MechMate CNC Router Forum

Go Back   MechMate CNC Router Forum > Miscellaneous hardware and software > Miscellaneous / General / Whatever / Catchall
Register Options Profile Last 1 | 3 | 7 Days Search Today's Posts Mark Forums Read

Reply
 
Thread Tools
  #271  
Old Wed 13 June 2012, 18:46
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Kim,

You may have given us the main reason that we get excessive "chatter" when cutting curves.

All stepper motors have an area on their "speed curve" where they resonate. When I design process control computers that use stepper motors, I make sure that the motors ramp through that "resonating" space as quickly as possible so that the "process" can be accomplished.

On a CNC machine, one axis often has to move very slowly (primary resonating part of the speed curve). When you look at how a circle is cut, each line segment involves two axes. It's fairly easy to plot things out using a spreadsheet. You'll end up with four nearly identical segments of a circle. The only thing that changes in those four segments is which axis is master and which is slave (or which is dominant and which is recessive). Using a spreadsheet that only has sixteen line segments to make a very rough circle shows the interaction between the x and the y axis.

Anyway, back to resonance. If one axis is moving very slowly, it may actually NOT move when commanded. It may suddenly move several steps out of sequence as the motor resonates.

If you plot the circle being cut, you'll probably notice that the "chatter" occurs when one axis is barely moving.

The cure for the problem is to use motors that don't resonate (or motors that are perfectly controlled by the stepper driver to eliminate all resonance), or to use servo motors with expensive feedback systems.
Reply With Quote
  #272  
Old Thu 14 June 2012, 02:30
IN-WondeR
Just call me: Kim
 
Randers
Denmark
Well, I did try to change the power to the stepper drives, it changed nothing. I can now probably only conclude that the stepper drives them selves are of poor quality.

The drives I have used is from Longs-motor in China, and they spoke highly of them, but it seems like they are just not of a good enough quality to be able to reduce chatter in my motors. So now I have opted for 2 Leadshine AM882 Drives for my X axis, hoping that this will help. Those drives can be programmed to reduce chatter in the motors, and they can be programmed via computer to give the best performance in an overall use. So croos your fingers, that I'm right about the drives.
Reply With Quote
  #273  
Old Thu 14 June 2012, 05:10
javeria
Just call me: Irfan #33
 
Bangalore
India
Kim ever thought of EM508 from leadshine? don't they have them in stock yet? the Am882 are almost 141 US in India.
Reply With Quote
  #274  
Old Thu 14 June 2012, 08:24
IN-WondeR
Just call me: Kim
 
Randers
Denmark
Quote:
Originally Posted by javeria View Post
Kim ever thought of EM508 from leadshine? don't they have them in stock yet? the Am882 are almost 141 US in India.
The EM508 can't take more than 50VDC, so it will not be usable. As the 48V PSU I have borrowed goes back, it was just to make a simple test to see if the PSU were the problem.

So I'm still gonna use the 70Vdc PSU, and the I'll just use the better AM882 from leadshine instead. I found them at around 300$US including shipping for the pair of them for the X axis here in Europe...
Reply With Quote
  #275  
Old Thu 14 June 2012, 09:35
domino11
Just call me: Heath
 
Cornwall, Ontario
Canada
Kim,
Since x and y axes are used when cutting a circle, each one slows down and resonates at their own points on the circle, would you not need to upgrade both x and y to eliminate the chatter on the circle cuts?
Reply With Quote
  #276  
Old Thu 14 June 2012, 09:59
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
Mike, here is a proposal for a definitive experiment:

Take a 200 pulse per rev stepper motor on a test bench,
with a large & accessible coupling driving to
a 2000 step/rev encoder

Drive motor with a 1/10th microstep drive. (In theory, 2000 steps into the drive will give one full turn and 2000 steps out of the encoder)
Use a variable frequency generator to get a steady stream of pulses into the drive (which will theoretically give smooth rotation of the motor)

Then, put in a "comparator" to see the difference between the input versus output pulse counts. Run at low speeds and with loading on the coupling. (I don't know how the comparator will work, but it should tell us if there is lag sometimes)

Would this make sense?
Reply With Quote
  #277  
Old Thu 14 June 2012, 13:27
IN-WondeR
Just call me: Kim
 
Randers
Denmark
Yeah I would probably need to change the y axis drive also. But since the machine has been sold, then I will not put anymore money into it than nessecary.. And the Y axis does not sound as high as the x axis does...
Reply With Quote
  #278  
Old Fri 15 June 2012, 06:41
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Gerald,

You've made a good suggestion.

