TravisR100
Cast Iron
- Joined
- Aug 7, 2006
- Location
- Houston, TX
I recently bought a surface/tool grinder from a fellow member here. I haven’t been able to get a good finish with it. Shows some scalloping on the finish. It was pretty obvious there was a fair amount of vibration coming from the motor. It had, if not the original, the motor that was designed to go on it. It was a single phase Rockwell 3/4 hp motor.
I decided to replace it. I got an 1800 RPM 3 phase WEG motor and matching WEG VFD. They arrived yesterday and I promptly wired up the motor on the bench. The literature for this grinder says that the original motor and pulley are balanced from the factory and that should the pulley be removed that it should be rebalanced.
After searching the Internet for information on balancing I came across a site where someone had published an article and spreadsheet for a method of balancing using a real time analyzer, accelerometer, or displacement gauge. At the heart of it he says that the device used to measure the vibration can be almost anything that gives a quantifiable vibration measurement, even a speaker with a weight placed at the center of the cone to generate an electrical signal that can be measured. I got the feeling the article was written before smartphones were around.
Being that I have an iPhone I looked and found a fairly sophisticated vibration measurement app. I decided to use his spreadsheet and the phone to attempt to take some measurements and improve the balance.
Here’s the motor on the bench...
What’s interesting about the app is that it doesn’t just show a composite vibration measurement, it gives values for vibration in 3 axis, x, y, and z. The way I placed the phone on the motor the x axis would be from the top to bottom of the picture above (lateral axis), the y axis (longitudinal) would be axial from the front to back of the motor shaft, and the z axis (vertical) would be up and down, normal to the bench.
The app can give measurements in a variety of units but I chose to take readings in m/s/s. The values are so low that what I’m reporting below are the values times 1,000. So below where I say the value was .4 it was actually .0004.
With the phone sitting on the bench, motor not running, it reports x .4, y .4, and z .5. This is the inherent noise in the sensor I believe. No matter where I place the phone I can’t get a reading less than those values.
Placing the phone on the motor while the motor sits directly on the bench, no pulley, key installed in shaft, motor running I was getting
X 2.4, Y 8.3 and Z 1.8. It occurred to me that if the motor were rigidly mounted that the vibration values would go down since the motor wasn’t free to move and thus generate any acceleration. So I then set a piece of semi-rigid insulation foam on the bench and set the motor on top of that. Again, no pulley. Values were
X 29.7 Y 11.3 Z 27.5.
The foam obviously gave the motor some freedom to move. My guess is that a softer piece of foam might work even better.
As I would expect most of the vibration is in the X and Z axis which would be vibration up and down and side to side. The Y axis would be axial. I’ll leave out Y values from here on out.
I then added the pulley and ran it again. Values were
X 36.3 Z 37.9
So the pulley increased vertical and lateral vibration which is what I would expect. Keep in mind that this thing FEELS dead smooth. If you rest a hand on the motor you can just perceptibly feel what I’d characterize as the slightest hum through your hand.
The process then calls for you to place a weight on the pulley at a zero reference, then a 120 degree reference, then a -120 reference and take measurements with the weight at each of those 3 positions. The amount of weight can supposedly be anything however it was obvious that too much weight would induce a vibration that was too harsh and I didn’t figure would give good results. I ended up using a weight that was 1.06 grams. My measurements were as follows in X and Z.
No weight
36.3, 37.9
Zero
48.6, 47.9
120
33.6, 35.2
-120
77.5, 76.6
You then plug these numbers into a spreadsheet and it spits out an amount of weight to add or remove and the position at which to add or remove it. The spreadsheet assumes a composite vibration number so I had to pick an axis to use. As you can see the X and Z axis moved almost the same amount and so I decided to use the Z axis measurements.
The spreadsheet said to add 1.08 grams at 72 degrees. Now, I didn’t go mark the degree positions with any high degree of accuracy. I just eyeballed them. I took my original 1.06 weight and added it at approximately the 70 degree position. I then took a final measurement to see if it had reduced the vibration. My results were
21.7, 15.5
Compared to the original measurement with the pulley and no weight of
36.3, 37.9
Seems to be a significant improvement.
A couple of things to note, the app is very interesting. It can not only spit out a graph and values of the realtime vibration going on it can do a FFT to isolate the frequencies at which the vibrations are occurring. It then plots those on a graph marked in either Hz or RPM (same thing, different expression). As you’d expect here’s what that looks like
It’s got a resonant vibration at exactly the motor rotation speed of 1805 RPM. If there were other things such as ball bearings, etc. causing resonance at other frequencies it would show those as well.
And here’s a screen shot of one of the measurement screens. I did use the RMS as opposed to peak values.
This can also be done without a spreadsheet just using trial and error. As opposed to taking measurements at 0, 120, and -120 degrees you could take them at 0, 90, 180, and 270. No weight initially, then each of those positions, and see which position makes the most improvement. Then take 2 more measurements either side of the most improved position, etc, until you hone in on the best reading.
Just FYI the iPhone is only good up to around 3000 RPM for doing this.
I also want to qualify this post by saying I no little to nothing about balancing. I also realize this is only a static balance. Since the pulley itself is the only thing I’m actually balancing here yes, it could be done by putting it on a static balancing fixture but I don’t have one and haven’t made one. Additionally the title should have been changed but I can’t change it now. I thought this was an interesting experiment at the least.
If nothing else it gives a way to quantify the vibration of one machine or system compared to another. The sensor in the phone is incredibly sensitive.
