Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: eabrust on December 24, 2013, 10:01:06 AM
-
Hi all, my name is Eric, and I would like to share a probing wizard I've been working on (for a long time...).
I saw Jimster's comments in Terry's 'probe wizard' thread about some of his wishes (2D probing, save to dxf, etc), and didn't want to hijack that thread to introduce myself. I've been a long time reader/lurker, and pretty busy with my day job, kids, and moving recently, but I've also been working on a probing wizard for Mach3 which does 2D perimeter probing and saves to either DXF or CSV files. It started out as a project inspired by Stirling's old routines, and to see what I could accomplish for fun (I'm not a programmer by any means...). It's coming along into something that could be useful for some, so I'd like to share it.
The wizard is called 'Probe-It!', and you can find more info about it at www.craftycnc.com (http://www.craftycnc.com). It is done completely within Mach3 VB macros. To give a brief run down:
It has a page of basic operations (corner, center find, side/edge find, etc much like other existing wizards) for setup operations, but you can record the points to DXF or CSV (sort of handy for CMM like or reverse engineering of simple parts).
There is a 'Z probing' page, which doesn't do much in its current form except it helps tell you if your work piece is 'flat' and leveled.
There is a perimeter probing page, which I think would fit Jimster's request quite well. On this page, the routine will 'trace' around the external (or internal) 2D perimeter profile, and make a DXF (or CSV file) of points/lines which can later be cleaned up and used to reconstruct your part in CAD. It isn't perfect and occasionally has issues with sharp direction changes, but it is coming along into something that works.
Finally, what I think makes this wizard unique from currently available routines is found on the setup page. You have the option to probe the ID of a 'ring gauge' of known diameter, and this builds a 'tip correction' table that is used when probing if you wish. This takes out some of the error associated with backlash, probe deflections, etc. Once you've built a correction table, errors are reduced in your measured parts, and you are also able to measure bore diameters more precisely and even get 'roundness' measurements.
I don't want to come across as being here only to 'advertise'. I would genuinely like to share what I'm working on, have people try it, and let me know what they think. I would appreciate suggestions/comments/requests/ etc.
ProbeIt requires a later version of Mach3 (3.43.66 has been my test platform for a while). I've also done all testing on a machine run via parallel port. I've had mixed response from users running SmoothSteppers, and I don't have a SS run machine to confirm/deny issues. So any feedback I could get from a SS owner would be appreciated! And like anything related to probing, I'd like to think the wizard is pretty robust, but everyone's setup is different and it can be easy to 'crunch' your probe, so please use at your own risk and probe slow until you get comfortable with the settings.
ProbeIt is free to download and play with and is fully functional, except it limits the number of points that the tip correction is on and how many points you can perimeter probe to a file without a license. If you find it useful for basic stuff, you can use it free forever.
Thanks for letting me share, and please let me know of any quetions/comments/suggestions.
Regards,
Eric (CraftyCNC)
-
HIYA ERIC Welcome to the probing fun(;-).
You do know that Mach3 HAS probe tip corrections built in right ??? Most never knew it existed.
Just a thought, (;-) TP
-
Hi TP, Thanks for the welcome!
Let me explain a little more about how my wizard is setup to calibrate the probe tip. I don't just mean being able to enter the 'ball diameter' so as to compensate measurements to the radius of the ball, if that is what you refer to about Mach3 built in correction.
The way my routines are written, so as to allow the 2D tracing of a perimeter with some accuracy, probing may be done in any possible direction in the X-Y plane, not just in +/- X and +/- Y.
If you have a really fancy probe made by Renishaw, and your machine is super tight and has no backlash, then just entering a single ball diameter is pretty good. However, most touch probes in reality deflect more in some directions than other before they trigger (ie, a +x probe event might trigger after deflecting the tip .001", but a -X probe event might deflect .003" before triggering). Also, most of us hobby guys also have machines with a little backlash. All these little errors add up when trying to measure something with some good accuracy, as I've found on my equipment.
The calibration routine in ProbeIt is set up to probe the ID of a ring gauge in 10 degree increments (36 pie slices if you will), and generates a corrected 'ball tip' diameter for every direction probed. Then when the actual probing routine is running along, the lookup table is utilized to correct the probe tip diameter to compensate for all the little inaccuracies in your machine and probe. If the angle being probed at falls in between the entries in the table, the correction is interpolated between values. The idea behind this is similar to the way a real CMM is 'calibrated' around a reference ball, I just chose to go around the ID of a ring rather than around the OD.
Regards,
Eric (CraftyCNC)
HIYA ERIC Welcome to the probing fun(;-).
