Detecting Washing Machine Drum Out Of Balance Limits

[madsi]

I am a noobie and I am not really sure this is the right forum to post to, so please excuse me if I am posting in the wrong place. I had considered General Physics and Classical Physics, but I hope I have made my best choice and I don't think it is acceptable to post in more than one forum.

The problem to be remedied: Shake, Rattle and Roll in my front loader Rock and Roller.

I am really fascinated by how really complex the physics and mechanics governing the operation of a seemingly simple appliance like a washing machine can be.

Out of hobby interest and a love of science and also finding some joy in hacking home appliances, I have hacked my own store-bought washing machine, tore out the existing computer controller, added my own MCU and now I can control almost all necessary washing functions.

That is, except out of balance conditions. These events have shown to create an excessive vibration sufficient to be a self-destructive event or cause the machine to wander from its placement or else create so much vibration as to upset me, my girlfriend or my neighbors in this apt. building.

To attempt to try to remedy this problem, I have bought a GY-521 MP6050 accelerometer module that can detect F.S. accelerations of +-2g, +-4g, +-8g or +-16g and delivers 250/500/1000/2000 deg/Sec gyro readings in all XYZ directions. This accelerometer/gyro can deliver up to 1000 readings/sec of all measurements.

What I know so far:

If the washing machine drum/motor unbalanced moves up >1.5-in or to the left >1-in or to the right >2-in there is a great destructive banging to be heard.

I have to detect an imbalance that is trending towards an out of balance limit before the limit is reached.
By the time the unbalance is detected, the drum Asm has already gained momentum and will continue to travel in its trajectory, even after all power is cut from the motor.

Movement is restrained in some directions by shock absorbers.. I even added two more to the concrete blocks at the top in an attempt to quench excessive vibration. However the mechanics of the suspension develops cyclic paths of vibration that align with the least restrained directions of movement.

With some calculations I can quantify pitch, yaw and roll of my wannabe flying washing machine.

I know how to create a moving averages of each of the XYZ accelerations as data to make a decision for the MCU to interrupt motor power when an out of balance condition is detected.

What I think what I am trying to do, at least in part, is to detect absolute displacement or a dangerous accelerating trend away from dead center from the acceleration/gyro readings.

The rotating drum/motor + wash clothes +water sloshing around has a mass of 40 to 60kg.
And while the total rotating mass has a total of 40 to 60 kg, the only out of balance portion of this mass is the wash and water in the drum/tub itself. I also have observed that the out of balance mass often forms itself more or less into a flattened ball-like shape inside the drum/tub. But every out of balance event is different and so it creates it's own cyclic pattern of vibration depending on speed, washload and the organization of the washload and water in the tub.
I have also observed that the rotation of the tub creates a spooling effect that creates the clumping that causes the imbalance and wonder how to prevent this by manipulation of the direction/acceleration of the tub from stopped position.

I know that acceleration is the rate of change of speed and that one g = 9.81 meters/Sec.
From that I think I know that if the drum Asm. is accelerating at 9.81 meters/Sec in any or all directions, that it will have reached 9.81 meters/sec of speed and the displacement from dead center resting position in one second would be 1/2 x 9.81 meters?.

So I need to know how to combine the various readings of acceleration and gyros to detect an out of balance condition.

But how do I attempt to combine this information into a remedy?

It have taken a physics and calculus course at the university, but I didn't too very well in my one and only general physics course.

I do know I could use a cut and try approach to achieve some amateur sloppy solution, but I hope to do better if I can get some good advice from you experts!

I think I need some help with the physics and math at this point.

Please help.

 

[Baluncore]

Is the unbalance only a problem on the spin cycle ?

At the start of the spin cycle you need to start spin slow and allow time for the majority of the water to drain before increasing speed. If you overload the machine or start the spin too quickly you should expect high amplitude displacements.

The ballast weights and the springs are selected to be resonant at specific speeds. If you change the springs or weights you will change the spin-up profile required to avoid high amplitude excursions.

