Predicting Peak Displacement in Imbalanced Rotating Drum

[madsi]

If I am attempting to prevent a suspended motor-rotating imbalanced drum of high mass from colliding with its enclosure, how can I attempt to predict the maximum displacement from a balanced center of rotation if I have only a single sensor that can accurately measure(1000 samples/sec) acceleration and rate of rotational change in XYZ planes?

What I need to know how to do is to predict when a rotating drum is on a collision course with it's enclosure.

Case in point: Washing Machine, front-loader, a rotating heavely weighted drum/motor assembly filled with a little water and a lotta clothes that is ballasted by heavy weights, encumbered from excess travel by a few incompetent shock absorbers and suspended from two large springs, but is otherwise balanced in a vertical plane when unloaded with water and wash.

Can someone help me with the calculations? I haven't studied physics except for a single course in general physics. I am not a mechanical engineer. I do understand algebra, trig, math and I understand well the qualitative dynamics of mechanical systems and even remember doing well in calculus I and II, but what I do best is programming embedded MCU's. I code.

 

[Nidum]

Most washing machines work on simple basis of detecting whether drum vibration is above or below an acceptable threshold value .

If below threshold continue with spin .

If above threshold stop spinning .

Often there is a recovery procedure where clothes are tumbled slowly to redistribute them before a restart .

 

[CWatters]

+1 to just limiting the amplitude.

I very much doubt you could calculate the trajectory of the drum and even if you did how much warning would you get? Perhaps less than half a revolution? Would that be enough time to stop the impact given the inertia of the drum?

 

[madsi]

Thanks Nidum and CWatters.

By either better design more robust and thought out suspension of the drum and maybe also by more clever computer code, some washing machines, particularly the more expensive variety, seem to be better able to control violent vibrations.

I bought an inexpensive machine and expected to get what I paid for, but with the hope of hacking it to improve on its performance..that was my challenge. I also later realized that I needed the machine to always be available to wash my clothes..this sometimes conflicted with dealing with downtime in improving it.

I bought the cheapest, highest rated load capacity machine that I could find on sale. Sadly, its performance under stress of imbalance is shockingly incompetent. I am not prone to be believe in wild conspiracies, but I thought perhaps the designers intended to make this washing machine of mine to work in an inferior way in comparison with their more expensive models offered. So, I thought I could maybe improve on it.

That is why I tore out the existing controller and built my own controller that can control the motor and all pumps and solenoids and whistles.

But I don't know how to use my MCU hooked to the acclerometer/gyrometer sensor to maybe make things better.

I can write code that detects peak acceleration in all XYZ directions. I can create an short array of readings over a short period. However, the memory limitations of the MCU can only hold maybe a total of less than 200 hundred individual XYZ measurements and I have maybe at most .3 seconds to react.

My best guess is that my MCU must detect and react in less than a third of a second because the magnitude of momentum reached in an imbalance condition cannot be prevented from creating a collision if my MCU takes too long to make up its mind on when to cut power to the motor.

At the same time, any control strategy mustn't interrupt normal washing with nuisance tripping.

 

[CWatters]

I suspect the algorithim they use to detect and correct the out of balance situation is quite complicated.

As per post #2..

Our machine starts by spinning slowly for say 10-20 revolutions and then accelerates in stages. If it starts getting "out of wack" it stops and goes back to turning slowly again, this tumbles the clothes to redistribute them before it tries to ramp up the speed again.

Unfortunately I've no idea how they detect the drum is vibrating too much. They may use limit switches, or accelerometers or a combination of both?? If they just use an accelerometer they must set a different limit for each speed because. At low speeds you can have low acceleration at the same time as unacceptable amplitude. At high speeds you can have high acceleration even though the amplitude is acceptable.

If you come up with a new algorithm and it copes with this you might try patenting it!