You do know that Mach3 HAS probe tip corrections built in right ??? Most never knew it existed.
Just a thought, (;-) TP
-
Yep the calibration technique normally used is the ball and post method. Where you set the center of the balll to the centerline of the spindle by circling around the ball with a pin to move it into perfect alignment of the ball center to spindle center THEN probe the outside diameter of the ball to create an offset table that the machine uses to compensate when probing also the spindle MUST be locked in the same position each time so that it cannot rotate while probing or others wise the table is useless.
With a 3point supported probe there will always be 6 zones of error in relationship to accurate probing.
You can use a ring as you describe but are you centering the ring to Spindle first ??
Just a thought, (;-) TP
-
Hi again TP,
Yes, my calibration routine centers the tip to the ring before calibration, builds an initial correction table, then re-runs again to find center using the initial correction table and calibrates once more. Error is reduced every time you run through the calibration utilizing correction from a prior run.
The one thing I don't do is compensate for the tip of the probe being off center from the spindle axis. Rather than do that in software, I prefer to use touch probes that are mechanically 'centered' by adjustments. I much rather see there is no runout when measuring the tip with an indicator (I'm a mechanical guy, not a software guy :) ). For general measuring/reverse engineering, the tip doesn't have to be centered, but for setup prior to machining (edge/corner/hole center finding) it is obviously required to have the tip concentric with the spindle axis.
I would really be interested in hearing more about the Mach3 built in calibration that you mentioned existing. It must be pretty well undocumented, as I did a lot of searching about probing before starting to write my own routines, and I still can't find anything about it. ??? Can you share any more about how it works, how you use it (how to call it, variables used, is there a table created, etc)? thanks in advance.
regards,
Eric (CraftyCNC)
-
HIYA Eric , Mach3 has probe tip compensation which will account for tip radius AND switch travel. It does NOT have a comp table.
I think you missed the point about the ball and post test. The idea is to mechanically center the spindle to the ball. You install a pin in the spindle and then circle the outside of the ball (or inside of a ring) at the same diameter as the ball. This physically centers the ball under the centerline of the spindle . THen probe the ball to find any differences in values from the physical value of the ball compared to the probed value of the ball. This gives you the actual COMP value to use. It can be different for each and every tip combination you use. Short tips are different from long tips,etc,etc.
Also you must always install the probe in the same position in the spindle. Commecial units use the spindle index position.
Nice looking Wizard, (;-) TP
-
Hi TP,
I appreciate the discussion and feedback. Thanks!
I didn't completely miss your point about the ball and comp table, I just had a slightly different take on it when I started about writing my routines ;D Here's my reasoning:
My thought was this: to mechanically center the 'ball' or 'ring gauge' to the spindle before calibration of the probe, you will have to manually center it by sweeping a DTI or using a co-ax indicator, etc... there is no automating this with a macro unless you already have a calibrated probe. So to me, it made more sense to mechanically center the ball to the shank of your probe (assuming you can), and then use the comp table to clean up the probe deflection errors and any minor backlash effects, etc.
I think I could easily make it an option in my calibration routine to allow what you suggest depending on what the user wants to do:
1) If the user has already centered the calibration ring to the spindle manually (as you suggest), the routine could just touch off all around the ring and create a full comp table as you suggest. This would account for probe tip runout along with the deflection and other errors.
2) If the user has already centered the probe tip to the shank of the probe, then the routine auto finds the center of the ring prior to creating the comp table (as it already does), and repeats the routine to further reduce errors. This really only accounts for deflection and backlash errors, as the user has manually addressed tip runout in the probe already.
I also completely agree, the probe must be indexed in the same direction every time for the comp table to be worth anything. As most of us hobby guys don't have repeatable indexing spindles, that is why I would recommend the calibration technique I currently do, which is to center the probe tip to shank first. Then the comp table is much less sensitive to the probe not being perfectly indexed in the same direction.
If the user uses a 'full comp table' which corrects for significant probe tip runout, then any change in probes indexed position relative to how it was calibrated will create a significant error for edge and center finding routines.
Anybody have feedback on if it is worth while to make the option to pre-center the ring and calibrate out significant runout from the probe tip? It would put more burden on the user for installing the probe clocked the same way every time, but allow correction of probes that don't have any adjustment built in.
regards,
Eric Brust / CraftyCNC
-
HIYA ERIC if you use the method I described you WILL center the spindle over the ball exactly without the use of any other device. Then you get precise calibration standards to go by.
IF you center the probe tip first you STILL have the problems of the trip being OFF in 6 possible quaudrants while you try to center the ring. This will introduce error into your table.
