Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: simpson36 on May 13, 2010, 12:10:49 PM
-
I am looking at acquiring the capability to do some simple dynamic balancing, and I need some help from the electron guys who are also familiar with Mach scripting.
A very simple form of dynamic balance is simply a sensor (accelerometer) that triggers a strobe light. The light is pointed at the spinning object and similar to a timing light, the strobe freezes the part at a specific point. You throw on some weight and go thru "worst/better" iterations until the balance is close enough. Not efficient, but it does work if you have the time and patience.
A new USB based balancer that runs on a laptop is over $6,000 and has capabilities far in excess of what I need, so I am looking at DIY (read cheap) alternatives. Basically, all that is needed is to fire a strobe based on movement (vibration) of the part.
First, are there any oscilloscopes or other off-the-shelf devices that can do this?
Second, how difficult would it be to design a circuit to do this?
Third, it occurs to me that Mach can take an input (tachometer) and of course can generate outputs. I think I can figure out how to fire a strobe from a 5v output pin. How would one go about connecting a digital or analogue accelerometer to Mach as if it were an index? And can the index input signal be detected by script?
-
What are you going to try and balance?
Crankshafts? What size motors?
-
Not sure how an accelerometer would help you. I believe the way tire balancers typically work is to have strain gauges on the spindle support, which will give you a cyclic strain reading that can be correlated with the angular position of the wheel, to determine the location of the heaviest point, while the magnitude of the strain gives an indication of the amount of imbalance.
For lower precision, I'd try using a proximity sensor. If the spindle is mounted on compliant mounts, you should be able to get enough motion through compression of the mounts to get a good reading from the proxiimity sensor. Since you'll inherently know angular position courtesy of Mach3, and a simple calibration with a few known weights will tell you how to scale the proximity sensor reading, the math to calculate the weight required, and the correct position, should be pretty trivial. A quadrature setup should give high sensitivity.
Regards,
Ray L.
-
Thanks for the replies. I am interested in balancing R8 and 5C machine spindles running in the 8,000 RPM range.
'Spin' balancers for tires (fairly crude actually) and similar devices for industrial dynamic balancing are designed to balance a 'disembodied' rotating part. There are balancing services that do this, the most obvious would be a tire store, but there are balancing shops where I could send the spindles out to be balanced . . . but where's the fun in that? It seems to me that it would be very useful for the Mach community to have an inexpensive but effective way to smooth out the operation of their CNC machines.
Strain gages are used for low RPM unrestrained shafts. A tire 'spin' balancer actually has one end of the shaft completely unsupported, but that application is under 1,000 RPM. I am not an expert in this field, but to my knowledge, higher speeds are done with constrained shafts and rely on vibration analysis.
In-place or 'field balancing' (pumps, motors, fans, etc) relies on vibration analysis. This is done with accelerometers which detect minute movements . . probably several orders of magnitude more sensitive than a proximity sensor, I would imagine. There are basically two methods (that I know of). Both methods use accelerometers attached to the machine. The more sophisticated method uses two accelerometers 90 degrees apart and also a photo (or proximity) sensor to count revolutions (and create a ref azimuth). These devices then tell exactly where and how much weight to add, and can do each end of the shaft separately, but simultaneously.
I am not expecting to mimic this type of balancer. The simpler form has one accelerometer and a strobe. It simply fires a strobe in responce to the vibration, thereby identifying the imbalance spot, but without quantification. In the hands of a very experienced person, only a few 'better/worse' iteration usually get the machine dialed in pretty close.
Accelerometers can be digital or analogue. A simple analogue 3 axis accelerometer is contained in a single IC and is about 20 bucks. These can be hooked up to a scope and the magnitude of the imbalance read directly. What is needed then is a trigger for a strobe. I have been talking to a company than makes an inexpensive USB oscilloscope and that looks promising, but I think it can be done for 'free' with Mach if there is someone out there with the combination of electronics and Mach3 knowledge to do it. I would be willing to work it out and share the process, but while I can contribute to the project, I do not have all of the knowledge needed, hence this thread.
-
I balance crankshafts.
Ended up buying a vibration analyzer on Ebay, Balmac model. 2 plane with strobe. Then just built my own holding fixture.
An oscilliscope will give you way to much noise from what I was told, I looked into that a bit.