 

[Nidum]

http://archives.sensorsmag.com/articles/0803/34/main.shtml

http://www.memsic.com/markets/Washi...D-E56A-AF37-3D197176A6821730&cmsStatusAdmin=0

nb: Download the application note from second link .

 

[madsi]

There are many ways I've thought of to detect excessive vibration, but a MEMS accelerometer seemed to me to be an easy solution.

Many different possible ways to detect vibration:

A very inexpensive tiny elctrolet microphone mounted on the tub or coupled to the plastic hose detecting water level can detect vibration sound and feed its output to an A2D input on the MCU for it to make sense of.

There are pneumatic pressure water level sensors in most newer machines that provide pulse width feedback to the MCU to sense water level. When the machine is not at rest, movement of the drum will cause the water or even air trapped in the bottom of the sensor tube at the bottom of the drum leading up to the water sensor mounted on the interior of the enclosure near the top of the machine to vary in pressure. By detecting these small pressure variations in the air trapped in the plastic hose, the MCU can detect vibration which causes rapid fluctuations of fed back PW.

The motor tachometer will show oscillations in motor speed as the motor reacts to variations in speed when powered at a fixed power because the motor power is being precisely controlled by PWM controlling the AC power by duty cycle. Easy enough for an an MCU to detect as the tach is a small motor mounted on it rear shaft of the washing motor sends back sinewaves at 8 x tub revolution's speed.

There are also some less reliable totally mechanical methods to detect acceleration such as a tiny weighted pendulum mounted on the drum monitored by IRF sensors, or simply a ball bearing presssed down upon a weak spring mounted in an enclosure on the tub and this sensitive switch will open and close due to inertia of the ball bearing with heavy vibration, or a small 24um gold-plated surface of a ball-bearing with 3-degrees of freedom in a small enclosure making different electric contacts on the sides of its enclosure.

But I want to try to figure out how to interpret my MEMS accelerometer feedback.

 

[madsi]

Thanks CWatters, I really enjoyed the youtube!

Now I know how a properly built machine should handle out of balances.

I will definitely implement these control strategies in my newest revision of the MCU washing machine controller I have built.

But first I have to find out how to properly detect imbalances and attempt to predict displacement caused by imbalances with my $2 MEMS accelerometer module.

In an effort to carefully orchestrate operation of a machine according to its selling price, I see good reasons why these well thought out strategies to deal with vibration were not implemented in the out of carton, before my MCU was used, cheap machine I'd bought.

.

 

[madsi]

Thanks Nidum, this is the first time I've seen this article and it was published in 2003.

Funny enough, this applications engineer at Analog Devices seems to be parroting the essence of all of my research and discoveries about vibration detection.

Although the app note (written in very high-quality Chinglish) I've downloaded clearly shows the suitability of the MEMS device Analog Devices is touting, this ap note gives only vague qualitative guidance and leaves the washing machine engineer to figure out how to use and interpret XY acceleration readings to implement a MEMS device in a real-world washing machine..that is what I want to learn how to do too!

Interesting, dated 2009, with a MEMS not able to monitor the very important acceleration in the Z-axis.

 

[madsi]

I need some help to create a quantitative understanding of a classical physics problem.
I have Maple on my PC ready to study any solutions or help offered.
--------------------------
I know that acceleration is the first derivative of velocity. Therefore I know: velocity= g T

This is, of course, a simple calculation, when accel. g is constant and T elapsed is precisely known and when the coeff. of static or moving friction are not used to further precise the final XYZ velocities.

I also know that the forces accelerating a rotating mass can be resolved into XYZ component vectors.

