Safarir
Plastic
- Joined
- Jul 21, 2013
- Location
- Qc, Canada
Hi all,
I recently got my hand on my first surface grinder, a Brown & Sharp 510. After doing some initial testing, I was really unhappy with the surface finish I was getting: I am getting a lot of "wave" in line with the Y axis. After reading a lot on around the internet and on this post, I decided that the best first step is probably to balance my wheel. To do so, I made a simple wheel lock nut and used a microphone taped to the spindle and a spectrometer app on my phone to measure the vibration. I then used the 4 run technique to find the appropriate balancing weight.
The result was a lot better but not fantastic. To make sure that the problem come from vibration and not the table roller, I made this test cut:
On this block, the top part mark (with a M mark) was cut at a medium feed speed, the center part was cut at a slow feed speed and the lower part at a fast feed speed. My conclusion from this test is that since the line spacing depend on the feed speed, the vibration must be coming from the wheel/spindle/head and not the table.
This is when I went down a rabbit hole: I decided to play with some vibration analysis. What I did is mount a cheap adxl345 accelerometer to the spindle housing and connect it to a esp32 which allow me to log data remotely.
I did some signal processing with python. I am running the capture at 400hz which give me a 200hz bandwidth once I do a dft analysis. Here are some of the results:
This first plot is the vibration with additional screw added to my balancing adapter. You can clearly see the spike at 47.5hz cause by the imbalance on the wheel, running at 2850rpm. One interesting thing is that the vibration seem way stronger along the height and not so much side to side.
This second graph is after I balanced the wheel. You can see it is quite a bit better around 47.5hz. One interesting fact is the spike around 112 moved to around 105, I still have not explanation about that.
At this point, you probably noticed the giant spike at 60hz. Fun fact, the vibration is actually totally off the scale at around 12, I had to chop it down a bit so we can see the rest of the graph. I think this is actually cause by the motor, running at 3600rpm. I decided to do some more testing and remove the o-ring that drive the spindle to test the vibration caused by the motor itself.
In this test, I did not move the sensor at all from the previous test. The really interesting part is that the entire noise floor dropped significantly ! I am really surprise how much vibration across all frequency range is coming from the spindle. The 60hz component is still there which to me mean my motor need some balancing.
Where am I going with all of this ? No idea ... I just thought this was quite interesting and wanted to share my finding. My next step is to remove the motor from the machine and test it by itself. I would really like to get rid of this 60hz vibration before doing more testing. I will check the bearings and might try to balance the motor.
I recently got my hand on my first surface grinder, a Brown & Sharp 510. After doing some initial testing, I was really unhappy with the surface finish I was getting: I am getting a lot of "wave" in line with the Y axis. After reading a lot on around the internet and on this post, I decided that the best first step is probably to balance my wheel. To do so, I made a simple wheel lock nut and used a microphone taped to the spindle and a spectrometer app on my phone to measure the vibration. I then used the 4 run technique to find the appropriate balancing weight.
The result was a lot better but not fantastic. To make sure that the problem come from vibration and not the table roller, I made this test cut:
On this block, the top part mark (with a M mark) was cut at a medium feed speed, the center part was cut at a slow feed speed and the lower part at a fast feed speed. My conclusion from this test is that since the line spacing depend on the feed speed, the vibration must be coming from the wheel/spindle/head and not the table.
This is when I went down a rabbit hole: I decided to play with some vibration analysis. What I did is mount a cheap adxl345 accelerometer to the spindle housing and connect it to a esp32 which allow me to log data remotely.
I did some signal processing with python. I am running the capture at 400hz which give me a 200hz bandwidth once I do a dft analysis. Here are some of the results:
This first plot is the vibration with additional screw added to my balancing adapter. You can clearly see the spike at 47.5hz cause by the imbalance on the wheel, running at 2850rpm. One interesting thing is that the vibration seem way stronger along the height and not so much side to side.
This second graph is after I balanced the wheel. You can see it is quite a bit better around 47.5hz. One interesting fact is the spike around 112 moved to around 105, I still have not explanation about that.
At this point, you probably noticed the giant spike at 60hz. Fun fact, the vibration is actually totally off the scale at around 12, I had to chop it down a bit so we can see the rest of the graph. I think this is actually cause by the motor, running at 3600rpm. I decided to do some more testing and remove the o-ring that drive the spindle to test the vibration caused by the motor itself.
In this test, I did not move the sensor at all from the previous test. The really interesting part is that the entire noise floor dropped significantly ! I am really surprise how much vibration across all frequency range is coming from the spindle. The 60hz component is still there which to me mean my motor need some balancing.
Where am I going with all of this ? No idea ... I just thought this was quite interesting and wanted to share my finding. My next step is to remove the motor from the machine and test it by itself. I would really like to get rid of this 60hz vibration before doing more testing. I will check the bearings and might try to balance the motor.