Try and get ahold of a manual for one of the commercial vibration analyzers... very informative.
-
http://www.datastick.com/products/bal2000.html?gclid=CKjJ_LO3zaECFR5rswodFXUafA
If one has USD$6,000 burning a hole in their pocket, the solution is easy. Actually I just watched an old analyzer/balancer sell on eBay complete with sensor, strobe and 'lots of cables' . . . but the fellow did not know if it worked or not and I don't have the resources to fix something like that if it didn't, so I passed on it. It sold for $135
Most likely I will end up with the USB scope which has outputs and a scripting language that allows it to be read and programmed from a PC. Add an accelerometer (or two) and a strobe and it's Miller time. Plus then I have the scope for other uses.
I though it would be interesting and useful to use Mach3 since it seems capable of the task and everyone already has it, but so far it seems there are no Gurus up to the task. :'(
-
Simpson, you are the guru up to the task! And we are waiting for the results and following videos. ;D
-
Simpson,
Just a thought, but look into what the RC helecopter guys do. Balance is lost once you put something into the spindle.
Make note that the active system must be looked at, not just a single item, so even tool chatter can create a condition for resonance.
FWIW....RICH
-
I think it could be done as you originally discribe, given some "adjustment" capability of your sensor (accelerometer).
Machs current loop time is 10hrz, so in theory your max Trigger time (RPM) would be 6000 rpm, I would probably
limit your spindle to no more than 5000 rpm, since real world loop is approx 1.0x to 1.1x dependig on the speed of your comp. That is important since to act on the information with scripts, you have to capture that data, and act on it, within
the loop time.
On your accelorometer, I would pick one that you could adjust its trigger, since each thing you try to balance may have different out of balance amounts. I would put the Unknown object in the spindle, then turn it on, then I would "increase" the trigger force until the strob quits lighting all together, then come back in, till the stobe lights again, at the very edge of max detection. that would show the "Top of the bell curve" of your deflection.
At any rate: I would use the PP, and bring in the Index trigger, and use the Index OEM LED as my trigger object.
I wold then send out an Output that would power your strobe.
Here is a Brian, that would do what you want, if you can get your Accelarometer to adjust properly (a digital one).
here is the logic in Brian words(using Output 4 in this example that represents the signal to trigger your strobe light, brain is attached).
IndexSpindle(LED) ->NoOp ->Output4
Here is a MacroPump that would do the same thing in VB.
'MacroPump.m1s
If GetOEMLED(826) Then
ActivateSignal(OUTPUT4)
else
DeActivateSignal(OUTPUT4)
End If
'scott
-
Simpson,
Just a thought, but look into what the RC helicopter guys do. Balance is lost once you put something into the spindle.
Make note that the active system must be looked at, not just a single item, so even tool chatter can create a condition for resonance.
There is nothing attached to the rotating part. The sensor goes on the housing (magnetic usually) and measures the movement. The balancing is done with the spindle running free, so there is no cutting chatter or other influences.
As it happens, I used to be an RC helicopter guy. I flew a gasser. Goal was a camera ship. but I started tinkering and next thing I was into aerobatics. Still have it on a shelf in the garage.
(http://www.thecubestudio.com/pictures/Hobbies/Hobbies09WEB.jpg)
-
On your accelerometer, I would pick one that you could adjust its trigger, since each thing you try to balance may have different out of balance amounts. I would put the Unknown object in the spindle, then turn it on, then I would "increase" the trigger force until the strobe quits lighting all together, then come back in, till the strobe lights again, at the very edge of max detection. that would show the "Top of the bell curve" of your deflection.
Could this be done with an analogue accel. using a pot to adjust the input until Mach just 'sees' the high points of whatever signal is being produced?
More than likely there is a way to filter and condition the signal from the Accel, but that's over my head. That's why I was trying to scare up a Guru or two.
I think I can figure out how to fire the strobe with a Mach output.
-
Simpson,
I was assuming the R8 or 5 spindles would be used in your 4th axis you have been working on.
So comment referred to end use.
RICH
-
Hi,
I am also looking for the solution..We had a manual balancing machine for compressor/Turbines rotar etc.. I also want to make one with Mach3.. Will closely watch this thread...
-
Ok, lets make some "Assumptions", since your not providing what you want to use.
