Reply with QuoteAwesome! So how accurate are your cuts? Did you have to tram and adjust the mill significantly?Originally Posted by asifjahmed
Thanks
The primary function of microstepping is to provide smoother motion, NOT to increase resolution. The ultimate resolution, and accuracy, of any CNC machine is FAR more a function of the mechanical characteristics of the machine than it is the drive system. In the real world, it is quire difficult, and VERY expensive, to even get close to +/-0.001" true accuracy and resolution. It is done on large commercial machine by FAR more expensive components, a FAR more massive (stiffer, with better damping) machine structure, and more complex software, which compensates for some of the many remaining sources of error and inaccuracy. Having a drive that is theoretically capable of high resolution will do nothing to improve the overall system accuracy or resolution, unless the many errors contributed by the many other parts of the machine are all already below the error contributed by that drive. On these machines, you typically have ballscrews with an accuracy of perhaps +/-0.003" per foot. Flex of the machine itself can be several thousandths of an inch. Thermal expansion can contribute that much more. You always have stiction, and backlash, which can completely consume small movements. So what does reducing the step size from 0.0001" to 0.00001" actually accomplish? Now, you can try to address all these deficiencies, but a single high-precision ballscrew will cost more than you paid for the whole the machine, and the other parts you'd need to swap out won't be much cheaper.
Awesome! So how accurate are your cuts? Did you have to tram and adjust the mill significantly?
Thanks
So far my machine looks to be within +/- .005" for the X/Y axis, and somewhat better than that for the Z axis. However, with the Z axis I experimented with different size balls in the ballnut, and was able to stuff .1257" balls in there, where the X and Y axis have .1250" balls. I have of course placed an order for more .1257" balls for the other axes.
See my build thread for a ton of pics - http://www.cnczone.com/forums/bencht...uild_lots.html
As for tramming, I honestly have not even gotten around to doing a full tram yet. I did adjust the column/table tram by placing some shims between the column and base (I have the solid column version of the SX2), and it was then trammed to within a thou. However I have since discovered that there are more adjustments in the head that I should have checked first. I can tell when face milling that my tram is not 100% perfect (seems like 1 of the 2 cutters on my indexable mill wears faster than the other). I will have to take the plunge and tear down the mill and adjust everything. I also want to make some slightly oversized brass gibs to replace the stockers. I am hoping that this will help with repeatability.
HK is correct, you have to have a machine that is capable of utilizing the increase in microstepping. 99.9% of desktop and benchtops only benefit from high microstepping to smooth out the motion if need be. My bench top mill on the other hand costs more then my some new cars, so it can utilize the increased microstepping to its advantage. There is a % of loss but its umm.............very small LOLOLOL most wont get that, but its ok.......
Just for the record, my jewelry mill with THK ballscrews and .9 degree steppers runs at 10 micrstepping with the keling non digital drives.
My granite based desktop (difference between desktop and bench top is size.... think the size of a X1) its THK everything its tight and solid so it can use it.....
AND if you are using steppers in "your highly accurate machine" it isn't all that accurate anyway. Mariss explained on the gecko forum that if you attach a laser pointer to the shaft of your .9* motor and aim it on the wall, you will see larger and smaller steps as it goes through the motion of micro stepping.
so micron level machining is not accurate enough? works for me.
I actually have some positioning issues with by setup whereas the steppers skip 7 of the 8 microsteps and jump like crazy on the 8th. This really drives me nuts because it completely desynchronises X from Y and I'm getting some weird squished and deformed ovals instead of circles and if it has to change directions several times on a small and complex shape part, there's really no telling what kind of deformation I'll end up with - it all becomes just random.
So, anyhow, microstepping may be nice for keeping the noise and dynamic load on the machine low but for accuracy as such, it may be detrimental rather than helpful, despite what the term implies.
I'm about to mount a small mirror to the shaft and try the laser pointer trick. No space for the pointer itself but a small mirror will fit and I really only need just a fraction of the whole rotation to get the idea of just how bad the things are...
I am interested in seeing a picture of this mill that you use. Do you happen to have any that you are willing to show? I have never heard of this type of machining being done successfully in a home-shop environment, but would love to see it if it's been done.
How have you *proven*, by direct measurement, that you're actually getting the accuracy you claim? Calculating the theoretical resolution and accuracy is all well and good, but it does not mean you're actually *getting* that accuracy and resolution on the actual machine in actual use. It is also extraordinarily difficult, and VERY expensive, to achieve true "micron level" accuracy in ANY real world machine, and it requires, among MANY other things, a temperature-controlled environment for the machine to negate thermally-induced errors, which are FAR exceed "micron level" (~10 microns per degree C for cast iron, higher for steel, MUCH higher for aluminum!). So, if you're not operating your machine in a tightly temperature controlled environment, you are most certainly NOT achieving "micron level" accuracy. And, the fact is, steppers are NEVER perfect. Microstepping NEVER creates perfect, equal-sized steps, and the higher the microstep ratio, the greater the % error in step size. The error is also highly dependent on the load, and increases significantly with increasing load.
Regards,
Ray L.
I'd also be interested to know, if this is a hand-built machine, how you would square up the axes to achieve such accuracy. The equipment to do so would be well beyond reach of most machinists, hobbyists, etc.
