Orthoganol balancing - twinwheel grinders

I've recently had to fit a new wheel and rebalance my 8" twin wheel grinder. The method I use is pretty straightforward but may be of interest because it doesn't seem to be well known

A good start is to dress both wheels and then remove and individually statically balance each wheel. This only a partial cure because even a small amount of eccentricity in remounting the wheels can result in significant vibration.

Balancing the wheels when mounted on the grinder is much more difficult because it's not easy to find out how much weight is needed, where, and on which wheel.

This normally needs a proper dynamic balancer but, with a bit of patience, the job can be done with an oscilloscope and an old 2" computer loudspeaker.

The speaker is converted into an accelerometer by epoxying a

5/8" dia x 5/8" brass slug to the front of the voice coil. This drops the bass resonant frequency to about 75Hz. A small 50 to 100uf electrolytic is shunted across the voice coil to get rid of most of the high frequency bearing noise.

The rear of the speaker magnet is anchored to the pick off location on the grinder. One wheel is removed and the scope is then used to measure the vibration level generated by the remaining wheel.

The grinder is located on a piece of thick carpet or bungy rubber feet.

The balance weight holder that I use a is cup shaped similar to tin lid. This is bolted facing outwards on the side of the wheel and the balance weights are lumps of modelling clay pressed to the inside of the cup rim. It's convenient to roll the modelling clay into a long cylinder which can then be cut into appropriate lump sizes.

The balancing method is based on orthoganol balance weight placement.

Mark positions 1,2,3,4 each 90deg apart on the cup. Put a sufficiently large lump of clay on position 1 to roughly double the vibration amplitude. Then by successively halving the lump size and alternating as necessary between positions 1 and 3 reduce the vibration amplitude as much as possible. Unless you are lucky with your first choice of position this first reduction may be fairly small.

Now repeat the exercise but using positions 2 and 4. This will allow further reduction. Because these weights are positioned at 90 deg to the 1-3 axis (orthoganol), this weight change on the 2 - 4 axis does NOT affect the correctness of the weight added on the 1 - 3 axis.

However with the reduced vibration amplitude you can now increase the 'scope sensitivity and rebalance 1 - 3 more accurately. The balance can be improved with smaller weights and higher sensitivity settings until the display becomes too noisy to detect the basic sinusoidal unbalance signal or shows a consistent slow beat.

This beat results from the beat frequency between the 2F torque vibration produced by a single phase motor and the slightly below 1F frequency of the unbalance signal (F=supply frequency). It's easily distinguished by observing the unbalance signal change as the motor is switched off. Once you're down to this level further balance adjustment is pointless.

It's vital to be methodical in steadily reducing the size of the correcting weights that you add as you approach correct balance. This is because the 'scope only tells you the amount of the unbalance signal - not heavy or light. An improved reading can result from either not quite enough added weight or a bit too much - the result is the same. This ambiguity is first resolved by trying the same added weight at both the 1 & 3 (or 2 & 4) position - the result tells you which way to go.

Once you're getting near balance the 1 & 3 readings may be similar because the added weight is just enough to almost symmetrically bracket too heavy and too light. This is resolved by trying 1 & 3 again with the next smaller weight.

Once balanced, dress the wheel and check that the balance hasn't changed

With the first wheel safely balanced, add the second wheel and repeat the process.

P.S Pressed down and properly located on the inside rim of the cup modelling clay is OK as a permanent solution. However in the enthusiasm of making useful progress it is easy to carelessly dab a piece on and to find it later flying across the workshop. Be warned - a small piece of clay may not be life threatening but it's better not to stand in the line of fire!


Reply to
Loading thread data ...

Jim, this was very interesting, but I have a question.

It would seem to me that the amount of vibration in a grinder, would depend on two things:

1) Eccentricity of wheels 2) Mass of the grinder (or the grinder and its pedestal)

In other words, if the grinder has a thick enough shaft, and is rigidly attached to something heavy, that act by itself would dampen vibration.

Reply to

The vibration we are trying to minimise arises when the centre of mass of a rotating disk is not coincident with the axis of rotation.

It is the mass distribution not the shape that matters. The ONLY reason that a rotating eccentric disk generates vibration is because the weight distribution is wrong. The actual shape and the eccentricity have nothing to do with it.

The mass distribution error can be considered as perfect wheel with a single error mass attached at a known radius. The rotating force generated by this error mass tries to move the mass of the grinder plus pedestal. The grinder mass plus anything it is rigidly attached to acts reduce the amount by which the grinder body moves.

Things get a bit more complicated if you have a wide wheel and there are axial variations in mass distribution. However the same arguments apply - the heavier the grinder in relation to it's unbalanced wheel the smaller the observed vibration.