Lets assume your Analog device "accelerometer" will put out a 0-10v, or 0-5v analog signal the high end of the range would represent the "Max",
acceleration, that you trim out with your "adjustment" pot or what ever, for that piece your spinning.
I would purchase something like an Automation Direct DL05 PLC, and a 12 bit analog module that has at least 2 inputs. I would have one of the inputs that is feed
by your Accelerometer, and the other input, I would use to "adjust" high or low, the Raw data coming in on the accelerometer. This would allow you to Fine tune, your device "On the Fly" (so to speak). You would have to purchase and do some reading and set up your ladder logic in your PLC. I would set it where, I could tune (using the 2cd input with a POT, so as that the max signal will just barely exceed a "Threashold" value. At that point (Using a compare statement), I would go to a "Control Bit" as an output, that will write directly to the Serial Plugin Modbus as an Input. When that Input goes HOT, (representing max deflection on your accelerometer), I would have mach trigger an Modbus OUTPUT in a brain this is triggered off of the SErial modbus input. That would come back to your PLC and trigger an OUTPUT what would fire your stobe. Also, since the Trigger could technically be "inbetween" machs scans, I would my my Accelerometer control bit a "Set" control bit, and when the Strobe output went HOT, I would use that output as a RESET for your Set control bit.
Cheaper solutions may be that you purchase a: M1/M12 from CNC4PC (M12 has 4, 8 bit analog inputs), OR, you can purchase Peter Holmans ModIO device he also has Analog input capability. But since those devices dont not have "Programable" space on them, you would need to do your calculations in VB script in the Macropump, OR, in Brians, or perhaps a combination thereof, or a plugin.
scott
scott
-
Ok, lets make some "Assumptions", since your not providing what you want to use.
Scott, I am not providing what I want to use because I don't know what to use. I don't know enough about electronics to make an informed decision on the best way to proceed. So I ask questions that hopefully are not terribly stupid and leave it to experienced people like yourself who offer up ideas that perhaps would have never occurred to me.
Cheaper solutions may be that you purchase a: M1/M12 from CNC4PC (M12 has 4, 8 bit analog inputs), OR, you can purchase Peter Holmans ModIO device he also has Analog input capability. But since those devices dont not have "Programable" space on them, you would need to do your calculations in VB script in the Macropump, OR, in Brians, or perhaps a combination thereof, or a plugin.
Here is a good example of what I just mentioned. The solution you outlined with the PLC would be too complicated and expensive for a layman, certainly for me, but this looks like a palatable alternative and gives me another idea. I have seen some accelerometers that are already installed on boards with some controls built in. Some are pre-assembled and some are in 'science kit' form. Perhaps that would be a place to look to get the sensitivity adjustment needed.
-
I was assuming the R8 or 5 spindles would be used in your 4th axis you have been working on.
Yes, the R8 is the mill spindle and runs 8,500 RPM max. The 5C is the 4th axis and would only run in the 4,000 RPM neighborhood (hopefully), but it has a 4" steel disc brake on it that is not balanced, so even with the lower RPM, it is a larger problem.
I have had the mill spindle over 7,000 RPM so far and while it is smooth enough for most of the range, it has some harmonics at certain RPM, so there are some 'illegal speeds' that I would like to eliminate. Only yesterday I finally was able to get the 4th axis spindle up to about 3,000 (estimate). It has no harmonics, and is pretty smooth, but if you slap a 4" 3 jaw chuck on there you get quite a vibration. I could send chucks out to be balanced, but I have lots of other uses for balancing capability, so I still need a solution.
There are two other reasons that I am persuing this capability; One is that if the spindles are well balanced and I have some form of rudimentary 'in place' or 'field balancing' capability, I would then be able to balance (within reason) any part that I made on either spindle. Second is for production purposes; if a badly unbalanced part, say a casting for example, needs to be turned, it can be done much faster if a counterbalance weight is added to the fixture.
As I mentioned earlier, for my own purposes, I believe I have a solution in hand with a particular programable USB scope. However this is still a very interesting topic to discuss and it would benefit a lot of Mach users if an inexpensive and reasonably effective solution was developed that utilized the CNC PC and MACH software that's sitting there already.
-
Hi Steve,
Just wondering what kind of bearings do you have on your mill spindle and how is it cooled to run at 8500RPM?