I have several "b" type motors (1/2" shaft extending out the back), but I don't have any 1/2" encoders. If I can find a good 1/2" encoder, I could hack things together. The front shaft could be coupled to a load to simulate actual operation.

As far as the comparator goes, that would just be a simple microcontroller that "clocks" whenever a pulse step is sensed and whenever an encoder pulse is sensed. Then, it would drive a simple two-digit LED display that showed the variation. For example, if the stepper pulsed, it would show a '1', if the encoder pulsed, it would subtract the '1' and show a '0'. That way, it would should how many pulses out of sync things were.

The variable speed pulse generator would just be an analog to digital signal converter in another microcontroller (like the Arduino Uno). Turning a "pot" would increase or decrease the speed. Software would shape the pulses to "square" them up.

It sounds like a good project for testing resonance. I'll see if I can find a good 1/2" encoder with 500 steps. The encoders are usually quadrature, so they output 4:1. A 500 'line' encoder would match a 2,000 step per revolution motor (driven with a 10:1 gecko driver).
Reply With Quote
  #279  
Old Fri 15 June 2012, 10:54
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
From http://www.geckodrive.com/support/st...or-basics.html, there is this part:

Clipboard02.jpg

It would be interesting to see how our typical Nema 34 motors behave themselves.

How can we help with such an experiment Mike?

Last edited by Gerald D; Fri 15 June 2012 at 11:19..
Reply With Quote
  #280  
Old Fri 15 June 2012, 14:19
MetalHead
Just call me: Mike
 
Columbiana AL
United States of America
Would we have to see the motors running on an oscope?
Reply With Quote
  #281  
Old Fri 15 June 2012, 14:30
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
The best way to check a motor for good linearity is to listen to it run as its speed is slowly increased. The smoother the sound, the better the linearity.

I was lucky. I had a bunch of PK268-02AA motors on hand that have excellent linearity when used with a G203v stepper driver. (They can also be driven by other Geckodrive products with excellent results.)

AFTER the stepper drivers were tuned to the motors (read the instruction sheet that is available on the Geckodrive website for instructions on how to tune the stepper driver), those motors ran about 2X faster with much better torque than they had run with the Oriental Motor Unipolar driver that had been packaged with the motors. Unfortunately, those size 23 motors are too small for the MechMate, but they were a standard to which I could compare size 34 motors.

I had several different size 34 stepper motors on hand. I found that it was hard to tune the round motors to the Gecko stepper drives. The "square" motors worked much better. My favorite motor is the PK296-F4.5 followed by the PK299-F4.5. Both of those motors can be wired half-coil or parallel. Both of those motors, WHEN DRIVEN BY THE CORRECT VOLTAGE had advertised torque and excellent speed. Both motors had almost no 'dead spots' on the speed curve, but I'm almost certain that both motors have resonance in the lowest speeds even when stepper driver tuning has been completed.

That's why I've always recommended Oriental Motor stepper motors and Geckodrive stepper drivers. That combination has always worked for me (as long as the motors were the "square" design). Other motors have worked very well. Mike (Metalhead) had me test the motor that he recommends. It was an excellent motor. Based on the test of that single motor, I would recommend it to anyone who either couldn't get the Oriental Motor steppers or where the nearly double price of the Oriental Motor steppers was a big factor.

As far as I understand stepper theory, linearity is highly dependent upon the load and the acceleration rate used. Ramping too quickly or too slowly can affect linearity. Trying to drive too heavy of a load can affect linearity.