First question:
I also think I know that if I integrate the known acceleration over an interval of time, I should be able to determine the instantaneous value of velocity, then finding the achieved displacement=velocity x time to get an estimate of imbalance displacement from center of rotating axis?
Velocity= integral of g during time dt
2nd question:
Assuming the 1st assumption is correct, if I can easily use a sensor to obtain multiple sampled and accurate values of XYZ acceleration and Deg/Sec gyro info from a rotating drum over a few hundred mSecs of time, is it practical or reasonably possible to determine or predict the present or future displacement from the axis of balanced rotation or determine the displacement of a 20-in diam drum from its balanced axis center or when a displacement of only an inch or two will be reached.
Assume the rotational speed of the unbalanced drum would be approx. >0 to 10 rps
3rd question:
What is the math to accomplish this calculation?
There is no problem in the system in question to obtain a sampling rate is up to 1000 samples per second of instantaneous acceleration and Deg/Sec gyro XYZ feedback and there is an embedded computer competent, if correctly programmed to handle the calculations very quickly.
----------------------------------------------------------------------------------------------------------------------------------
Note: Please don't read the info below and bother to offer qualitative advice about how to wash clothes. I've already know how to wash clothes.
---------------------------------------------------------------------------------------------------------------------------------------
I am neither a physics expert nor a mathematician, I am an embedded systems programmer, but I think I do have a strong qualitative understanding of mechanics and I have achieved a top grade in Calculus I and II at an university level, and at the same time I also took a single introductory class called Physics for Engineering Students( but it has been some years since I studied these subjects and this knowledge is rusty.)

In my specific problem, consider now a large balanced hallow drum mounted in the horizontal axis, partially filled with an amorphous mass that can organize itself dynamically during startup and accelerated rotation and so causing an imbalance whose XYZ magnitude of displacement may be trending toward an unacceptable limit.
-----------
Consider this imbalanced accelerating horizontal rotating drum is suspended by springs, ballasted by large horizontal weights(2-in concrete blocks with approx total of 2x the mass of the drum/motor) upon the top and bottom of the drum and suspended vertically upon large springs.

In a further incompetent attempt to stabilize the system that only achieves some dampening of oscillation, there is attached two (ea top and bottom) shock absorbersm which are attached at approx. 30 deg angles to the left/right sides of the top/bottom concrete blocks that are mounted on the drum.
.
An Out of Balance (OOB) condition often develops, resulting in an oscillating XYZ displacement from the center horiz. axis of the drum which produces peak displacements that are inversely proportional to drum's rotational speed.

There is a then a need to monitor the movement of the drum to reliably detect the amount of imbalance (OOB) and predict the instantaneous displacement in order to determine a safety trip point to momentarily cutoff power to interrupt an OOB condition and signal a need for redistributing the contents of the drum before resuming operation.

The amount of displacement caused by imbalance oscillations cannot exceed one to two inches before an excessive OOB condition exits.

The rotating drum is otherwise well-balanced, but is contents(some water and fabric) can organize itself as an imbalanced mass that sustains an imbalanced condition.

What is known is the approximate mass of the drum, ballasting weights, horizontal rotational drum speed and
an accelerometer sensor mounted on the horiz top center stabilizing concrete block upon the drum yields instantaneous XYZ accelerations and gyroscope Degs/Sec readings up to 1000 times per second.

From the sampled XYZ accelerations, I would like to understand how to make useful calculation using these XYZ sampled readings to determine and predict the displacement from center of balanced rotation or at least determine an excessive OOB state.

 

[madsi]

If I am attempting to prevent a suspended motor-rotating imbalanced drum of high mass from colliding with its enclosure, how can I attempt to predict the maximum displacement from a balanced center of rotation if I have only a single sensor that can accurately measure(1000 samples/sec) acceleration and rate of rotational change in XYZ planes?

What I need to know how to do is to predict when a rotating drum is on a collision course with it's enclosure.

Case in point: Washing Machine, front-loader, a rotating heavely weighted drum/motor assembly filled with a little water and a lotta clothes that is ballasted by heavy weights, encumbered from excess travel by a few incompetent shock absorbers and suspended from two large springs, but is otherwise balanced in a vertical plane when unloaded with water and wash.