This process has been around for MANY years now to calibrate a probe. Most Manf use this method.
For those without an index I always suggest the paint marker method. Align the spindle to the Quill and MARKIT with a dot to dot then install probe and align it to the spindle then MARKIT with a dot to dot to dot mark. Then always align the marks to use.
Just a thought, (;-)TP
-
IF you center the probe tip first you STILL have the problems of the trip being OFF in 6 possible quaudrants while you try to center the ring. This will introduce error into your table.
Hey TP, We still agree on pretty much everyting ;) !
You're completely right that when you initially probe to find the ring gauge center, the '6 quadrant' error is in the 'found' center location. But by running the calibration routine two (or more) times, the '6 quadrant' error gradually diminishes and goes away, because the correction table just generated is used in the subsequent finding of the gauge center for re-calibrating. If one wanted, you could keep pushing the calibrate button until the correction table didn't change anymore. I've found that two times pretty much does the job for me.
Now everyone looking for a probing tool for Mach3 has to have serious expectations, this wizard this isn't going to turn your average mill or router with a little backlash and a home-made touchprobe into a CMM good out to millionths or even tenths... I'll be the first to admit this. But it can make a pretty good improvement in accuracy for measuring things in the thousanths, and it opens up the option for corrections in any direction in the X-Y plane beyond entering a simple ball radius compensation. Long story short, it is what I needed as a basic platform for the way I wrote my perimeter tracing routine for basic reverse engineering that I wanted. So Probe-It is what it is right now pending requests and comments for changes or new features from other users or interested touch probe users out there. I'm all ears, but I can't promise anything requested can be made a reality (that is a jab at my own programming skills by the way ) :)
regards,
Eric (CraftyCNC)
-
Hi Eric, This sounds like this is just what I'm after.
Using this wizard is it possible to probe the outside of an object, and also the inside of a few pockets and save to the same DXF? For example if I wanted to probe a ring spanner and created on single DXF with the outside edge of the spanner and the shape of the internal rings at the end??
-
Hey Jim,
Using this wizard is it possible to probe the outside of an object, and also the inside of a few pockets and save to the same DXF?
Yes, you can do that. The way the wizard is setup, you open a CSV or DXF file for writing to, and you can start/stop, switch between pages in the wizard, etc, and keep recording to the same file until you close the file or close the wizard.
I went out to the shop tonight and did a quick example recording of a part of a picture frame I made a while back, but lost the CAD file in a crashed computer, as an example for you. The part is full of pockets and sweeping curves. I was going to video it also, but my camera died in the middle of it. Please see the attached picture of the part being 'traced' as well as the raw CSV and DXF files as saved by ProbeIt wizard with no edits. (NOTE, forum doesn't allow CSV uploads... ?! I resaved as a .txt, just rename back to a CSV when you look at it).
You will see the profiles in the DXF are made of many little segments, as defined by the settings in ProbeIt. There are a few breaks in the profiles, that is locations where I started/stopped due to either a faulty Estop, or so I could enter a smaller stepover to get in the tight little corners better. There are some jagged corners with circles highlighted also, this is where sharp corners occured, and the routines 'guessed' at the corners because my steps were to coarse. Ideally, you would take this raw DXF, then make a smoothed poly line out of the geometry and clean it up a little to use for a design/CAM.
Hope this helps give a little more info about it.
regards,
Eric Brust / CraftyCNC
by the way: I would like to do a better video of the usage and settings, if anyone has a recommendation on screen recording software, I would apreciate it. I've tried CamStudio, but it usually gives me poor output or crashes my computer while Mach is running.
-
I've been using Eric's plug in for a few weeks now. I must say I'm very impressed. It works very well and must have saved me hours and hours of work already
-
I've recently put up an update to ProbeIt, to version 1.1.0.
Some of the updates include:
-added secondary method of calibrating probe per TP's comments. (it is up to user to manually center the spindle in the gauge ring using whatever method they like, but it creates a full 'comp' table that also includes probe tip runout/offset in the correction table). Thanks TP for the comment, I think it is a good addition and gives the user a choice, especially if their probe is non-adjustable.
-reworked all the routines to be compatible with the CSMIO (thanks to Jimster for his help/patience/cooperation!)
-added additional 'probe active' checks in the perimeter routine
-added a counter to the perimeter probing screen so you know how many points have been collected
-various bug fixes to some of the routines.
