is needed here)
- Prize

Am I forgetting something?

OK so what is it we don't know but would like to know?

- The noise level at different voltages - 5, 7, 10, 12.

- The noise level at certain fixed CFM. This will make it possible to compare the noise levels of fans when performing the same regardless of different specifications. (this is the actual goal)

- Lowest starting voltage.

- Lowest spinning voltage (how low can you go...)

- RPM measure to complete the picture.

- Effects of PWM fancontrols. (is the fan clicking)

How to compare the fans

If the testbench can provide all the information above it would be possible to gather the information in a table and a noise / performance comparison would be straight forward.

In the table below the "fixed" cells are supposed to have the same number for all fans. All the blank cells should be filled out.



All the questions (please help me out here)

- What should the fixed CFM number be. I am thinking about finding them by using a reference fan running at 5, 7, 10, 12. This could be more than four readings.

- How do I measure CFM (this is the million dollar question)

- How do I measure RPM? The problem is measuring the lower numbers that motherboards don't read. I have been thinking about using a small magnet similar to those used to measure RPM with a bicycle computer.

- Can anyone recommend a simple PWM fancontroller.

That's all. Please comment my plans if you can help.

[cAPSLOCK]

For using a bicycle speedometer to measure rpm, I don't think it's a good idea, the magnet is heavy enough that it might make strange effects on the fan... The best way is probably with some stroboscopic method, which normally needs fancy equipment, but if you have a Lego RCX with a light sensor, you could put something white on one of the fan blades... well I haven't tried this even though I have the RCX, I'm not sure it the light sensor is quick enough, but worth a try...

For the CFM I only have a crazy idea that probably isn't very useful, but ok: if you attach a big tube to the end of a fan, and put the fan at a smoke source and see how long it takes to fill the tube with smoke, you could calculate CFM... Otherwise, once again you have to get the fancy kit, like an anemometer or something...

[mk]

For using a bicycle speedometer to measure rpm, I don't think it's a good idea, the magnet is heavy enough that it might make strange effects on the fan...

I think you're right... it was just a thought

For the CFM I only have a crazy idea that probably isn't very useful, but ok: if you attach a big tube to the end of a fan, and put the fan at a smoke source and see how long it takes to fill the tube with smoke, you could calculate CFM... Otherwise, once again you have to get the fancy kit, like an anemometer or something...

The smoke idea is probably too inaccurate, but it would look great in a video review!

I have been checking the anemometer idea and it seems that it is the way to go. This one claims to measure within the range of 0 - 999.999 CFM. The prize is $ 249. This could be a solution.

[jaganath]

Webmaster of StillePC.dk - silent computing in danish.

Good to know that silent computing is an international phenomenon. I think there's a www.pcsilencieux.fr as well? Best of luck with your fan measurements; this is a really promising project.

[Le_Gritche]

The manufacture is normally providing a lot of useful information. So before we start testing lets see what we already know.

- Voltage and ampere

As was pointed to me by Elixer on another thread :

What is generally listed on fans is the max current draw. You probably won't ever see the fan draw this much power unless you stall the fan at 13.2 volts. Also, fans have wide tolerances; I wouldn't be suprised to see a 10% resistance difference between apparently identical fans. The fan will need to be measured.

So I'm afraid you will have to measure the real current draw too. It's of use for people willing to put a resistor in-line with the fan to lower the RPM.

[EndoSteel]

mk

- CFM (can we trust this one?)
No.

- Noise
Useless, as well.

- Voltage and ampere
The latter is also not trustworthy, not to mention that starting, operating and maximum (locked rotor) currents differ.

- Effects of PWM fancontrols. (is the fan clicking)
I would also add "bearing noise" (for ball bearing fans).[/b]

[Aleksi]

Hi,

If we're throwing ideas out in the air, I would recommend a mylar bag over an anemometer. Anemometer's aren't really the best suitable choice for axial fans due to the lack of laminar airflow.