Daniel
-
Just wondering what kind of bearings do you have on your mill spindle and how is it cooled to run at 8500RPM?
No hijacking, naughty boy.
Start a new thread on medium speed spindle cooling and perhaps I will participate.
-
has a 4" steel disc brake on it that is not balanced, so even with the lower RPM, it is a larger problem.
The spindle should be balanced as a unit, thus it would include the disc brake.
no harmonics
Yes it does, the spindle has a natural frequency ( all things do even the human body) and a critical speed, and whether you get vibration or worst yet resonance can be
dependant on the dampening effects of the mounting, bearing, machine frame, runout, etc, and could be influenced by repetative driving forces and transmissions.
Usualy you deisgn the part or system to not operate near a harmonic, in some analysis it is looked at to the 16th mode and design
changes made to move the natural freqeuncy away from a resaonant point for a given or fixed rpm. In the analysis, the harmonics of the system are used
and it is important to go through the resonant points / modes in a timely fashion such that resonance cannot occur.
You can have a spindle balanced to satisfy some spec, ( there are ISO and ASME spec quidelines for rotating machinery ). As you have become aware, as soom as you
add a different component the balance changes, runout changes, leveling plays into this, the natural frequency has changed. As a component in the makeup, a good
example is the addition of say a set screw / screws, or weight additive attachment can be used, or even a movable balance plate can be used to advantage, but ,
like all things it becomes a matter of degree and what your trying to do ( your unlikely to be removing material to accomplish it).
The 4 jaw chuck can be balanced, but then as soon as you put something into it there goes everything to poop again.
From a practical viewpoint, lets use consider what we do instinctively on say a lathe. If the lathe starts to get some vibration while we are trying to turn
something mounted off center, we slow it down and take cuts to make things symetrical and at some some point we crank up the speed and do more finishing to the piece.
Instinctively we deal with balance, runout, and vibration.
Now if you were to do some study on a part relative to it's natural frequency and as compared to degree of out of balance / loading conditions you could probably
relate back to some calcualted speed and balanced adjustments. Not something you would want to invest into if not in a large production quantity.
Time spent trying to follow a procedure may not pay when an additional roughing cut may be quicker.
I guess my viewpoints are more from a design point of view but the topic is of interest.
Back to the sidelines......
RICH
-
The spindle should be balanced as a unit, thus it would include the disc brake.
We are talking about balancing in-place, so the entire rotating assembly would be balanced by default. If the imbalance is too far from the center, then you would need to a true dynamic, which is sort of beyond what I am talking about in this instance.
The 4 jaw chuck can be balanced, but then as soon as you put something into it there goes everything to poop again.
I was talking about a 4 inch chuck, not 4 jaw
From a practical viewpoint, lets use consider what we do instinctively on say a lathe. If the lathe starts to get some vibration while we are trying to turn
something mounted off center, we slow it down and take cuts to make things symmetrical and at some some point we crank up the speed and do more finishing to the piece.
You make a valid point, but limited to something that should be and eventually will be symmetrical. I was thinking in terms of a workpiece (or assembly) of some odd shape that will have a monstrous imbalance before during and after the cut. Cutting such a part in extreme slow motion is one solution, but often that will result in an intolerably low SFM or intolerably high production time.
You would agree, methinks, that the closer you can bring the part into balance, even a very rough balance for example by bolting a bob weight to a rotating fixture or even just a face plate, the faster you can spin it for the needed operation. That's all I was talking about, in practical terms. I would tell a story about huge eccentric hinge pins, but I see my time is up... ;)
-
You make a valid point, but limited to something that should be and eventually will be symmetrical. I was thinking in terms of a workpiece (or assembly) of some odd shape that will have a monstrous imbalance before during and after the cut. Cutting such a part in extreme slow motion is one solution, but often that will result in an intolerably low SFM or intolerably high production time.
You would agree, methinks, that the closer you can bring the part into balance, even a very rough balance for example by bolting a bob weight to a rotating fixture or even just a face plate, the faster you can spin it for the needed operation.
I have followed this with interest and right off the bat let me say I know nothing about the physics of off balance rotating objects.
BUT I did find this explanation on the web.
What is an unbalanced wheel?
An unbalanced wheel is any device capable of repeated rotation in which
any given point along the circumference is either
A. prone to stop in a given position along that circumference as opposed to
any other, regardless of the amount of force inputed into rotating the
wheel, or
B. prone to rotate indefinitely according to the principle of having a
constant and recoverable force which acts on one half of the wheel more
than the other.