Below is another screen shot. Notice the spike in the Z axis (orange) right in the middle. For this I set the phone on my concrete shop floor, walked back 6’, and stomped my foot moderately.
Sent from my iPad using Tapatalk
I decided to replace it. I got an 1800 RPM 3 phase WEG motor and matching WEG VFD. They arrived yesterday and I promptly wired up the motor on the bench. The literature for this grinder says that the original motor and pulley are balanced from the factory and that should the pulley be removed that it should be rebalanced.
After searching the Internet for information on balancing I came across a site where someone had published an article and spreadsheet for a method of balancing using a real time analyzer, accelerometer, or displacement gauge. At the heart of it he says that the device used to measure the vibration can be almost anything that gives a quantifiable vibration measurement, even a speaker with a weight placed at the center of the cone to generate an electrical signal that can be measured. I got the feeling the article was written before smartphones were around.
Being that I have an iPhone I looked and found a fairly sophisticated vibration measurement app. I decided to use his spreadsheet and the phone to attempt to take some measurements and improve the balance.
Here’s the motor on the bench...
What’s interesting about the app is that it doesn’t just show a composite vibration measurement, it gives values for vibration in 3 axis, x, y, and z. The way I placed the phone on the motor the x axis would be from the top to bottom of the picture above (lateral axis), the y axis (longitudinal) would be axial from the front to back of the motor shaft, and the z axis (vertical) would be up and down, normal to the bench.
The app can give measurements in a variety of units but I chose to take readings in m/s/s. The values are so low that what I’m reporting below are the values times 1,000. So below where I say the value was .4 it was actually .0004.
With the phone sitting on the bench, motor not running, it reports x .4, y .4, and z .5. This is the inherent noise in the sensor I believe. No matter where I place the phone I can’t get a reading less than those values.
Placing the phone on the motor while the motor sits directly on the bench, no pulley, key installed in shaft, motor running I was getting
X 2.4, Y 8.3 and Z 1.8. It occurred to me that if the motor were rigidly mounted that the vibration values would go down since the motor wasn’t free to move and thus generate any acceleration. So I then set a piece of semi-rigid insulation foam on the bench and set the motor on top of that. Again, no pulley. Values were
X 29.7 Y 11.3 Z 27.5.
The foam obviously gave the motor some freedom to move. My guess is that a softer piece of foam might work even better.
As I would expect most of the vibration is in the X and Z axis which would be vibration up and down and side to side. The Y axis would be axial. I’ll leave out Y values from here on out.
I then added the pulley and ran it again. Values were
X 36.3 Z 37.9
So the pulley increased vertical and lateral vibration which is what I would expect. Keep in mind that this thing FEELS dead smooth. If you rest a hand on the motor you can just perceptibly feel what I’d characterize as the slightest hum through your hand.
The process then calls for you to place a weight on the pulley at a zero reference, then a 120 degree reference, then a -120 reference and take measurements with the weight at each of those 3 positions. The amount of weight can supposedly be anything however it was obvious that too much weight would induce a vibration that was too harsh and I didn’t figure would give good results. I ended up using a weight that was 1.06 grams. My measurements were as follows in X and Z.
No weight
36.3, 37.9
Zero
48.6, 47.9
120
33.6, 35.2
-120
77.5, 76.6
You then plug these numbers into a spreadsheet and it spits out an amount of weight to add or remove and the position at which to add or remove it. The spreadsheet assumes a composite vibration number so I had to pick an axis to use. As you can see the X and Z axis moved almost the same amount and so I decided to use the Z axis measurements.
The spreadsheet said to add 1.08 grams at 72 degrees. Now, I didn’t go mark the degree positions with any high degree of accuracy. I just eyeballed them. I took my original 1.06 weight and added it at approximately the 70 degree position. I then took a final measurement to see if it had reduced the vibration. My results were
21.7, 15.5
Compared to the original measurement with the pulley and no weight of
36.3, 37.9
Seems to be a significant improvement.
A couple of things to note, the app is very interesting. It can not only spit out a graph and values of the realtime vibration going on it can do a FFT to isolate the frequencies at which the vibrations are occurring. It then plots those on a graph marked in either Hz or RPM (same thing, different expression). As you’d expect here’s what that looks like
It’s got a resonant vibration at exactly the motor rotation speed of 1805 RPM. If there were other things such as ball bearings, etc. causing resonance at other frequencies it would show those as well.
And here’s a screen shot of one of the measurement screens. I did use the RMS as opposed to peak values.
This can also be done without a spreadsheet just using trial and error. As opposed to taking measurements at 0, 120, and -120 degrees you could take them at 0, 90, 180, and 270. No weight initially, then each of those positions, and see which position makes the most improvement. Then take 2 more measurements either side of the most improved position, etc, until you hone in on the best reading.
Just FYI the iPhone is only good up to around 3000 RPM for doing this.
I also want to qualify this post by saying I no little to nothing about balancing. I also realize this is only a static balance. Since the pulley itself is the only thing I’m actually balancing here yes, it could be done by putting it on a static balancing fixture but I don’t have one and haven’t made one. Additionally the title should have been changed but I can’t change it now. I thought this was an interesting experiment at the least.
If nothing else it gives a way to quantify the vibration of one machine or system compared to another. The sensor in the phone is incredibly sensitive.
Below is another screen shot. Notice the spike in the Z axis (orange) right in the middle. For this I set the phone on my concrete shop floor, walked back 6’, and stomped my foot moderately.
Sent from my iPad using Tapatalk