According to the experiences of others, a motor that has about 600 oz*in will work very well with the MechMate. If a belt-drive is used, then a motor producing 300-450 oz*in of torque works very well, depending on the gear ratio. Those motors are "middle of the pack" motors. When you study the torque curves of those motors, you'll see a very smooth torque curve EXCEPT of the resonance part of the curve. In traditional applications, the pulse generator would be configured to ramp through that resonance part of the curve as quickly as possible to avoid resonance problems. In a CNC application, unless you are cutting lines along the x-axis or y-axis or if you're cutting a 45-degree angle, you may face resonance issues, i.e. possible chatter when one axis is moving in the resonance range of the motor.

(Read the discussion above, especially the test that Gerald recommended to test for resonance to see how we may be able to 'quantify' the amount of resonance in a motor and the speeds at which that resonance occurs.)

Last edited by Richards; Fri 15 June 2012 at 14:33..
Reply With Quote
  #282  
Old Sat 16 June 2012, 00:02
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
Seeing it on a scope would be pretty, but not essential. For me it is enough to know how many pulses the encoder output lags behind the input under certain conditions, and to know what conditions cause the max lag.
Reply With Quote
  #283  
Old Sat 16 June 2012, 03:51
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
When cutting a gentle curve with two motors, one of the motors may not actually be driving - that motor could be acting as a brake, counteracting the sideways cutting force of the spinning cutter. The braking motor changes position by "slipping" or "releasing" a microstep at a time....in theory that is. On top of this we have the motor wanting to go into a natural detent position (What we feel when we turn a motor by hand). I have always been suspicious that microstepping drives could really produce smooth motion under these conditions.
Reply With Quote
  #284  
Old Sat 16 June 2012, 07:04
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
I've been playing with a spreadsheet to show the number of steps that a MechMate would move as a circle is being cut. The PDF shows the first few lines of that spreadsheet when a 5" circle is being cut. Remember, a stepper motor can only make full steps. Just as Gerald said, one axis is not moving (and possibly going to its natural detent position).

The top portion of the spreadsheet shows how far an axis moves for each step when a PK296A2A-SG7.2 geared motor is used.

The bottom portion of the spreadsheet shows the distance that each axis moves per angle increment (as specified in the top portion of the spreadsheet).

From this little example, it looks like the "chatter" could easily be caused by "natural detent" + multiple "moves" with no actual movement on an axis.
Attached Files
File Type: pdf partial_x_y.pdf (31.6 KB, 32 views)
Reply With Quote
  #285  
Old Sat 16 June 2012, 09:41
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
I had to travel about 100 miles to fix a problem at one of my computer customer's installations, so I had time to think a little more about "chatter" and about full-step vs micro-stepping.

Let's assume, that under certain conditions, that the 0.000273" micro-step position can be out of place so that it turns into an 0.00273" error (although it would probably be 1/2 of that amount because the motor would probably "slide" to the closest natural detent).

Let's assume that BOTH axes could "slide" into full step positioning under certain conditions.

Then, looking at the height of the "chatter", does it look like the "hills" and "valleys" are about 0.00273" high (or low)?

I no longer have a CNC machine to play with, but when I had my Shopbot PRT-Alpha with retro-fitted 7.2:1 geared motors that had 1,000 steps per revolution instead of 2,000 steps per revolution, I remember that the "chatter" was minimal, but still clearly visible. Just how much chatter was there? I don't know, but there was enough that I had to do some hand sanding before I could send the work out for painting.

Given the fact that we may have discovered what causes most of the "chatter" and given the fact that no one is going to throw away his stepper motors and stepper drivers to solve the problem, what can be done to minimize the problem? Think of motion pictures and persistence of vision. When we watch a motion picture, we're really seeing "still" frames racing by at 24 or 25 frames per second. Our brain can't process 24 or 25 frames per second, so we have "moving pictures" caused by persistence of vision.

