How to dynamically balance EDF unit with a cell phone



Introduction

We all heard this – loud whining sound of unbalanced EDF unit. There even were cases when pilots were banned from RC clubs for loud noise their EDF jets are making. The first advice people are usually getting is to balance a fan. There are magnetic balancers on the market that can do that job. But even if fan is perfectly balanced it still makes noise and vibration is all but gone. Reason for that is unbalanced rotor of the motor. One can choose a difficult path of balancing motor’s rotor but there is easier way to achieve smooth, quiet, vibration free running of EDF unit.
This article will describe step by step process on how to dynamically balance an EDF unit at your home. Before we dive into details a little theoretical explanation will help understand the nature of vibrations in EDF unit and what’s need to be done to eliminate it.


Theory

To achieve our goal we need to look at EDF unit as a one unit of motor and impeller together.

This picture shows typical distribution of centrifugal forces in EDF unit.



Sum of two forces is not equal to zero and this is what causes vibration. To stop vibration we need to make one force cancel other force. Here are two steps to do it.

Step 1: Align force vectors along one line, i.e. find fan’s position where heaviest point on the fan corresponds to the lightest point on the motor’s rotor.





Step 2: Add weight to the lighter side to equalize forces, i.e. make fan’s heaviest point as heavy as motor’s rotor heaviest point. The sum of two forces will be zero or in other words two forces cancel each other resulting in no vibration unit run.





Practical implementation

To prove this idea works I needed something that can measure vibration level and give me a number so that I can compare levels in my experiment. The easiest solution I could find under my hand is a smart phone. Latest smart phones like iPhone and Android have accelerometers in them. This is a perfect tool to measure vibration level. There are several programs available on the market that will collect accelerometer data and produce output file. I purchased Sensor Insider Pro for my HTC Incredible running Android operating system for less than two dollars. This program will export collected data along all three axis to csv formatted file I can use in Excel for analysis. This is all I need.

So here is my setup:

1. Cell phone attached to the EDF unit with two rubber bands
2. Laptop with Microsoft Excel

Step 1

This is what we need to do to complete step one of the dynamic balancing.

a) Mark motor’s rotor with a sharpie
b) Mark fan at the same line as motor’s mark, i.e. top of the motor is top of the fan
c) Start sensor’s data recording
d) Gradually increase motor’s RPM to the max and back during approximately 10 seconds interval
e) Stop recording and save result file.
f) Rotate fan to approximately 45 degrees. Direction is not important as far as you rotate in the same direction each run.
g) Repeat steps c-d-e-f until you reach angle of 315 degrees, which will be the last measurement point.

All this should take about 10 minutes or so. There is no need to be precise at this point, few degrees more or less is not that important. Let’s export collected data to Excel, make a few calculations and build a spider chart to look at the result.




By looking at the temperature diagram and spider chart we can say that minimal vibration level is 1.18 at the angle between our original motor and fan mark of 315 degrees.

So what this information means to us? It means that at 315 degrees centrifugal forces of the impeller partially cancel centrifugal force of the motor’s rotor. And at approximately 315 degree angle two forces align along one line. The remaining vibration level of 1.18 is the difference in weight between the motor and the fan. Step one is complete!


Step 2

In step one we found relative angle between fan and a motor where centrifugal forces align. In step two we find absolute location of that point where we need to apply more weight to make one force compensate another. How do we do that? You already know the answer – magnetic balancer! But we don’t have to use precise and expensive balancers. If you have one – great! The one I have has too large shaft diameter and I made another balancer out of a drill bit and a couple of magnets placed between vise jaws.

Now, place a fan rotor in the balancer, spin it a little and wait till it calms. With Sharpie marker place a small dot on top-most point of the fan. Repeat this step five times. By the end you’ll see an area of dots close to each other. Just round this area with Sharpie – this is where we are going to place extra weight.

I used white masking tape cut in ¼ inch squares. Based on construction of your fan and RPM you may need to use stickier tape or use different methods to add weight, like drops of CA glue or some other way. My fan is GWS-style where I can place tape pieces on top of the fan’s hub for measuring and when done move tape pieces inside the hub at the same location.