About PWM... would be a nice to have different freq options on it, just to determine how the fan reacts with different freq PWM instead of PWM with one single frequency.

I'd also recommend a strobe light for the getting the RPM figures, although not all like to look at blinking lights.

Those just as a few quick ones.

[cpemma]

Flow and pressure are very difficult to measure. Taking flow, the usual vaned anemometer must cause a resistance effect, unless it's far from the fan and then dispersal angle comes in.

Perhaps a "standard case", a large deep box with front panel for a fan and a rear moderately-long tubular duct to create a fairly laminar flow pattern leaving the box. A few inches diameter to reduce back pressure and wall effects. Then stick a hot-wire anenometer in the air stream

[mk]

Thanks for all the comments.

CFM

I am starting to realize that it will be very difficult to measure the exact CFM. There is a standard for measuring CFM but the test setup required is beyond my reach. It is called the AMCA Standard 210.

However I still believe that it is possible to create results that makes it possible to compare fans to each other.

My idea is to make a tube and place the fan in one end and an anemometer at the other end. If using the same setup for all fans a relevant comparison should be easy to make.

I still have some questions though.

- How will the volume of the tube affect the measure?
- Can the same tube be used for all fan sizes?

RPM

Can someone explain the idea with the strobe light a bit further.

- What equipment is needed?
- and how should the measure be performed?

Anemometer

There seems to be two different kinds.

- What is the difference between a vane and a hot-wire anemometer?
- Why would one be better that the other?

Please keep posting useful comments. You're really helping!

[mk]

No help here at all?

[Devonavar]

CFM is very, very difficult to measure accurately. I have no help there.

However, I can help you measure RPM.

You'll need three things:

A Video Camera, preferably one that can decrease shutter speed ("Sports mode" should work well)
Some paint
Editing software that will let you see individual fields.

The basic idea is to paint one of the blades of the fan white or some other easily identifiable colour.

Then, shoot a video of the fan.

Now, the tricky part. You need to separate the video signal into fields, so that you can see a "stop motion" recording of the fan. With PAL, you will have a field rate of 50 fields per second, which is enough resolution to capture motion up to 3000RPM (50 fps * 60 seconds)

If you're not familiar with video signals, a field is half a frame. Every frame is actually made up of two fields, which are interlaced with each other in the final frame. Since PAL is 25 frames per second, there are 50 fields.

Once you've laid out the fields in front of you, you can count how many rotations there are in a given time frame (say, ten seconds) and them multiply appropriately to get the result in RPM.

[IsaacKuo]

Painting the fan will add weight and that'll affect its performance. It may also affect balance.

Instead, just point a cheap laser pointer through the fan and take a video of the laser spot. Even better, shine the laser spot onto a photoresistor and hook that up to an audio card's mic input. Then you can record the signal and determine its period with audio software.

[IsaacKuo]

Can someone explain the idea with the strobe light a bit further.

The idea is to run your fan in a dark room with a strobe light. This strobe light flashes with a regular beat, at an adjustable frequency. You adjust the frequency with a knob until the fan appears to be moving still. At that point, the period of the strobe light is equal to a multiple of the period of the fan blades (which is the fan's period divided by the number of fan blades).

[Tibors]

and determine its period with audio software

or with Oscilloscope for windows.

Which might even work directly with the RPM signal, for fans that have them.

[Rusty075]

I think you may be over thinking things a bit. There's no need to reinvent the wheel when there's already a wealth of tools available to test these things.

RPM:
Buy a laser tachometer for $50. Even a cheap one will have the range and resolution necessary for this sort of testing.


CFM:
Anemometers will never get you satisfactory data, they're just not designed for this. Instead, why not build yourself your own AMCE210 testing rig? All you need is some lumberyard materials, a blower (just a fan that is more powerful than anything you'd be testing, installed with a variable speed controller), and a pair of air pressure meters (or one differential pressure meter). You can find used laboratory grade equipment on ebay pretty cheaply.