Can someone help me with the calculations? I haven't studied physics except for a single course in general physics. I am not a mechanical engineer. I do understand algebra, trig, math and I understand well the qualitative dynamics of mechanical systems and even remember doing well in calculus I and II, but what I do best is programming embedded MCU's. I code.

I need some in understanding the mathematics and the physics and the physics mechanics equations to do some calculations to help in creating an algorithm.

 

[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.

 

[Nidum]

Start here

 

[madsi]

Thanks Nidum, I now have some link towards better working on this problem.
I realize that I didn't know what key words to query Google for some help,

I know that I will find links to some PhD physics level of elaborations of my problem of interest.

How will I be able to further get help from this site if need help to start to clearly understand the presented notation/equations/explanations I will find?

Is everyone saying, easy..go back to college and get your master's in Physics..then ask your question??

 

[Isaac0427]

Displacement= integral of g during time dt

Try using dimensional analysis for that. Taking such an integral will give you the units of m/s, which is the unit for velocity. Should you integrate that acceleration over time, you would get the equation v=gt, and if you integrate it again, you would get d=gt²/2. That is your displacement.

 

[Nidum]

and then look here

Edit : Better link

 

[madsi]

Thanks Isaac0427,
This is a most useful start for me to approach understanding and creating a solution for this mechanical problem.

I don't quite understand how Dimensional Analysis helps..isn't this just about the actual units of measurement, or are you giving me a hint to look at a scientific field of study, some reference to a physics analysis of the dynamics of a mechanical system?

 

[madsi]

Thanks again Nidum!

I feel so dumb in suddenly realizing there exists is a wealth of info on this specific subject if only I googled my query correctly! It does take some time to understand a problem sufficiently to ask a viable question!

I also realize that I will likely discover that these links usually lead to rather esoteric, complex but vague explanations that have been carefully edited to protect any valuable trade secrets and then may fail to offer any real practical knowledge that would give possible competitors any advantage.

Or else these links are to published papers that require a paid subscription to view or else are just published to gain proof of concept for a patent or just published by someone wanting to obtain some notoriety to post on their resume.

But I mustn't be so negative before taking some time to explore these many links.

 

[Isaac0427]

I don't quite understand how Dimensional Analysis helps..isn't this just about the actual units of measurement, or are you giving me a hint to look at a scientific field of study, some reference to a physics analysis of the dynamics of a mechanical system?

Whenever making/testing out an equation, you should always start at dimensional analysis. If the dimensional analysis doesn't work (i.e. the two sides of the equation do not have the same units) then you can do one of 2 things:
1. Put a constant out front
2. Find out that the equation is wrong
It is downright nonsensical to put a constant in front of your integral (if you think of it qualitatively, you will realize why it doesn't make sense), so number 2 must be the case. By integrating twice, dimensional analysis does work out, which shows that the equation could possibly be correct. It turns out that displacement is defined as the integral of velocity over time, and velocity is defined as the integral of acceleration over time.

 

[madsi]

Thanks again, Isaac0427
Of course,
I get it, I hurriedly show g as a constant in my equation, but I meant if the g was a steady value/measured, so assumed to be unchanging(constant) during the interval in question, not a constant fixed value that would not be changed in any other instance of use of the same equation.

 

[madsi]

So I now know the equations: v=gt, and if you integrate it again, you would get d=g(t^2)/2 (displacement.)

Seems simple enough, but the demons are in the details:

Do I just sum the instantaneous g readings * dt, (they are changing over the fixed sampling intervals and fixed sampling interval)

1)then take the moving average of the result..or the average of the sum of each sample product of Accel*dt
2)..or is it more correct to take the average value of Accel * total measurement interval to make calculations that will yield the integral?
3)or is it that: V= Summation (sample Accel in g's)*dt(sampling period, a constant value, i.e. some value like 2 to 10 mSec) ...or something else?
---------------------------------------------
Since the measured acceleration are happening in XYZ, I know any value is instantaneous, and I have observed the waveform of the XYZ accelerations, it is signed, monotonic, but slowly changing in value and direction(at least at low speeds of drum rotation) and acceleration readings are very quickly changing in value at spin speeds.