Perhaps that same basic principle could be used when cutting curves and circles. Most stepper motors have listed (by symbol on the torque charts) the maximum starting speed for that motor without ramping. Let's say that that speed is 400 full steps per second. With a Gecko stepper driver, that means that we could try running that motor, without ramping, at 4,000 steps per second and see if it stalls or if it runs smoothly. If it runs smoothly, then, in theory, we could run that motor at 4,000 steps per second without ramping. Looking at the chart in my post above, that means that would could cut at 1" per second without ramping. Using the "persistence of vision" principle, we could try cutting at 1" per second and hope that the natural vibration of the machine, of the motors and of the cutter would produce the "smoothest" possible cut, that that speed would let everything work together.

There is one more thing that much be addressed. The CAM program (Vectric V-Carve or whatever you're using to create the tool path) should be calibrated so that the minimum move is equal to the distance that the stepper motor moves an axis each time the stepper is "stepped". Trying to get 0.0001" moves from a geared motor that could only produce 0.000275" of resolution will not work. On the other hand, setting the software to create arcs with 0.01" segments would not let the motor do all that it is capable of doing.

Last edited by Richards; Sat 16 June 2012 at 09:44..
Reply With Quote
  #286  
Old Sat 16 June 2012, 12:31
sailfl
Just call me: Nils #12
 
Winter Park, FL
United States of America
Mike

Where in Vectric software does it allow you to calibrate the software to the motors? I use Aspire.

I don't believe that is available.
Reply With Quote
  #287  
Old Sat 16 June 2012, 12:41
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Nils,

I don't know. I'm 100 miles from my home so I can't check things from here. I'll post when I have a chance to look at the V-Carve settings.
Reply With Quote
  #288  
Old Sat 16 June 2012, 13:17
sailfl
Just call me: Nils #12
 
Winter Park, FL
United States of America
This is an interesting discussion on motors. Thank you for sharing your expertise with us.
Reply With Quote
  #289  
Old Sun 17 June 2012, 09:36
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Nils,

I've got egg on my face. The Circle Segment Resolution is part of Shopbot's SB3 program. I was certain that V-Carve Pro also had a parameter setting, but my old mind failed me, proving, once again, that memory is fleeting.

However, I ran some circle tests on my test bench with my two "best" motors, a PK299-F4.5 and a PK296-F4.5 attached to the X-axis and the Y-axis. I drew a single circle in V-Carve Pro and then created a tool path file. I exported the file as a Mach-2/3 compatible file. The circle had an 8" diameter. V-Carve Pro created a file that used four arcs to cut the circle, using the "I" and the "J" parameters.

When I ran the cut file, there was a very distinct "stutter" from the motors at the ends of each arc. The motors ran smoothly, as expected, when they approached the 45-degree portion of each arc.

Of course, this test was not scientific, but my fingers and my ears detected harmonic problems when cutting the circle, which probably would have caused "chatter".

I looked some saved posts from the Shopbot forum where I had posted some circle test cuts (years ago). The "chatter" on those pieces fairly well matched what my fingers felt and my ears heard.
Reply With Quote
  #290  
Old Sun 17 June 2012, 10:15
sailfl
Just call me: Nils #12
 
Winter Park, FL
United States of America
Mike,

Since this has been an interest I have had for some time, where is the chatter coming from. It sounds as though you have determined that the software it self is part of the problem.

I also notice there is a jerking or stutter while cutting but I have attribited to the machine.

If this is the case how do we fix this problem? Pick a different post processor in Aspire.

The Vectric people are not eager to admit there are problems in my personal experience but I am willing to address the problem.

What do you suggest or do we need to do more testing.

MetalHead or domino11, may be this motor topic needs its own thread and we release it from Kim's build thread.
Reply With Quote
  #291  
Old Sun 17 June 2012, 11:12
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Nils,

I don't think that there is much that can be done with either the software or the motors.

We use stepper motors because they are accurate and inexpensive. One of their inherent problems is that they "resonate" at certain low speeds. That is a characteristic of stepper motors that can be minimized, but not eliminated, at least if we need both low speeds and high speeds.

The CAM software (tool pathing) can be optimized (by the software engineers who wrote the software) so that it generates arc segments that are suited to the minimum step distance of the stepper motors.

