Machsupport Forum
Third party software and hardware support forums. => SmoothStepper USB => Topic started by: cv580 on January 11, 2010, 07:54:21 PM
-
I have a smoothstepper on a large mill with stepper drives and what ever code that is being run through the smoothstepper it will not make the whole stepping distance ie: if the code is X-1.250, Y1.250, Z-.750 the actual distance the smoothstepper will make is X-1.2499, Y1.2502, Z-.7499. Can anyone explain why this happens and what the fix is? If I run the machine thru a parallel port this does not happen. The drives are being run in step and direction mode. I have used a dial gauge on the axis and it confirms the over travel and under travel with the smoothstepper.
cv580
-
What are your steps per unit?
Hood
-
Steps per unit are 10,000 (ball screw makes 5 turns to 1 inch of travel), I have tried different step per unit settings but then distance travelled becomes incorrect in a big way. The undertravel or overtravel remains constant until the axis passes thru 0.000 and then the change of plus or minus a step will happen.
cv580
-
I cant remember exactly what it is but Greg did explain it to me a few years back, its something to do with the way things are calculated but it doesnt accumulate. In fact I cant even remember if its just a display thing or actually a real value but either way 0.0001 is an insignificant value. I agree it is f a visual nuisance but you soon forget about it when you realise how small it is :)
Hood
-
It is actually .0002" as it will go plus or minus .0001" and it can be more than a nuisance especially when you look at the DROs while running code and see a bunch of 9's because it has rolled back .0001", also I have been doing some very small engraving of serial numbers and losing a .0001 step makes for some difficulties also so it does depend on what you are doing will make it a small or big issue.
Does Warp9 still support the smoothstepper as I have tried to send them e-mails about this problem but never receive a reply?
cv580
-
Hey cv580
I have the exact same problem with my smoothstepper, I have been told that it is no big deal as it is such a small step and there is more back lash in the axis than what travel the smoothstepper cannot make but I disagree with that thought as we are now working with electronics that can be super super accurate, basically I see what the smoothstepper doing here is a form of electronic backlash. I have talked to other people that are having this issue with the smoothstepper and it is only happening in the step and direction mode, I know one shop that have this issue and they were talking to Warp9 about this and the explanation came back something along the line about it being a problem with the algorithm not being right for the step and direction mode and using a hysteresis for the step and direction mode which is not used for the quadrature mode, from this I gather the smoothstepper is not really designed for step and direction mode. I do hear a lot of comments of it is just a small step and not to be concerned about it, which I do not agree with because here is a product with great potential but it is put on the market as is with no feedback from the machining community then it will remain a half built product that is not right, I do not know of another motion controller on the market today with this sort of lead and lag. I gather that the designer of the smoothstepper is not a machinist and if everyone in the machining world comes across with the attitude of plus or minus .0001" is not a big deal then nothing will be done by Warp9 and you will be stuck with a motion controller that has lead and lag. He has to know what is not working properly, so yeah + or - .0001" steps is a concern for some of us chipmakers. As for support for the smoothstepper from Warp9 I am not sure what their take is on this issue, I understand they are a moderater on this forum maybe they could let us know what their thoughts are!
millmaster
-
Your steps per unit are 10000, so if all was perfect ( which it is never true ) the very best you can hope for in
position is 0.0001" plus or minus. To do better than the 0.0001" the mechanical axis needs a required steps per unit
at 20000.
It's an emotional issue to see say 1.9999 instead of 2.0000.
I never had a display problem with any of the machines, SS or PP, if the steps per unit was reasonable above 10000.
Do small engraving and say it makes a difference on small engraving of letters.......how small and how deep and what is your runout at the tip of the point. Best i could ever do is 0.0005" tip radius which won't even last in scribing butter. And you will not scribe a single point / line width of 0.001" since there will be spindle runout and ovality of the point. Sorry, just don't buy a practical impact on lettering because of the .0001" . FWIW my engraving
machine requires 253000 steps per unit , slow, but accurate as can be in positioning. Small quality letters or numbers .010" high are a real challenge, almost not practical.
RICH
-
can I ask the specs of your machine please?
Hood
-
Hey Rich
I think we are looking at a electronic issue here, you are thinking of a mechanical issue with the machine. What mode are you running your machines in? Step and direction or quadrature? The mode will make a difference in what the display shows.
millmaster
-
millmaster
what machine do you use the SS on?