OK, presuming I understand this to mean that in an unbalanced rotation condition, the rate of rotation varies at any point on the circumference?
What I am leading up to is if you have a spindle that has an encoder, and you capture the anticipated position on several points on the circumference, due to imbalance, the following error on the spindle will vary, and the captured positions will vary.
Although Mach per-se is a bit of overkill to achieve this, if you already had, say Mach with a Galil card, Galil has a feature called high speed capture that possibly could be used with a high speed capture input, lets say you had 4 sensors on the spindle input to the high speed capture, they could be referenced off of the marker pulse so you could identify the quadrant that required the 'Bob' weight, which if the system works, could be done initially at low rpm?
Actually I just though of a method that may not need sensors, this is the use TE tell error at given positions at degrees of rotation.
Of course I could be right out to lunch with this theory, if so please ignore it.
Nosmo.
-
Balancing a rotating object is done to reduce the loading / forces that the machine will see. Need to really call it a system. In that light the bearings will have greater life,
improved machining, etc. You can run in an unbalanced condition as long as the forces can be tolerated by the system configuration or the task at hand.
You ever wonder why they put that adjustable collar on an Albrecth chuck? What do you do with it?
Why level a lathe? Don't they use them on a ship at sea?
Is the mill table leveled such that when a horizontal 4 th axis is used as a component of the system, if balanced, takes advantage of the additional efforts done to it?
Why balance the rotor of a motor or the impeller along with it's shaft for a pump?
How come some machines are massive and some are not but they have the same stated turning capacity?
What happens when you add all the unbalanced crap onto a balanced HP car engine? What do you get for all that additional money and effort?
Why spend the money for a hand spindle foriegn made and not spend an extra bunch of money for a name brand ( could be as much as 5X in cost)?
Just some things for ya to think about.
Lets give some vibration and resonance examples form true experiences:
- 10 foot diameter x 60 foot high vessel filled with liquid lifting it's base ring and shearing the anchor bolts.
- Watching a 8" schedule 40 stub end on a 20" pipe sheer off
- a heat exhanger still attached to it's concrete base shear the pad and bouncel up and down when resonance occured ( 4 ft dia x 10 foot long heat exchanger )
- my favorite and relative to what Steve mentioned about a unbalance during turning:
So the 1200# lathe is leveled and anchored to a 4" thick table which is bolted to a steel frame which is anchored 5 places to a concrete block wall.
An odd shaped piece is mounted to a face plate and the tail center used to add some hold / rigidiy. I will guess about 5 lbs for the piece and turning off center some 2".
Slowly machine off some of the casting, crank up the speed and refine the cuts, ok, hech turn up the speed and lets get her turned ........
Well before i could reduce the speed, the inbalance created vibrations which happen to hit a resonant frequency of the sysytem, and in about 3 seconds,
the lathe lifted the base and pulled the anchoring out of the wall. Not something you want to be standing by or experience.
Even if the piece was counter balanced before starting, the eccentric loading while spinning, the balance will vary as mass and rpm change
thus the affect relative to vibration or resonance is not constant due to a changing natural frequency of the system.
RICH
-
While the topic of balancing is interesting to discuss, the thread was intended to be specific to exploring the possibility of using Mach facilities to perform some rudimentary balancing, for the benefit of all MACH users.
I have purchased an accelerometer, programmable controller and strobe that can do the job autonomously. That is how I will proceed from here.
Thanks for the contributions.
-
You might look into the relatively inexpensive sensors used on the servo driven
washing machines. Whirlpool cabrio in particular. This is a top load model with
a glass door. During the spin cycle if an out of balance condition is sensed ,
the unit stops and repositions the load by turning the drum in one direction
while turning the agitator in the other. Then a test run up, and if green
lighted, continues up to 2000 rpm. Pretty impressive for a 100lbs of wet
cloth.
I like this thread, and hope your idea works.
Slieghtly out of balance,
Bill
-
I like this thread, and hope your idea works.
No question about it working with the components that I purchased as they already work together.
A Guru was need for intefacing an accellerometer to Mach. Unfortunately, this task was apparently beyond the available Gurus, so no balancer with Mach :'(.