For PWM controllers look at some of the Matrix Orbital LCD units. They have PWM features with programmable frequency controls.

[QuietOC]

Thanks for all the comments.

CFM

I am starting to realize that it will be very difficult to measure the exact CFM. There is a standard for measuring CFM but the test setup required is beyond my reach. It is called the AMCA Standard 210.

CFM: Why not just time how long the fan can fill up a large inelastic bag? I think someone mentioned mylar, but a normal plastic bag should be accurate enough. If you figure out the volume of the filled bag it you could even get a real cfm number.

[nomoon]

I would vote to have curves for CFM vs SPL. I want the best cooling (most CFM) with the lowest noise. If I'm using a fan controller so that I'm running the minimum CFM needed to cool my system, I want to know how much noise a fan will make. I don't care how many volts I'm using.

I recently posted at message here which showed graphs made using SPCR data. That small bit of data suggests that all 120mm fans create roughly the same amount of noise when run at low CFM, even for a wide 38mm fan.

Jason

[jaganath]

Why not just time how long the fan can fill up a large inelastic bag? I think someone mentioned mylar, but a normal plastic bag should be accurate enough. If you figure out the volume of the filled bag it you could even get a real cfm number.

Um, this might sound like a dumb question but how can the fan "fill up the bag" in this scenario? Won't the bag already have air in it (after all, you're not proposing to create a vacuum in the bag first, right?)? How will you know what proportion of the air is from the fan and how much was in there to begin with?

[Captain John]

The two places I've seen needs to measure airflow accurately are
a) my ENT doctor (for asthma)
and
b) my car, in the form of a Mass-Air-Flow-Sensor (MAF).

The car (any Bosch MAF car) has an output from a MAF sensor of between 1.7v and 5v. Now, each MAF is different, but it strikes me that if you had a known fan, with a known rate of flow, you could determine the mass-flow at various voltages. Mass flow is different than CFM, as it takes into account the temperature and air density, and comparing CFM at different temps & air pressures would yield an inaccurate view of flow.

The doctor gave me a goofy device to measure airflow, I think it measured in liters/minute, and had an upper range of ~1000. I don't think this would be useful in this particular scenario.

MAFs, on the other hand, can be had for ~$80, although calibration would all lie on your shoulders. If you're a bit more daring... you can pull the existing MAF from your car...

-John

[mk]

Thanks again for all the comments.

Buy a laser tachometer for $50. Even a cheap one will have the range and resolution necessary for this sort of testing.

Thanks Rusty! This is by far the easiest way to measure RPM. I have ordered a laser tachometer. Found one at $17 on ebay!

Anemometers will never get you satisfactory data, they're just not designed for this. Instead, why not build yourself your own AMCE210 testing rig? All you need is some lumberyard materials, a blower (just a fan that is more powerful than anything you'd be testing, installed with a variable speed controller), and a pair of air pressure meters (or one differential pressure meter). You can find used laboratory grade equipment on ebay pretty cheaply.

Again you hit the nail. Instead of working on half solutions why not go all the way.

I have checked the prizes on air differential pressure meters and they are not to expensive. Found this on ebay. Will it be good enough for my needs?

I do have a question for the setup: What is an auxiliary blower? Maybe I just don't understand this English word. I tried to google for it and everything from big ships engines, car engines and windmills turned up??

Also any information on how the AMCA box should be scaled would be useful. I am thinking about building it in plexiglas. That would look nice.

If you know of or stumble upon any building guidelines for the test setup please let me know.

For PWM controllers look at some of the Matrix Orbital LCD units. They have PWM features with programmable frequency controls.

I will check that out.

Thanks again to you all!

[QuietOC]

Why not just time how long the fan can fill up a large inelastic bag? I think someone mentioned mylar, but a normal plastic bag should be accurate enough. If you figure out the volume of the filled bag it you could even get a real cfm number.