Hood
-
This just a rounding issue with the display of the axis DRO and the fact that you have a low step resolution. If you have 10,000 steps/inch you'll never be accurate to 0.0001" (ten thousandths). It has nothing to do with the actual SmmothStepper HW.
Seriously folks getting your shorts in wad because a DRO shows a 0.0001" discrepancy is ridicules. 99.999% of the machines that folks have on this forum can't resolve down to a ten thousandths of an inch. Also 99.99% of folks here do not the equipment to accurately measure a ten thousandth of an inch. You'll have loads more variation in your part from tool deflection, tool wear, your machine heating/cooling, etc than 0.0001" !
-
I suggest the people that are worried about this have a look at some of the big names in machining and see what their positioning accuracy is, for example I just looked an Okuma MD series which by all accounts is a very nice machine, positioning accuracy is plus/minus 0.000120"
Hood
Edit
HASS VF11, first one I looked at from them, positioning accuracy plus minus 0.0003
-
This just a rounding issue with the display of the axis DRO and the fact that you have a low step resolution.
Echo the above.....
Millmaster, Step and direction, no encoders, just plain Jane machines here.
I know how you feel :o as i gave been playing around with the an indexer / rotary and when it shows some fraction of 45 deg like 44.9987 it becomes an motional issue..........practicaly it's meaningless....but wanted to have more exact display so bumped the gearing and now I am non-emotional. ::) But it's still may be meaningless depending on what your doing with it. ;)
Off to the sideline,
RICH
-
Hood,
I have a smoothstepper on a bridgeport style mill with the fourth rotary axis.
millmaster
Jeff and Rich
Can you explain why the DRO rounding issue disappears when I run the mill through a parallel port instead of the smoothstepper?
Rich
Just curious (seeing as you do not have display issues) what are your step settings for the axis on your machines with the smoothstepper?
millmaster
-
millmaster,
On the lathe the X axis is 12700 steps / unit and i get that silly .0001" difference with the PP or SS.
It means nothing, and there is .0003 or 4 backlash anyway and there is nothing that can be done to eliminate it.
All the other axes on the other machines are over 20000 and never had a DRO problem. I am currently not using the SS since i got a new pc, can swap very easily if i wish. Thus i haven't even tried the latest SS plugin.
What i remember from what someone posted some time back on the DRO rounding, if the dro display was changed to show 10 places + you would see how things are being rounded. If a number was rounded, that small difference would be added to the next move .....don't take this as an absolute statement and it was some time ago and may be out of context. So no i can't tell why that is happening as it's beyond my understanding of the details. Greg would need to be the definitve source on the rounding manner for his HW.
Are you loosing a step / pulse ( .0001").....A friend came over with a custom built piece of electronic equipement ( borrowed from his workplace lab ) when we were investigating something on threading, i don't have anything that will measure a single lost step electronicaly. I doubt most folks do.
For position / mechanical movement i do it opticaly as it is easier only because i have the equipment to do it and it is much better than any indicator .
RICH
-
Hood,
I have a smoothstepper on a bridgeport style mill with the fourth rotary axis.
millmaster
Yes, so even if it is a real error rather than just a display issue you are never going to see it or be bothered with it.
I do agree it was at first an annoyance seeing it, as I see it on the lathe at times, but for me it is 0.0001mm which makes me think it is a display issue as that is 0.000003937"
I think there is a good chance if you increase the digits displayed after the decimal in your DROs you will see the 9's go further back.
Hood
-
when we were investigating something on threading
This had to do measuring how many pulses were sent to the stepper as compared to output pulses from the index
relative to triggering the axis move. The electronic device measured to so many parts per million of pulses and could
add, subtract, or totalize the pulses from three or four inputs.It was used because,even very high end ditgital scopes are not
as good as that device / it's a custom counter. So you could see the affect of something put in line and compare and see if a pulse was lost. Something proved with pulse counting, was how belt tension and allignment can create a non movement ( call it backlash in the timing belts) of the axis which you would think was due to the ball screw or bearings when in fact it was lack of belt engagement to the pulley and even had an effect over time period ( small but real ). So where's that lost ( for lack of a better word ) pulse relative to display and mechanical movement can be get very complex. I'll leave the details to those that know what they are doing and just accept at times that all is well. Again i find out how stupid i am when i get into the expertise or discipline of another.