Reply to

I love the method, am way too lazy to implement it. No O-scope either. I wonder if a sufficiently sensitive AC voltmeter would suffice?

I will remember using the speaker as a sensor!


Reply to

That's brilliant, Jim.

How about adding some sort of synchronization signal? Paint a white dot on the cup and use a photo transistor connected to the scope's other channel to get a reference to make it much easier to figure out where to add weight.

Is a grinder fast enough to use self-balancing similar to what they do in CD-ROM drives? I think it's simply some sort of drum with loose steel balls inside.

Reply to
Robert Roland

As far as I know, all grinders of this type operate well above critical speed. That means that only an insanely massive shaft, maybe almost the diameter of the wheel, would be enough to raise the critical speed so that the vibration could be constrained by a massive base. Otherwise, no reasonable bearing could force the wheel to run at the shaft center. In general, any rotating object above critical speed should be expected to spin about its center of mass, and the shaft centerline be damned!


Reply to
Jon Elson

A millivoltmeter is, in some ways, more convenient than a 'scope but it needs to be analog and sensitive down to about 10mV full scale. A digital version would drive you to drink because of the rapidly fluctuating readings that the analog meter happily averages out.

For part of this work I actually used a home brew millivoltmeter based on a single CA3240. If there is interest in this I'll post the circuit diagram. However I do not recommend building this unless you're already well experienced with this sort of kit.


Reply to

I have played with that sort of system but it gets complicated. For the comparatively small balance improvement needed for a twin wheel grinder the stark simplicity of the speaker plus 'scope is all that is needed.

I'm pretty sure that would work but I've never tried it.I think the system is a bit like my cup but with the edge of the cup rim turned in to retain a partly filled complement of balls. Wen run up to speed the balls space themselves out into a pattern that cancels the unbalance.


Reply to

if you have a scope, and a strobe, and the speaker "accelerometer", then why not just "do it right" - make a paint mark on the wheel as an "index", synch the strobe to the rotational speed and synch the scope and see where the peak is WRT your accelerometer location and then just add weight at the one place where it is most effective?

Reply to
Bill Noble

I am always interested in seeing circuits. I do not think I have any CA3240 op amps, but do have some other varieties on hand.

Dan Caster

Reply to

Search on PC sound card oscilloscope. You will find several programs to use a sound card for a dual channel o'scope. Not good for high frequencies, but should work well for balancing grinding wheels. Might need a amplifier on the input in order to be sensitive enough.


Reply to

For extra credit, use some wire, a nail, and a magnet to construct a once- around sensor for the grinder that can be used to trigger the scope, so that (after practice) you'll know just where on the balance wheel to put the weight.

I thought you were just supposed to bolt the thing hard onto the bench and accept stuff falling on the floor 12' away.

Reply to
Tim Wescott

The CA3240 is nothing special. It was an early MOS input bipolar op amp. The CA3240A is better, and better protected but there are plenty of newer parts that are better. The first time I saw the CA3240 was after lightning hit the tower at a CATV headend, and wiped out every CA3240 that was fed from a sat receiver. Of course, it hit on a Friday afternoon at 4:00 pm. I had a dozen flown cross country, and had everything working by noon the next day.

formatting link

Reply to
Michael A. Terrell

The LM358 dual opamp in 8-pin DIP would also work well here. They're readily available for about half a buck.

Reply to
Don Foreman

The 3240 was used because I wanted a FET or JFET input opamp to permit the use of high value feedback resistors and a 1 Mohm input divider. Any modern FET or JFET opamp would be OK. The LM358 or similar would also work but, because of the input bias current of a bipolar opamp, there would be some loss of accuracy and zero shift with different attenuator settings.

The 1% resistors are only used to permit reasonably accurate voltage measurement. For use as an unbalance indicator, run of the mill 10% components are all that's necessary

The circuit diagram is at

formatting link

It's basically a x10/x30 gain stage followed by a feedback mean rectified 50mV to 0.5mA voltage to current converter.

The 0.5 mA meter was rescued from my junk box. 100ua or

1mA would be OK with suitable adjustment of the 100ohm calibration resistor. The exact value of the series 3.3 kohm series resistor is not important - its only purpose is to limit the maximum overload current through the meter.

A number of posters have suggested thatit would be helpful for the unbalance angular location to be identified by some form of optical or mechanical pickoff. As Don Foreman has already pointed out this does not help with this sort of setup. The uncontrolled error signal phase shifts resulting from the crude bungy mounting and the relatively low resonant frequency of the "accelerometer" make it impossible to identify the angular unbalance location with enough accuracy to be useful.

The whole point of the orthoganol method is that it makes accurate error phase angle measurement unnecessary.


Reply to

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.