Um, this might sound like a dumb question but how can the fan "fill up the bag" in this scenario? Won't the bag already have air in it (after all, you're not proposing to create a vacuum in the bag first, right?)? How will you know what proportion of the air is from the fan and how much was in there to begin with?

Mountains out of molehills!

It is rather easy to empty air from bags (how about turn the fan around and suck all the air out first?), and they normally come completely empty when you buy them. How much do several hundred large plastic trash bags cost? I've never seen them come with air inside (at least not anything that would throw off this type of calculation anyway.) They all hold the same volume if you can come up with a consistent way of attaching them to the fan (shouldn't be too hard.) Sure this method might only give a couple digits of accuracy, but this isn't rocket science.

This method would be quite accurate and more reliable than a MAF or any of the other flow sensors mentioned. All of those measure flow indirectly and require calibration. How do you measure the flow of a water pump? The most accurate way is to pump a known volume of water and record the time. Same thing works with air...

Spend the $$$ on a MAF if you need an good instantaneous guess at CFM, like say you need to know how much fuel to squirt into a combustion chamber. I don't see that that is what is needed for fan CFM measurments.

[cloneman]

I'm curious to know what your going to use a 12V source to power the fans. I've settled on a broken hard drive enclure.

[Rusty075]

Again you hit the nail. Instead of working on half solutions why not go all the way.

I think we all tend to overthink things some times.

I have checked the prizes on air differential pressure meters and they are not to expensive. Found this on ebay. Will it be good enough for my needs?

I'm no expert, but that looks like exactly what you'd need.

I do have a question for the setup: What is an auxiliary blower? Maybe I just don't understand this English word. I tried to google for it and everything from big ships engines, car engines and windmills turned up??

It's just a powerful fan whose speed you can control. It's purpose is to control the pressure difference between the chambers, so you can test at different points in the P/Q curve. Turn it down low and you get more backpressure, up high you get less.

If you know of or stumble upon any building guidelines for the test setup please let me know.

I went looking for details on the AMCE 210-99 spec and didn't find much. AMCE will gladly sell you all the specifics for $60 from their website though. If you have access to a good public library, or better yet a university library, they might have it right there on the shelf.


There's a discussion going on at procooling regarding this exact topic right now. Bill and Ben have already pulled together some of the tech. documents you're looking for: BillA's Blog at Procooling In particular, take a read through the linked Flow Bench Manual PDF

Or, if you wanted to save some work you could just buy a complete AMCE210 chamber, preassembled: for only $5000

[mk]

I'm curious to know what your going to use a 12V source to power the fans. I've settled on a broken hard drive enclure.

I am going to use a small universal power supply that goes from 3.5 to 14 volt. I will be using a potentiometer and a voltmeter to get the power I want.

There's a discussion going on at procooling regarding this exact topic right now. Bill and Ben have already pulled together some of the tech. documents you're looking for: BillA's Blog at Procooling In particular, take a read through the linked Flow Bench Manual PDF

Now this is useful. I will put the project on rest for a while and see what they come up with over at procooling. Thanks for all the help Rusty. I am sure I will end up with a great testbench thanks to you.

I will let you all know how the project turns out. I will probably start out doing only voltage, RPM and sound measurements and then add CFM when the testbench is complete.

By the way I just received my new (old) Sound level meter. It is the same type that Mike uses but a newer model (2209). From the pictures on this site it is a bit hard to see how big this thing actually is. So here a shot. It's a 120 mm Yate Loon next to the sound level meter.

[frankgehry]

That's an impressive looking piece of equipment. Do you plan to calibrate it or use a specific fan as a reference noise source?

[mk]

A calibration device was a part of package among many other things when I bought the sound level meter. The calibration device acts as a sound source and produces exactly 94 Db at 1000 MHz. A special barometer is used to correct for different airpressure. This is really good stuff.

I plan to calibrate it every time I run a test.