Not trying to curtail your quest for answers, just sharing experience FWIW,
RICH
-
I disagree with Jeff about it not being a problem with the smoothstepper, if I run the machine with the parallel port I do not see the leading and lagging on the DROs, when I run the machine with the smoothstepper then the DROs do not match the g-code hence the .0001 lead and lag which leads me to think there is a issue with the smoothstepper. Can you shed some light on that Jeff?
I have three other NC machines, 2 machines have MicroKinetics controllers and the 3rd has a Centroid control, none of these machines have a lead or lag issue, the DROs on them match exactly to what the g-code that is being run at the time (makes for easier troubleshooting g-code issues when things appear to be going sideways), it is just the smoothstepper setup that gives mismatching readings between g-code and DROs, if it was a rounding issue wouldn't other controllers do the same, everything still points to the smoothstepper being the culprit which I think could be made right.
millmaster
-
There is no doubt that the issue is in the SS and the reason you see it with the SS and not the PP is that when using the PP Mach puts the numbers in the DROs, when using the SS it tells Mach what to put in the DROs.
Obviously if this issue is not present with Quad outputs then it is the way the SS does things with step/dir only.
Whether its a real position error or just a DRO error I dont know but as said above due to me occasionally seeing it and it just happening to be 0.0001 when I use mm makes me suspect it is a DRO issue, annoying from a visual point of view but no way a problem with the machining even if it was a real error.
If it can be fixed then I am sure Greg will get round to it but it will be low on the list I suspect.
Hood
-
Finally someone admits that there is a problem with the smoothstepper. Now how do we get it fixed?? Why would any problem with the smoothstepper be low on Greg's list?? I see Greg is a moderator for this forum why can't he come on the forum and explain everything from the problem to a fix?
millmaster
-
millmaster, read my second post in this thread I said it was something about the way it was calculated by the SS, so dont know why you say
Finally someone admits that there is a problem with the smoothstepper
Now to answer your next question Why would any problem with the smoothstepper be low on Greg's list??
The reason it will be low on the list is because there are other things much more important to users that are needed to be worked on first. Are you seriously telling me that you can measure this +- 0.0001" out of dimension on anything you have machined ?
Hood
-
Nope never made mention of measuring machined parts for the +/- .0001". My beef is at the control level .. the smoothstepper as it cannot manipulate the g-code correctly, but I see everyone is looking at the machine first and quoting specs, tolerances and steps and how one cannot measure .0001" (there are several measuring tools on the market that will measure .0001", some real low budget import ones also), and if you read Jeff's post he states this issue has nothing to due with the actual smoothstepper, the opposite of what Hood states, so yeah it is somewhat confusing with the info you folks are posting in regards to wether the smoothstepper has an issue or not. Like I say I have not seen another control that connot count and display steps properly to the code that is being run. So now I am curious what other things more important to users that are needed to be worked on first that puts this low on the list?
millmaster
-
So now I am curious what other things more important to users that are needed to be worked on first that puts this low on the list?
The evolution of Mach to rev4 and all that is required to get it done while also maintaining functionality and usability along with step change improvements to Mach3. The order of effort can only be determined by those actualy doing the work and in that light i am sure the demands on those folks are great.There are more important things that need to be addressed than what has been discussed in this thread for sure.
So i will bluntly say that surely a concern will not go un-noted but will be addressed when appropriate
and not necessarily in the order we as users would like it. Such is life!
As I replied before, those who are dealing directly with the hardware and software would be the definitive source.
Thus we have tried to provide insight relative to your basic question as we relate to it.
RICH
-
Ok so what it all boils down to is the display sometimes shows this 0.0001 out. That 0.0001 seems to be the same numerical value whether in metric or imperial mode which would suggest that it is just a reporting issue to the DRO. Even if it was a real error none of us here or most in general industry for that matter, have machines capable of position to that accuracy and even if we did we would need temp controlled environments and a means to measure to that accuracy.
If you read Jeffs post again I think you will see he said it was a DRO rounding issue and not a SS hardware issue, in other words it is a software issue and the software being the SS software as it is telling Mach what to put in the DROs.