I owe Jimster some thanks for working with me to get him up and running. It turns out that that not all of the motion control devices out there behave the same, and was giving my routines some issues. Both the para-port driver (what I have to test with) and the Smoothstepper seem to return actual X/Y/X coordinates into Vars 2000, 2001, and 2002 whether or not a touch is made at the end of the G31. I do the math and checks in my routines to decide if the probe hit or missed. After some testing with Jimster, it was confirmed the CSMIO plugins return X/Y/Z coordinates into Vars 2000, 2001, and 2002 on a hit, but it returns X=0, Y=0, and Z=0 on a miss. This took a little re-working of my code to make it compatible, but I think it is sorted out now.
I hope to keep it a work in process with more updates to come.
Regards,
Eric Brust / CraftyCNC
-
I would think that the CS Labs guys should be made aware of this and change their plugin to match the parallel port if possible.
-
I would think that the CS Labs guys should be made aware of this and change their plugin to match the parallel port if possible.
Actualy, through the process of troubleshooting and trying to figure it out, Jimster did email CSLabs (as he was their customer and also a very helpful guy!) I think based on the initial response he got back from CSLabs, they thought he was using another product and they weren't really receptive to his question/issue. I'm not sure he got any further feedback from them yet or if he clarified to CSLabs what the issue was in the end. I'll leave it for Jimster to clarify if there was any closure for him beyond the first email he got back.
Regards,
Eric Brust / CraftyCNC
-
Sadly CSLabs didn't come back with any information after the questions I initially asked them, very disappointing after the price I paid for their controller.
-
I think they will reply to you Jim, it usually takes time but I have always had a reply. a few things I assume are the reason for slow responses.
One is English is not their first language and although they seem to have a good grasp of English it still must be a factor. would you manage to reply quickly if you got an email in Polish ? ;)
Also I know they are extremely busy with enhancing their products, getting the plugin done for Mach4 and I would imagine they have a very full inbox from people wanting information and requesting enhancements.
Of course these are just my thoughts on the reasons and from dealings I have had over the last year or so.
Hood
-
Right you are hood, (again!!!!) Cslabs did get back to me last night. I should have been more patient.
-
by the way: I would like to do a better video of the usage and settings, if anyone has a recommendation on screen recording software...
Try Fraps (http://www.fraps.com/)
Screenshots are free but it appears the video recording will require $37.
Lots of folks in the gaming world use it as seems to be the case seen on therr website.
Just a suggestion as you requested.
No, I'm not affiliated with them and have no idea whether it will work or not.
-
Thanks for the suggestion on FRAPs, I'll give it a look.
Eric
-
has anyone tried this with machstdmill I like the idea of this and would like to use it with mach standard mill
Thanks andrew
-
I've put out a new revision of ProbeIt wizard, V1.1.2 is available for download.
Some of the changes:
· Records Z postions in DXF now (can do 2.5D parts with various profiles at different elevations).
· Warn of 'zero mode on' when recording , turn off DXF temporarily
· Updated Calibration Option 2 (now sets an x/y offset instead of allowing negative tip vallues for correction)
· Update setZ (change ZDRO set to center of ball)
· update touchz (change surface to bottom of ball)
· Added Z up/down buttons for manual control
· Constrain slope change on perimimeter routine added
· Added rapid FRO control
· Fix tip correct lookup error (negative angles caused issue)
· Fixed perimeter DXF write of interpolated Corners
· Altered calibration routines to discard prior data.
· General screenset cleanup and bugfixes
· Manual updated and improved to describe usage.
Some known issues:
*This wizard IS NOT COMPATIBLE with MachStdMill from Calypso Ventures, and cannnot be run while using MachStdMill Screenset due to conflicting use of DROs/LEDs/etc. A workaround to use this wizard is to setup another Mach3 Profile that uses a standard screenset (1024.set, 2010 screenset, etc).
* This wizard is known to work on machines driven from the standard parallel port, a Smoothstepper, or CSMIO. There is NO guarantee that it will work properly with any other external motion control device, as actual probing is handled within the motion controllers firmware and not Mach. Test at your own risk.
* You must have Mach3, version 3.43.19 or later. I recommend the latest lock down or 3.43.66 or later. This wizard is untested with any 'modifed' version of Mach3, such as the version provided by Tormach. Test at your own risk.
For any existing users: to update, just replace your existing 'probeit' folder in addons with the new version, but keep and transfer over your license file if you have one.
-
Eric,
I have been trying to purchace a license for your probe-it though your website and have emailed you a couple of times.
Are you still selling licenses?
Is there anyone that could sell me a lincese?
-
Are you still selling licenses for this? Has anyone tried this with an ethernet smoothstepper?
-
I hope he is still selling it as it works amazing well, I use it often, I'd be lost without it