You need to convert your acceleration readings to linear displacements. To do that you will need to know the frequency of rotation.

 

[madsi]

Thanks, Baluncore.

One thing I've learned is to drain the water before any fast tub speedup. I have full computer control of motor torque, speed and direction and my MCU controller can adapt it's motor control at will.

After some many experiments, I realized that I shouldn't even try to control the speed of the tub, after achieving a speed that is out of the stall range, just apply a constant level of power to achieve a slow speed that results in stall-free smooth drum rotation.

I realized, that with a constant rotational torque applied to the drum, the drum speeds up naturally as water is expelled, because the same rotational energy is being applied to a lighter and lighter load.

I have added two extra shock absorbers salvaged from another washing so I have changed the dynamics of the drum in respect to oscillation. This enhancement of additional shock absorbers only resulted in a shift of resonant behavior and the machine can still vibrate and oscillate violently upon imbalance.
This must be the case, since the motor is trying to rapidly oppose the XYZ positions of 40-60kg rotating load that is oscillating.

I have noticed there are low speed resonance points, but attempting to accelerate to a higher speed to move out of these resonances can end up with smaller amplitude higher frequency XYZ vibrations and result in very noisy vibrations that will lift the machine off its placement and it will try to slowly stroll around the washroom in a circular pattern.

The imbalance can happen anytime there is a great imbalance, but the greatest chance of this happening, even when underloaded, is always when the tub speeding up to rinse or spin.

Sometimes, regardless of the acceleration in speed to reach slow or fast spin, an imbalance will be created to create high amplitude displacements.

I can control the speed of the rotating drum, and I can see the oscillation caused by imbalance is complex.

The drum asm itself is balanced, only the wash and water combo create the imbalance oscillation around three axes..this is what I want to detect in time to prevent collisions.

Please help me with the math to be able to detect trend to an imbalance condition before its too late, predict this with monitoring of accelerations.

I don't know the physics and math needed to convert an array of XYZ acceleration readings over a short period and use these readings to predict linear displacements. If I knew this info, I could create code to attempt to manage this problem.

I do know that my computer control of the drum speed must react fast to prevent violent oscillations.

 

[gjonesy]

The one thing I've noticed about all washing machines is that the "load" has the most effect on the spin and when they are way out of balance its do to uneven weight distribution. When I wash my uniforms since its winter I sometimes wash my coat is very bulky so I usually have to place all my uniforms on one side and the coat on the other. I find the coat even after the spin cycle has retained enough water that it drips. I have to hang it up and wait till it drips dry and place it in the dryer or redistribute the load as best I can and spin again. The same case happens when I wash pillows or covers, I usually guesstimate where to put what when I load the machine. If its a horizontal drum the way those things are designed id try layering the load,(light item, heavy item, light esc.)

 

[madsi]

Note: Please don't read and bother to offer any more anecdotal or qualitative advice about how to wash clothes. I've already know how to wash clothes.

Please offer some quantitative, some explanation of the physical mechanics problem in terms of the math.

 

[Baluncore]

I don't know the physics and math needed to convert an array of XYZ acceleration readings over a short period and use these readings to predict linear displacements. If I knew this info, I could create code to attempt to manage this problem.

For small amplitudes, your displacements are probably sinusoidal. Many higher harmonics will be generated once the physical displacement becomes great enough to hit the buffers and the machine starts to jump around. Your aim should be to keep the analysis sinusoidal. When you go from single dimension measurement of acceleration to three dimensions, you may need to combine the three dimensions into a single 3D vector sum.

Analysis of your real data has a complexity in that the mass imbalance changes over time as items move about and water is lost. But, relating amplitude to acceleration only needs to know the rate of rotation.To understand that relationship requires you understand the dynamics of Simple Harmonic Motion. At some point you will recognise that mass does not come into the relationship between sinusoidal displacement, acceleration and rate of rotation.
Start here; https://en.wikipedia.org/wiki/Simple_harmonic_motion#Dynamics_of_simple_harmonic_motion

 

[berkeman]

Thread closed pending Moderation...