These measuring tools that can measure to 0.0001" that are cheap I dont know of, I have bore mics that have graduations to 0.001mm but their accuracy is only +- 2um, if you need to measure closer than that you will need something that has a resoltuion of 0.0001mm and then you will need to make sure all measurements are taken in a controlled environment.
So what does this all boil down to, we all seem to be agreed that it is unlikely that it is an actual positional problem and if it is it doesnt matter as we cant work to that resolution accurately anyway. You seemed to be disagreeing in your first post you said so yeah + or - .0001" steps is a concern for some of us chipmakers
but I will now assume in that you are agreeing with the above and it is only a visual annoyance rather than any physical problem which would affect our work.
So what are the other things you ask, well for many backlash comp is wanted and Greg has been working hard to get that implemented, to me it is not important but to many it is and I think it is top of the list.
THC is another, again many people waiting for this and I believe that too is near the top of the list.
For me personally the spindle control is not perfect when using a servo, it works but there are niggles. As most people dont have servo spindle I will assume this is lower on the list.
Swap axis is again wanted by some, me included but again probably of no concern to the majority of users so again presume it will be further down the list.
Greg has many things on the list but has to prioritise these things in order of what he and users think is most important and at the moment I think the backlash guys have the loudest voice and have been waiting the longest so I will presume that is top of the list, and it is probably the hardest and most complicated thing that will have to be done so I dont envy his task.
Hood
-
My beef is at the control level .. the smoothstepper as it cannot manipulate the g-code correctly
The SmoothStepper or any other motion control device that works with Mach DOES NOT interpret or manipulate the G-code, Mach ALWAYS does that. Mach's planner interprets the G-Code and breaks every movement down into the itty-bitty movements that needs to take place between each clock cycle; by clock cycle I am referring to the 'kernel' rate on the parallel port drive and the 'Controller Frequency' and 'Max Step Frequency' in the SmoothStepper configuration. In the SmoothStepper you can think of the 'Max Step Frequency' as sort of a kernel frequency that you can assign per axis. Just like with the parallel port driver, the steps/unit and max velocity determine the frequency needed.
In either case lets say you are starting from 0,0,0 and have a 'G0 X0.1234'; Mach will break that distance down into the small bits that need to be moved to each clock cycle (accounting for acceleration, deceleration and velocity, steps/unit). If your using the LPT driver then this clock is the kernel frequency, for the SmoothStepper is a combination of the 'Controller Frequency' and 'Max Step Frequency' as mentioned above. In some case breaking down a movement into all of these little steps leads to very minute rounding errors in the axis DRO. Since Mach runs a sort of cumulative axis DRO this is not an actual error as if it could not break a move down into an even number for display purposes the next move will be calculated properly, so any single step will be taken up on the next move. A crude analogy would be typing '1/3' into a calculator. You'll get 0.33333333...which we would just say is '0.3333' but if you type '2/3' you will get '0.666666666...' which we would say is '0.6667'.
If on the SmoothStepper has its 'Max Step Frequency' set to high for an axis it will limit the output resolution for that axis. You can get an idea of what this frequency should be set to by knowing your steps/unit and max velocity.
Steps/Unit * Units/Min = Steps/Min, Steps/Min / 60 = Steps/Second
This Steps/Second is the needed step frequency for that Axis, you should set the SmoothStepper 'Max Step Fequency' to the lowest value that is equal to or greater than your Steps/Second. Once you get all your axis' 'Max Step Frequency' set, adjust the 'Controller Frequency' to the lowest value that proved Smooth movement, the note of the SmoothStepper configuration page says higher settings will limit resolution.
So no, there is not a 'problem' with the SmoothStepper. Your worried about a non cumulative rounding error (i.e. display issue) that is in the ones column of your steps/unit (i.e. in the noise anyhow). It's possible that by properly adjusting the SmoothStepper it will go away, but as has been mentioned numerous times your getting all worked up over a display issue, not a positional issue.
-
Millmaster I think you are wasting your breath trying to get your point across to the virtual/armchair machinists here, I am sure they have more important things to do instead of making the smoothstepper work properly, like get the calculator out figure out the max amount of steps per unit so they can engrave their butter, collect some machine brochure data to post some useless gibble to make themselves feel like an expert, and search for a measuring device that can measure a ten thou. yep impressive! Meanwhile the high school project like smoothstepper is still a huge sponge for EMF with no built in filtering and was designed for servo operation (stepper was a afterthought) being produced in mom and dads basement, again I have sent a e-mail to Warp9 and no reply, they do NOT even have phone number on their site .... bottom line NO SUPPORT
-
cv580,
Don't test your luck, I am being tolerant.