I also know that violent vibrations are manifest ramping-up towards a spin speed and they need to be detected before spin/centrifuge speeds are reached.

How does this insight relate to a way to do my calculations?

So then what kind of value will I get for XYZ accelerations, is what I want is a peak or average or integrated value to plug into the second displacement-finding equation for some idea of present or future displacement?

Finally, even if some value of displacement is determined, how could I possibly know that I have obtained a non-drifting XYZ accurate instantaneous or predicted displacement from the center of rotation, which is, of course, the most important measurements, considering there are sampling error inaccuracies and number-crunching accumulated rounding errors and the jitter(some few uSecs) from sample to sample?

In other words I may always know a quite accurate value of XYZ instantaneous sampled accelerations, but probably can never know a correct instantaneous position of the drum after any short interval.. unless the XYZ gyroscopic readings are included give me a way to get a more accurate way to determine position?

Is the only way to accomplish solving my problem is to attach absolute displacement XYZ sensors?

Or must I resolve myself to settle for a cut and try empirical approach to finding someway to identify critical peak values have been detected in any of the XYZ accels that signal OOB?
Or if is it the case I must somehow combine XYZ to declare OOB, can someone tell me how I can combine XYZ accelerations over the sampled interval or consecutive intervals to somehow declare an OOB condition?

Is it really NASA grade Rocket Science I am trying to deal with?
I do know that more expensive machines know how to detect OOB and deal with it in a way that works.

I still seem to be quite confused on how to make my calculations make sense of my readings?

 

[madsi]

Nidum:

https://www.google.dk/search?q=use+...&oe=utf-8&gws_rd=cr&ei=-1TbVuKtG6HP6AStn56IBA

even seems closer to the problem.

 

[Nidum]

Yes - good source of information for applications using that specific device .

 

[madsi]

The citation below is quite a remarkable academic paper that doesn't fail, reveals all, showing extremely complicated equations, graphs and calculus I couldn't begin to try to understand until I get my phD:

From the first link Nidum suggested:

DEVELOPING AN UNDERSTANDING OF
WASHING MACHINE DYNAMICS
Clive Marsh∗, Steve Taylor†, Paul Milliken‡ and Galkadowite Senaratne

And after 150 pages of explanation, they come to the summary:

6. Conclusions
Modern washing machines with balance rings are complicated systems
and it was not possible to build a 3D model including balance
rings and an out-of-balance load in the time available.
[2] showed that balance rings can reduce eccentricity due to an
out-of-balance-load but cannot eliminate it.
------------------------------------------------------------------------------------------
In India, housewives take their baskets of laundry down to the sacred river and beat each item individually, violently against rocks, like they were exorcising demons from the cloth itself.

 

[madsi]

I know the equations: v=gt, and if you integrate it again, you would get d=g(t^2)/2 the displacement.

Seems simple enough, but the demons are in the details:

Do I just sum the instantaneous g readings * dt, (they are changing over the fixed sampling intervals and fixed sampling interval)

1)then take the moving average of the result..or the average of the sum of each sample product of Accel*dt
2)..or is it more correct to take the average value of Accel * total measurement interval to make calculations that will yield the integral?
3)or is it that: V= Summation (sample Accel in g's)*dt(sampling period, a constant value, i.e. some value like 2 to 10 mSec) ...or something else?
---------------------------------------------
Since the measured acceleration are happening in XYZ, I know any value is instantaneous, and I have observed the waveform of the XYZ accelerations, it is signed, monotonic, but slowly changing in value and direction(at least at low speeds of drum rotation) and acceleration readings are very quickly changing in value at spin speeds.