Yeah, that's exactly the point. I also thought I was getting 10 micron accuracy which would have been OK for me but looked at what's actually happening and turns out to be 8 times rougher. The biggest part of the miscalculation was to think that 1/8 microstepping is giving me 8 times the accuracy which it does not.
In fact, I'm not too sure if there's a way to know exactly the accuracy of a machine, especially one without feedback, such as driven by steppers. You can measure some surfaces but if they are complex enough, measuring them would be a whole task in its own right. My experience has been that complex shapes is exactly where the inaccuracy starts creeping up. You'd do a few perfect rectangular cuts and then go into a series of arcs which would get all messed up.
[B]109:24:23[/B] [I]Armstrong[/I]: That's one small step for (a) man; one giant leap for mankind. (Long Pause)[SIGPIC][/SIGPIC]
Microsteps are in-between positions that may only have fractional torque holding and going to the next step and varied accuracy in positioning. The 200 full steps are pretty much reliable, half steps are decent too.
Microstepping is for smoothness of operation. Take a big ratchet and give it 8 notches per turn and turn it fast. It's going to make a hell of a lot of noise and vibrate a ton. Take that same ratchet and give it 48 smaller notches and it will be much less noisy and feel smoother as it moves at the possible expense of the amount of torque it can hold against. It's easy to hear and feel the difference in a microstep driven motor and a full or half-step driven motor. But they may not actually move the table until a couple of microsteps are taken because they lack the full torque to move the load.
I microstep at 8 so 1600 steps per rotation and 5 rotations per inch or 8000 steps per inch, but I only count on the machine being able to position within 0.001". In most cases it *can* position to 0.0005" but there is no way in hell it's good to 0.000125" as the microsteps per would have you think.
ETA: I guess I missed the page 2 discussion before posting. I'll concur with the above. While I know I can get it to move 0.001" on command or very close to it. I can't get better than +/- 0.003 without going to some effort in how I approach the job. Use the whole table and it gets worse.
CNC: Making incorrect parts and breaking stuff, faster and with greater precision.
This is the core of the issue with micro-stepping while micro-machiningIf the CNC software is setup to assume that 1 step command = 1 step movement (and if there's no positional feedback it's not going to know it's not true) then it will issue, say, 4 steps on X, then 11 steps on Y thinking it's making a slanted line whereas in reality it will make a vertical line instead because the torque of the first four microsteps is not enough to actually physically move the table in X direction.
By the way, is it possible with EMC2 (or some other CNC software for that matter) to make sure it does not issue a step command unless the movement needs more than 8 steps? In other words, if I wanted to still use the smoother movement afforded by microstepping but did not want to allow for random and unpredictable movements on single micro-steps?
[B]109:24:23[/B] [I]Armstrong[/I]: That's one small step for (a) man; one giant leap for mankind. (Long Pause)[SIGPIC][/SIGPIC]
Hmm, If I could remember how I did it.... I think I set it up not to calculate a position beyond 0.0000 so no matter what you tell it to do it rounds up to at least 4 microsteps. It's something I vaguely remember. I'll see if I can figure it out.
CNC: Making incorrect parts and breaking stuff, faster and with greater precision.
here is my machine.... polished granite base and column thk rails trucks and screws. I should also note that this machine setup is also used world wide in the medical field for cancer research, micro-fluidics, micro structures ect.
I am also going to say that its not entirely the machine capabilities that can produce good quality parts, but the person running the tool.
There is also an article written by the Department of Micro and Nanotechnology. In the article it states that and i quote.
The step motors controlling the stage employ micro-stepping, and a XYZ resolution ofaround 1µm is achieved. The repeatability of the system, including thespindle run-out, has been experimentally found to be less than 5µm when working on areas a few centimeters across, which is the dimension typically used for microfluidic systems.
Now I know someone is going to spout off that the accuracy error along the entire length of the screw ect ect.
I got called out, and I delivered. I have been here a long time helping as much as I can when I can. Im not the smartest person here either. Definitely not a real machinist.
That is a very nice machine. It looks like a MiniTech or some variant of such. I must say, however, that I thought we were talking about "micron-level" accuracy, and that is why I inquired.
I don't know if your response was directed partially at me, but I wasn't trying to challenge you - I was only interested in seeing the machine.
it is a minitech GX machine yes. I dont like to spout off manufacture names to much.
Micron lever accuracy is how you use the machine. Mach3 has the wonderful rounding feature where if it cannot make the requested move it stackes them until it can. I have been involved in some research with a university to find out the average stacking of different machines. When I say different I mean identical machines sitting side by side (all things being equal, no two machines are the same) we found that even tho the same components are used, each setup has different characteristics. Finding the rounding in each machine over different distances, then using gcode to compensate for the rounding errors.
So with that, it is possible to take these machines to micron level accuracy, even if it means going backwards to go forward. I hope that makes sense. it didnt to me at first. Also that is taking into account for the tool diameter ect
here are some pics of some test cuts I did for a university in Massachusetts.
the pins are 10 microns in diameter 20 microns tall/deep. The project was measured under an electron microscope. The final project was done in a special substrate that was positioned. Then each pin was crushed and the forces to crush the pillar was measured.
These were test pieces done in aluminum to get the code right.
I see what you mean. What are the encoder and ballscrew specs, just out of curiosity? So it makes several positioning "attempts" until the requested move is verified by the encoder?
thats the thing, Its an open loop system. no encoders.