The big problem, at least as I see it, is that the tool path for cutting arcs (circles and curves), if written as straight-line segments, would require partial steps most of the time. A stepper motor cannot make partial steps. It either makes a step or it doesn't. Add the "resonance error" to the equation, and sometimes that stepper motor will make or not make that step because of electrical/mechanical "bouncing".

In normal cutting (straight lines), resonance is not a problem. The stepper motor shoots right past that portion of its speed curve and no problem is ever detected. However, cutting curves and circles keeps one or both axes in that partial-step and "resonating" portion of their speed curve - and we get chatter.

What about using a servo-motor? Well, there are some huge problems with servo motors. I have one large servo motor that is about 3X longer than a PK299 stepper motor. It only develops 225 oz*in of torque, so it must be geared at least 3:1 to handle the expected load. It requires a 7A, 70VDC power supply to operate. It requires an encoder. If something happens to the encoder, it will move the axis at the fastest speed possible until it crashes against the stop. So, it would take 4X the power supply, at least 2X the cost, and it would only solve the "resonating" problem. A servo, just like a stepper, cannot make partial steps (at least not the brush type steppers that most of us would use).

One of my computer customers had a new sign made that had acrylic letters glued onto an acrylic base. Those letters were cut on a "big iron" machine that cost over $300,000. Guess what? Those letters had about as much "chatter" as the same sized letters would have had on my $10,000 Shopbot.

So, I wouldn't worry too much about eliminating "chatter". My goal would be to see what depth of cut and what speed minimized the problem. I would also see if a "clean-up" cut would help reduce the "chatter". After finding the best possible solution, I would just live with it and use sandpaper as necessary to handle the remaining "chatter". That's basically what I did when I was cutting. I gave my customers the choice of sanding the work or having me sand it at $30 per hour. Most of them sanded the parts themselves, after all, they had to do some hand work before painting. As long as we agreed on the quality that they could expect and the amount of hand work that would be required, everybody was satisfied.

Every machine has limitations. We can try to make the machine do something that it cannot normally do, or we can find the limitations of the machine, and then use another process to handle the remaining problems.

Last edited by Richards; Sun 17 June 2012 at 11:18..
Reply With Quote
  #292  
Old Sun 17 June 2012, 12:53
hennie
Just call me: Hennie #23
 
Roodepoort JHB
South Africa
I had a situation about two month`s ago where I had a client that wanted some arcs cut.I did the job and had some chatter that can be sanded.The client wasn`t happy and went to another chap that cuts with a multicam machine with vacume and all story`s .They had a lot to say about the chatter but while back at the Ranch there is 3 workers sanding away.Moral of the story anything to say when competition is lurking around the corner.

Sean D. I think you would know who I am talking about.They do what you do in your part of the world.
Reply With Quote
  #293  
Old Sun 17 June 2012, 19:48
KenC
Just call me: Ken
 
Klang
Malaysia
Consider the money & effort to build MM range of machine. Asking for 100% chatter free cut all the time on any material...
I consider that irrationally paranoid.
Reply With Quote
  #294  
Old Sun 17 June 2012, 23:27
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
This discussion has been repeated many times. We have yet to prove, and accept, that micro-stepping of stepper motors simply doesn't perform perfectly under all conditions. All of our theory on cut quality is based on about 2000 microsteps per revolution. If that is not true, then it will be useful to know how far our motors can drift away from the theory.
Reply With Quote
  #295  
Old Mon 18 June 2012, 11:04
Allegheny
Just call me: Brian
 
Massachusetts
United States of America
If I understand what Mike is saying - chatter occurs at the resonate frequency/rpm of a given motor. Any attempts at minimizing chatter, therefore, should revolve around keeping that stepper from turning at that rpm. This resonate rpm, again if I understand his reasoning, tends to be on the low end of the motor's range, hence we see chatter in an arc or circle when the requirements of the cut dictate that one motor or the other moves very slowly. Is that correct?

If so, how about finding a way to never allow a motor to spin below that resonate rpm - i.e., always keeping it at or above whatever sweet spot it might have? I don't know how practical this might be, but how about using an electronic clutch that could be programmed to "slip" when slower cut speeds was needed?