I also know that violent vibrations are manifest ramping-up towards a spin speed and they need to be detected before spin/centrifuge speeds are reached.

How does this insight relate to a way to do my calculations?

So then what kind of value will I get for XYZ accelerations, is what I want is a peak or average or integrated value to plug into second displacement-finding equation for some idea of present or future displacement?

Finally, even if some value of displacement is determined, how could I possibly know that I have obtained a non-drifting XYZ accurate instantaneous or predicted displacement from the center of rotation, which is, of course, the most important measurements, considering there are sampling error inaccuracies and number-crunching accumulated rounding errors and the jitter(some few uSecs) from sample to sample?

In other words I may always know a quite accurate value of XYZ instantaneous sampled accelerations, but probably can never know a correct instantaneous position of the drum after any short interval.. unless the XYZ gyroscopic readings are included give me a way to get a more accurate way to determine position?

Is the only way to accomplish solving my problem is to attach absolute displacement XYZ sensors?

Or must I resolve myself to settle for a cut and try empirical approach to finding someway to identify critical peak values have been detected in any of the XYZ accels that signal OOB?
Or if is it the case I must somehow combine XYZ to declare OOB, can someone tell me how I can combine XYZ accelerations over the sampled interval or consecutive intervals to somehow declare an OOB condition?

Is it really NASA grade Rocket Science I am trying to deal with?
I do know that more expensive machines do know how to detect OOB and deal with it in a way that works well.

 

[madsi]

I have now read several articles without getting any closer to modeling and solving the problem.

I have noticed that some appnotes show that MEMS sensors can clearly detect vibrational resonances in a OOB washing machine drum..but I already know that, I've washed clothes.

Some MEMS appnotes I have read use FFT to analyze the vibrations..but this doesn't work because it requires a vibration to complete several cycles..it ends up identifying an OOB condition too late to prevent a severe OOB condition.

 

[madsi]

From the attached picture, it seems evident that the position resulting from the double integration of samples of accelerations results in a graph of linear displacement..is this correct?

But this approach requires I identify a point in my data stream where acceleration crosses over in direction to begin calculations?

If this is so, then the bottom graph show that a double integration of acceleration samples results in a plot of a scalar, and that if I can identify the point half-way up the displacement graph I could prevent a severe OOB condition.

Then I do this for all three axes? Do I first create a three-dimensional matrix of XYZ acceleration data and do a cross product matrix operation to resolve the XYZ into a single vector?

First I must learn what is the math is required, what it means to double integrate the accel data.

Am I on the right track here??

 

[Isaac0427]

From the attached picture, it seems evident that the position resulting from the double integration of samples of accelerations results in a graph of linear displacement..is this correct?

But this approach requires I identify a point in my data stream where acceleration crosses over in direction to begin calculations?

If this is so, then the bottom graph show that a double integration of acceleration samples results in a plot of a scalar, and that if I can identify the point half-way up the displacement graph I could prevent a severe OOB condition.

Then I do this for all three axes? Do I first create a three-dimensional matrix of XYZ acceleration data and do a cross product matrix operation to resolve the XYZ into a single vector?

First I must learn what is the math is required, what it means to double integrate the accel data.

Am I on the right track here??

No, at least the attachment isn't.
1. When you integrate acceleration over time twice, you are not doing ∫∫adt, you are doing ∫adt² (integrating over time squared).
2. There is A LOT wrong with your graphs. Give me your equation for acceleration (or at least if you want a constant velocity, a constant acceleration, a constant jerk (acceleration growing linearly), or something else), and I'll help you fix it.

 

[madsi]

Thanks again, Isaac0407
The attached doc is where I got the charts and double integral formula. Please see p. 23

I don't want to create any acceleration that causes an out of balance situation. I am trying to detect out of balance events using a MEMS device.

 

[Isaac0427]

Thanks again, Isaac0407
The attached doc is where I got the charts and double integral formula. Please see p. 23

I don't want to create any acceleration that causes an out of balance situation. I am trying to detect out of balance events using a MEMS device.