Put fan back on the motor aligning mark we made in step one at 315 degrees, the angle of minimal vibration.

a) Stick one square of masking tape to the area we rounded as lightest point of the fan
b) Start sensor’s recording
c) Gradually increase motor’s RPM to the max and back during approximately 10 second interval
d) Stop recording and save result file
e) Repeat steps a-b-c-d few times

After a few pieces of extra weight you will quickly see a dramatic difference in sound of your EDF. At this point data recording is optional but I’ll do it anyway to measure the result. At the point when next weight addition increases vibration unstick last tape piece, cut it in half and glue back one half. You can get as precise as you wish at this point.

Let’s export collected data to Excel and take a look at what we’ve got.





We can see that we hit vibration minimum when sticker number six was placed on the fan. By experimenting with smaller tape pieces between weights five a seven it is easy to find absolute minimum. In this experiment I just removed weight seven and left weight six unchanged.


The result

So what is the measurable result of our balance exercise? The result is amazing 91% reduction in vibrations of EDF unit! This is accomplished without expensive dynamic balancers, no tedious (and some times ineffective) fan balancing, no motor’s balancing. We got tremendous improvements with simple and relatively inaccurate steps. This result also shows how horribly unbalanced my EDF unit was.

Subjectively I can say every time I rev up my four engine A-380 I can’t stop smiling and enjoy how she sounds. If earlier I was worried there will be complaints how loud this plane is now I am having hard time telling by ear what RPM’s my engines are working at when plane is in the air and have to rely on throttle stick position. And realistic jet sound during a takeoff is music to my ears. I also noticed significant increase in power of engines. The plane can climb at angles it could never climb before and stays level at 50% throttle, vs. about 75% when it was unbalanced. I wish I had thrust measurements when unbalanced to compare results.


Conclusion

It takes about 40 minutes to an hour to balance one EDF unit. One can go quick and “good enough” way to get 90% improvement or can be as precise and detailed to get that last percentage point.

There can be variations to this method, such as instead of using accelerometers it is possible to use microphone or simply rely on your feelings. There is no need to know absolute vibration levels. Step one goal is to find that minimum vibration angle which is relative to other angles. And step two can be done without any measurements at all. You will know when you get that perfect weight by the smile on your face.

I would like to thank my friend Dale who gave me initial EDF balancing idea from which it all started. Thanks buddy!




--------------------------------------------------------




Alternative balancing method


What if… you cannot rotate the fan? It may be glued to the shaft or impractical to unmount, the fan itself is statically balanced or you are allergic to a screw driver and magnetic balancer gives you a headache? Can EDF unit still be dynamically balanced? Unless laws of physics prove opposite the answer is YES!
Here is how…

Theory


Let’s take a look at the EDF system and centrifugal forces when unit is working.



Unbalanced fan produces force that is applied to the center of rotation (shaft) and directed toward its heaviest point (blue arrow). Unbalanced motor creates its own force which applied to the same center and directed toward its own heaviest point (green arrow). The resulting force of EDF system is sum of these two vectors (red arrow). In previous method fan unit is rotated to compensate one force with another. But what if… we create our own force to compensate?

Take a look at the image below.



By creating new counter force in our EDF system we effectively compensate existing force responsible for vibration. All we need is add counter weight at the right location. The steps remain the same. In Step 1 we find that location and in Step 2 we add necessary weight to compensate unwanted force.

Before we start practical implementation there is one important point that deserves its own discussion. Where, or more precisely how far from the center of the shaft counter weight should be located? This may seem obvious but wrong place can result in unexpected consequences. The problem is the force created by rotating mass is proportional to the radius of rotation. Another words, one gram added to the blade tip creates stronger force than one gram added to the hub. This results in one level of vibration at low RPM and completely different vibration at higher RPM. For outrunner motors correct distance from the center of the shaft is where magnets are glued to the rotor. For inrunner motors answer can be more complicated, or instead much simplier, and probably should be discussed separately.


Practical implementation


First of all we need to find place on the Fan unit where counter force needs to be applied. We need couple of small squares of sticky tape. How many exactly can be determined experimentally. The counter weight should be small enough to not create too much stress and heavy enough to make measurable difference in balance.


Step 1
a) Mark fan’s hub with a sharpie every 45 degrees. Make one of the marks different – it will be a starting point
b) Stick counterweight to the starting point
c) Start sensor’s data recording
d) Gradually increase motor’s RPM to the max and back during approximately 10 seconds interval
e) Stop recording and save result file
f) Un-stick counterweight and move to the next point
g) Repeat steps c-d-e-f until you reach angle of 315 degrees, which will be the last measurement point.