Brian
Taxachusetts
Reply With Quote
  #296  
Old Mon 18 June 2012, 11:12
danilom
Just call me: Danilo #64
 
Novi Sad
Serbia
It often can be heard, that rough motion. In those cases I up the feedrate a bit to avoid the spot where it resonates. It helps a lot to listen carefully sometimes on test cuts.
Reply With Quote
  #297  
Old Mon 18 June 2012, 11:20
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
Brian, when you cut circles, the speed of a motor has to go from zero to max, and it has to cover all speeds. You cannot decide to stay away from a certain speed.
Reply With Quote
  #298  
Old Mon 18 June 2012, 11:25
Gerald D
Just call me: Gerald (retired)
 
Cape Town
South Africa
PS. I don't think this has anything to do with a motor's natural resonance. I think we are talking about a motor's ability to use micro-steps to fight its inherent detents. We see an oscillating battle between the micro-steps and the magnetic detents and this oscillation sometimes resonates with the external machine's structure (gantry, etc.).

Classic resonance of a motor is internal to a motor and independent of external structure stiffness. No?
Reply With Quote
  #299  
Old Mon 18 June 2012, 13:11
IN-WondeR
Just call me: Kim
 
Randers
Denmark
Well to just let you all know.

I got the 2 Leadshine AM882 Drives today, hooked them up, and viola, no more resonance in the motors.
I actually had to take a look at the motors to make sure they actually moved when I tested them off the rack..

Incredible drive, which has a lot of settings. The fact that it can be fine tuned via the computer, just makes for an even more versatile drive.

I can highly recommend this drive for those having high humming sounds from their motors at low speeds.
Reply With Quote
  #300  
Old Mon 18 June 2012, 16:30
Richards
Just call me: Mike
 
South Jordan, UT
United States of America
Whether we use a Geckodrive that has 10:1 microstepping or whether we use the driver recommended by the stepper drive manufacturer (and get 200 steps per revolution), we really have very little control over how a stepper motor operates.

IF the motor "slides" to its nearest "detent" position (+/- 5 microsteps), there is really nothing that we can do about it. It is a characteristic of the machine.

We need to learn the machine's characteristics and then we need to realize that EVERY machine has limitations. My Delta Unisaw is NOT 100% repeatable. When I move the fence, I might be "off" by 1/16" or 1/32" because of the way that I see the numbers OR the way that the fence "settles" when I push the handle down. My bandsaw is even worse. I can clearly mark the cut line on the material, but the bandsaw blade may "wander". Even when I'm sanding with my 10" disk sander, I may push too hard and get a "flat spot" on the material.

A CNC machine has inherent "characteristics". Ignoring those characteristics will not improve the cut. Blaming someone else for those characteristics will not improve the cut. Learning that there are limitations to what the machine can do will let us tell the customer what to expect and then bill accordingly.

I wish that stepper motors did not have "resonance". I wish that every step of a stepper motor moved the axis exactly the distance expected; but, that is fantasy. Peter Pan lives in fantasy land; we do not. We need to learn the limitations of our tools and then accept those limitations. Endlessly trying to make things perfect will destroy our budgets and take longer than the time allotted to each of us on this earth.

Last edited by Richards; Mon 18 June 2012 at 16:32..
Reply With Quote
Reply

Register Options Profile Last 1 | 3 | 7 Days Search Today's Posts Mark Forums Read

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Working on Belt Drive - Randers Denmark IN-WondeR Motors & their mountings 0 Sun 09 January 2011 07:23
Ron's Spin - MM inspired cab. guy Archives 49 Wed 16 December 2009 13:07
Anyone have the MM mounted on wheels ? Stepinwolf General - MM Build 13 Mon 14 December 2009 02:33
My new MechMate, is now "officially" cutting! #27 - Randers, Denmark mikefoged MechMates already cutting 55 Wed 07 October 2009 07:36
How is Z-slide mounted to Y-Car? fabrica Archives 12 Mon 08 January 2007 06:13


All times are GMT -6. The time now is 10:18.


Powered by vBulletin® Version 3.8.3
Copyright ©2000 - 2024, Jelsoft Enterprises Ltd.