Page 23 is OK, however I would not use double integral, as that term is used for integrating multiple variables. Page 24, however, is very confusing. The graphs appear to me as nonsense.

 

[Isaac0427]

Actually looking at it again, I realized that it is mostly correct, however it is a little off on the scaling and such. Also, as I said, you are integrating over time squared (i.e. add dt² to a single integral). This is why:
http://www.stankova.net/statistics_2012/double_integration.pdf

 

[madsi]

Isaac0427, thanks for your reconsideration.
And such??
Little off?
Scaling..there are no scales on the graphs.
In any case, thanks for sending the doc on how to do double integration.
If I integrate acceleration, I should get velocity(rate of change of distance over time), if I integrate velocity I should get distance as function of velocity over time.

I can understand this, but I not able to express it on paper in terms of functions/equations.

Perhaps, I know you are the wiz that mom always wanted me to be, but I would be so pleased if you chose to express your answers a little more quantitatively to make clear to me what you are saying,

I know we are all busy, so much to do, so little time(I think I read somewhere that I. Newton was laid-off while he wrote his book on Mechanics).

Please try to say what you want to say more in terms of the algebra and calculus involved.

 

[Tom.G]

How about just putting three (four?) switches, mounted on the frame, around the tub. Position them to trip when the tub wobble is greater than desired.

 

[madsi]

Thanks Tom.G,
At first the idea of using three of more switches seems to be a simple and practical approach to perhaps solving the problem.
But there are reliability and cost drawbacks to the use of limit switches, especially mechanical ones:

A switch not rated for extremely hazardous environments is otherwise cheaper, but delicate.
The interior environment of a washing machine is quite hazardous to mechanical switch contacts.
Inside the inner chambers of a washing machine enclosure there is often very high humidity and dust and lint(corrosive sw. fouling mold can thrive), not to mention the hazards of a wandering insects(spiders ants and cockroaches might find a warm, dark and humid washing machine switch enclosure an ideal oasis.)

Secondly, any switches must be carefully mounted and aligned inside a protective assembly. This assembly must also allow actuator overtravel or the switch mechanisms will be crushed or knocked out of a necessary precise alignment, either instantly or over time by collisions with the immense mass of the ballasting weights/drum asm.

Making space for added switches may increase the size, weight and cost of mfg. The size of a washing machine is somewhat standardized and if added switches take up more space, the idea of not being able to allow enclosure size to be increased even slightly translates to less suspension wandering room for the drum asm. and so the drum size would maybe need to be reduced resulting in less washing load capacity.

Adding multiple rugged switches can significantly add to the BOM of a machine and increase the cost of its design and testing, it could require a MCU to be more costly, have more control pins, and there is the cost of extra wiring and safety testing issues, etc. All this affects the bottom line in a very competitive market.

Most importantly, if the displacement achieved during an OOB condition of the drum asm. is sufficient to actuate a limit switch and even though power is then cut, the acquired momentum of the drum asm is still likely going to be more than sufficient to allow it to continue on a violent and certain trajectory towards the walls of the enclosure.

There is also the problem of nuisance tripping, the more switches, the greater the chance.

Finally, turning off power by limit switches in an OOB doesn't fix the problem, it is just a surrender to panic, it may not give the control MCU any specific feedback to redress the problem, and It just might also create and demand unwanted excessive user intervention.

 

[Nidum]

@madsi

(1)

Cutting the motor power when out of balance is detected is common practice . Even machines with complex electronic control systems do this .

The more sophisticated electronic control systems detect out of balance at lower rpm and activate load balancing procedures . Thus they usually anticipate difficulties before they become serious . Nevertheless if load suddenly goes out of balance at higher rpm the electronic system just cuts motor power in same way as cruder systems .

(2)

There are standard computational methods for processing data from an accelerometer . You could just look these up .