RICH
-
Millmaster I think you are wasting your breath trying to get your point across to the virtual/armchair machinists here
Now that is funny, especially coming from a guy that wont tell us what machine he is using other than to say its a large mill with steppers and then goes on to say that when engraving with this machine the +- 0.0001" error is a problem because he is doing such small work. Ever think about things a bit deeper, like to get your resolution of steps/per unit to the 10,000 you have you are using microstepping, how accurate do you think that is?
For your information for the last 15 yrs I have made my living using machines, I am not a trained machinist but unlike you I live in the real world.
Hood
-
Funny how you can't be factual about how things actually work without hurting somebody's feelings.
-
CV580,
As i rocked in the arm chair and calculated if the additional mechanical axis reduction would provide for an
appropriate resolution, I cut the following letters in butter, so to speak. Sorry I didn't make note if the DRO
was within 0.0001" since that would be 1/20 th of the 0.002" hair shown in the picture MY TEXT IN BUTTER .
But in the 30x or so magnification it was not to be found. I originally tried cutting on my mill, but all the letters
just came out as a blob, comparatively speaking, because of how good the mill is as a system. Don't know how
i can get the periods after the letters to show as cut circles ( the period is a circular cut with a diameter of .001").
Do you think that a change in DRO will get me to the tolerance that i am looking for to those cuts?
RICH
-
Rich,
What does that coin say after in God we trust, ALL OTHERS PAY CASH. ;D
Phil
-
Rich
That is some very impressive work you are doing there. I am amazed that you can do all precision measuring with just a plain old ruler!
Imagine what you do if your smoothstepper was working properly!
cv580
-
CV580,
The ruler is just for comparison. Actual dimensions were taken with a Gaertner Toolmakers Microscope.
When i get lazy i use the fical micrometer with it.
I can also do the same engraving with the SS and a realy poopy PC at faster feedrates. ;)
In fact, SS + XBOX360 + COPYCAT = CATSMEOO......... at times.
Where's the little lettering ya talked about? I am interested in doing some at under 0.010" high.
RICH
-
I just read through this post and it remindes me of a control for a machinemI built.
The customer wanted this thing to have 2 micron accuracy.For the process and machine
I knew that this was a joke but what the heck.I bought some linear guides and servo/controls
and put it all together (2 guides and servos 25k$ usd.).Now this thing had a dro that the operator could look at to monitor the operation.I had a call a few weeks after it was all up and running.Seems that every time some one walked by the indicated value on the dro would bounce around.All kinds of crap started to fly.My solution ;
I had some one put black tape over the last 2 digits on the dro.Now the error was gone and the machine worked great.The process in reality was only measured with a .001" micrometer .
2 microns is about .00008 in.
-
Rich
You are the only guy here making noise about engraving .010" high letters. Looking forward to seeing a pic of them when you figure out how to engrave them with your smoothstepper.
Will they be 3D?
cv580
-
The SS is not the problem. The combination of spindle and cutter runout along with depth of cut makes
the individual cut lines run together such that there is no longer any space between them. Since the color
is lost ( visual spacing) you no longer have a legable letter. So the mechanical system becomes the limiting
factor and not the ability of the software and hardware.
If some fraction of a silly .0001" was added to or removed from a cut you would not see it. Which brings us
back to what we have been trying to say all along.
Now to see say 0.000100" you need to look at it under a lot of magnification. If you had a known optical
standard to compare to, ( yes i have one that goes to .000020" ) , took the time to calibrate a 1000x microscope in a controlled enviroment, along with proper lighting etc, then you can monitor the actual or non step mechanical movements to sent pulse for fractions of the silly .0001" inches. But in your case, why do all that since your resolution that you have for a step is .0001" and frankly, depending on your stepper motor the micro step won't exactly position in the electric field for that pulse. Of course you can try a different motor and would find out that they are not all created equaly and thus would find differences.
Let not the heart be troubled due to an emotional issue of display.
RICH