The result should look very similar to the spider chart produced by Step 1 in previous method. There will be a minimum vibration point where weight needs to be applied to compensate vibration.

Step 2

a) Stick counterweight to the minimum vibration point of the fan
b) Add one more piece of tape
c) Start sensor’s recording
d) Gradually increase motor’s RPM to the max and back during approximately 10 second interval
e) Stop recording and save result file
f) Repeat steps b-c-d-e few times

The data analysis is exactly the same as described in previous article.

This method is less labor-intensive and much quicker than previous one. No screw drivers, no fan balancers, just phone and masking tape.

Very interesting approach to balancing a fan, thanks for writting it up so well and sharing with us.

Maybe this also explains if one just randomly places a factory dynamically balanced rotor
on a motor shaft....you can still get some nasty vibrations?

Please explain this, I confess to screwing up and stupidly turned off the Spectrum Tx before unplugging the batt to ESC during set up of a 70mm fan. I set everything down and as soon as I turned off the Tx it went WFO or WOT and sucked up a paper towel.( looked like snow in the kitchen) Its the best balance job I've ever had to date, spins true from zero to 1750watts.
Go figure, any other day I'd of needed a new rotor.
And I do spool up then rotate the rotor to get it true over and over till its good.
Tom

One explanation from the top of my head is there was misalignment during the motor assembly (or fan's hub?). When paper towel suddenly created stress misaligned part "clicked" to its intended place.

Wow!!! you took a simple idea and turned it into a science.Wow!!
I am blown away impressed. All new edf guys need to know about what you've done with this. This thread will be invaluable to all those who haven't a clue how good these motors can sound (ergot increasing efficiency) when balanced correctly.
Now what time can I bring my airliner by for the "Sergey" treatment?
My "Thanks" are to you for taking it to a higher level my friend.

Dale

That's pretty cool, Thanks for sharing!

Now I'm going to have to borrow the wifes 'Droid.

"Sweetie, look I gave you a new app for your phone."

Thanks for the article Sergey!
Great info...I'm going to do this in the next few days...
Z

the phone is bascially taking the subjectivity out of the the equation. Instead of run/turn/run/turn/run ahhh that feels smooth, it is run/turn/run/turn...examine data, ahh there IS the smoothest position. Now balance and repeat...
I like this, when my BB is due for upgrade, I will have look at a Droid.

This is a great idea, although I think it is working for a different reason than the one you explained.

[EDIT Ignore the following. you already knew this...]
Your explaination is for static balance whereas you need to worry about dynamic balancing. However, your balancing method may well be getting your system reasonably well dynamically balanced , just like when you have your car tyres balanced (i.e. the guy spins the wheel which measures the vibration forces and calculates where two weights need to be placed - one on the inside of the wheel and one on the outside).

See http://en.wikipedia.org/wiki/Tire_balance. You may be able to get it even better! [/EDIT]

Well done.

Very nice! I almost want to try it the next time we need to do balancing on a real helicopter just to see what it does lol. One of our guys has a droid and we might compare the accelerometer readings with the readings from our chadwick 8500 balancing box.

What a great thread. Subscribing!

Ok this may sound like some dumb questions but...

1) What would an equivalent "app" be? I have an iphone and all I found was an app called "vibration" that takes X, Y, Z readings? there is some kind of "sec/div" value also given? would this work?

2) Are the "vibration" values in your diagram, the average of the X, Y, Z readings? I see in the photo of your droid it too looks like it's giving these readings, like an oscilloscope?

Thanks for your help...;-)
Z

Some possible app for the iphone

Showcase: Vibration Turns Your iPhone into a Spectrum Analyzer
http://www.macresearch.org/showcase-...ctrum-analyzer


Turn your iPhone or iPod touch into a vibration spectrum analyzer
http://www.macworld.com/appguide/app.html?id=67886

Originally Posted by Ken Lapointe

Some possible app for the iphone

Showcase: Vibration Turns Your iPhone into a Spectrum Analyzer
http://www.macresearch.org/showcase-...ctrum-analyzer


Turn your iPhone or iPod touch into a vibration spectrum analyzer
http://www.macworld.com/appguide/app.html?id=67886

Ken, the first one is the one I got;-)
Z