Dynamic Balancing in the Home Shop
In working on this T&CG, I'd like to remove a bit of vibration in the motor. Static balance isn't going to fix it; to much drag from the bearings and parts of the shaft are several thou out so static balance on anything but the bearings would seem to introduce error; besides i don't have a static balance with rails just the right distance apart to rest it right where the bearings would be.
The proper way to do this, and any spindle or rotor is dynamic balancing; for the unfamilair that is measuring how much the element is out while running at speed, adding or removing weight in the right spot and trying again....but i can't recall there being content on dynamic balancing in the home shop?
Any balancing experts here that can describe the process, or how it might be simplified for the home shop? I guess you need some sort actuator to measure the vibratory movement and way to monitor the shaft speed or angle?...and a V block bed for the bearings to sit in? I know high end commercial balancers can do everything but turn the lights out...but is there a doable way to simplify the things to the get worst of the imbalance out?
Anyone accomplished this at home or have ideas how to go about it?
crap, another title typo, can someone add the "h" in home? thanks
Last edited by Mcgyver; 05-04-2010 at 03:42 PM.
Iv never heard of doing it in the home shop. You should be able to find a motor shop that will do it for a reasonable price however (Cheaper then building your own dynamic balancer anyway)
The obvious way to do this for me is to attach an accelerometer to the frame and use the peaks from this to drive a strobe - a strong LED source will do - that illuminates the rotating item.
You'd make marks on the shaft, or attach a degree disk to it. Then you'd build an amplifier for whatever signal the aceelerometer was giving. The LEDs would light up the disc to show you where the shaft was when it was slinging the frame on one direction. You'd take weight off that side and try again.
I looked up transducers for this a week or so ago. I think my search came up with Nat Semi and Harris. I think I gave it up when I realised all the modern chips would be digital. My electronics is a bit old-hat now, so I gave up. I'm sure that deeper research would get you somewhere, even to the extent of suggesting some IC amps to do most of the job for you.
It all depends on whether you're up to playing with opamps and little circuits. You wouldn't have to go past the breadboard stage to get something like this going, I'd have thought.
I'm sure some experts will give you a better analysis than me.
If I remember, I spent the rest of the hour reading the post about why you didn't need to balance grinding wheels, and left it at that.
There is one method I heard from a man who used to work at the Alvis works. (makers of luxury cars and aero engines)
He described some procedure which if I recall correctly went like this:
The machine to be balanced was mounted on compliant mountings and a light metal disc fixed to the end of the rotating shaft.
The machine was turned over by hand while scribing a circle on the disc centred on the axis of rotation of the machine.
Then the machine was turned at speed and another circle scribed on the disc.
Depending on how far out of balance the machine was the second circle would be more or less out of round and off centre. This would also indicate the heavy side etc. Both ends of the machine were done.
When I had the Archer I had the engine dynamically balanced. Helped somewhat, but a 360-cubic inch opposed 4-cyl engine is just never going to be smooth...
The process was to attach a 3-axis accelerometer at the front of the engine and another on the back. A non-contact photosensor was set up to note rotational position with a reflective dot on the prop spinner. The tech had a portable readout box. We ran the engine up, the tech said OK, and we shut her down. Bolted a few washers to the flywheel using the already provided holes (the readout box said where and how much). Started it back up, checked again, much better, done. Took about 30 minutes, cost $150 which is dirt cheap for anything aviation related.
That said, Analog Devices makes a 3-axis MEMS accelerometer that would no doubt work fantastic for this application. Were I doing it, I'd get one of those (ADXL330, probably) and some kind of a once per rev index. I'd attach all 4 traces to my 4-channel scope and let her rip. The vibrations would be readily visible, and the relative position of the vibration to the index pulse would say how many degrees offset the weights would go.
If your scope has FFT capabilities (mine does) you could sort out once-per-rev vibrations which are easily balanced out against other vibrations, probably caused by faulty bearings, that can't be balanced out.
If fast hasn't replied yet, I think yes - FFT is covered by a spectrum analyser. FFT means Fast Fourier Transform - in computing and signal analysis it does anyway - which basically means you can see the spectrum.
Yes, FFT is conversion from time domain to frequency domain, which makes analyzing frequency signatures a piece of cake!
The ADXL330 should work, www.analog.com is Analog Devices web site. You might even get them to send 1-2 samples. There's even a nice evaluation board prewired you can get. Analog will sell direct, if they won't send a sample.
A 465B is a great scope, but might not work so well at the low frequencies because it's analog. 60Hz is a bit of a bear, the traces dim out quite fast so you've really got to crank the brightness. If you're good at using it you can pull the signals out. I'm hesitant to use my 453A at these low freq's because the tube's 40+ years old and I don't want to toast the phosphor.
If your specturm analyzer goes down to 20 Hz you'll be OK. Mine (HP 8591E) bottoms out at 9kHz so it's of no use at these frequencies, that's why I use the TDS2044's FFT analyzer.
You could try using the 456B in X/Y mode, instead of X/T mode. You'd hook up the two axes of the accelerometer that are perpindicular to the spindle to the two (normally) horizontal inputs. The, set it for X/Y which makes the top input drive the X axis and the bottom input drive the Y axis (normally driven from the timebase).
The 465B has a secret, sneaky third channel, which I'd hook up to the index pulse. That's the external trigger input. You've only got 2 possible input scales so you might need a voltage divider (x1 and x10). Then press the trigger view button (third one down? been a long time.)
I'm not sure this will work, but X/Y should produce a circle showing the disturbances. The trigger view trick should work even using the scope in X/T mode.
Last edited by fasto; 05-04-2010 at 05:50 PM.
Well, this is offered as a crazy shot in the dark but ought to be worth some interesting comments.....
Remember the gizmos sold to mount behind the rim of your car tires in order to maintain 'balance'? They were a circular tube, filled with oil and ball bearings. The idea was that the bearings would 'settle' in the tube at the point of needed weight application as the wheel spun. If you could mount a tube to your motor shaft and 'see' where the balls settle in during rotation (use an auto timing light synced to the shaft??), could you then determine were counterweights are required?
Disclaimer: I have no idea if those wheel balancing devices worked, and yes, I know if the car was stopped the balls would fall to the 'bottom' of the tube, unbalancing the tire.
Sparkfun.com sells the ADXL335 accelerometer mounted to a breakout board.