Re: [Emc-users] Hi All
Rainer Schmidt wrote: Hi there This is my first post. My name is Rainer Schmidt and I'm from Bloomfield, NJ. I build a 4x8 servo drivven gantry router and am currently using Mach3 as planer and to drive my servo drives. While having some mysterious problems I decided to give EMC a try. One cannot have enough alternatives. In Mach3 I am using a hardware to generate steps and direction signals to the servo drives. I read that there is a piece of Hardware doing the same thing for EMC. If someone could give me a shopping list, that would be fantastic. I have a rather high resolution on the axis and thus need higher pulse rates. it owuld make me feel better if I have a reliable source for those pulses than my PC... I ran the latency test and my Dell Optiplex GX280 is having a feisty processor but 27-30 is to slow The problem I am having with Mach3 is that I have flat's on my circles. And no clue where they can come from. I have no measurable backlash and different toolpath shows no change either. So I suspect Mach3 doing some strange stuff with the settings I applied. EMC to the rescue haha Cheers Rainer Check out pico-systems.com How are you measuring your backlash? Ed. -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Rattling steppers losing steps
Hello Jeff, as I am very reluctant with systems I don't really know, I first connected Halscope to Axis (before changing anything you suggested in the files). I can see all three stepper pulse rows and notice that, for X and Y axes, they sometimes come in tight bundles (or packets) with lots of time in between, but usually are generated apparently at very irregular distances. Halscope trigger can't get a steady hold of it. I can't find a system in this pulse generation. I' using the arcspiral.ngc program where action runs mainly in the XY plane. At beginning and end of the program the Z axis moves up and down (alone) and, astonishingly, those Z pulses come very regularly and equidistant. I am using this on a giant machine (with +/- 4m in each direction) I defined in order not to run into machine size errors with the example files. Some small steppers are connected, not being attached to a machine. The configuration of all three axes (stepconf) as far as speed, acceleration etc. is concerned is identical. I think this pulse behaviour explains the stepper rattling. What is the origin, and will changing the files as below (Base period, delete automatic reset, doublestep...) cure the case? Can vector interpolation of two axes be a problem, since a single axis move works regular? Thank you Peter Blodow At 14:20 26.01.2009, you wrote: starting from a stepconf file, you need to at least * delete all the parport.0.pin-##-out-reset lines in your hal file * delete the 'addf parport.0.reset' line in your hal file * delete the 'setp parport.0.reset-time' line in your hal file * set the steplen and stepspace parameters to datasheet numbers in ns to disable the doublestep mode and get the long step pulse lengths that are required for your driver board. Jeff -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Rattling steppers losing steps
Peter blodow wrote: Hello Jeff, as I am very reluctant with systems I don't really know, I first connected Halscope to Axis (before changing anything you suggested in the files). I can see all three stepper pulse rows and notice that, for X and Y axes, they sometimes come in tight bundles (or packets) with lots of time in between, but usually are generated apparently at very irregular distances. Halscope trigger can't get a steady hold of it. I can't find a system in this pulse generation. I' using the arcspiral.ngc program where action runs mainly in the XY plane. At beginning and end of the program the Z axis moves up and down (alone) and, astonishingly, those Z pulses come very regularly and equidistant. If you are looking at step pulses with halscope, you MUST use the base thread for sampling. If you use the much slower servo thread, halscope will miss many pulses, which could lead to the results you are seeing. If you are already using the base period, ignore this message ;-) Regard, John Kasunich -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
Rainer, There are plenty of options under EMC for directly interfacing with servos. Do I read you correctly that you are generating step signals through Mach and then use hardware to covert steps to either a pwm or analog signal? You can bypass generating step signals in EMC by using any of a number of inexpensive motion modules, such as Mesa's 5i20 or 7i43 controllers (http://www.mesanet.com/). These are FPGA controllers configurable to generate signals such as step and pwm. With the 7i33 module, which can be connected to either controller, you can convert the pwm signal to analog if that is the interface to your servo amplifiers. Or just keep what you have and configure step and direction signals from the FPGA configuration. Unlike when generating stepper signals directly, the FPGA does all the heavy lifting, so you can back off significantly on the base and servo periods. Thus the speed of your computer should not be nearly as critical. BTW, I have no association with Mesa other than having successfully implement a number of machines with their products. How are you generating your circles, using G2/G3 commands, or as a series of lines (G1) as can happen when importing from formats such as HPGL? Regards, Eric Hi there This is my first post. My name is Rainer Schmidt and I'm from Bloomfield, NJ. I build a 4x8 servo drivven gantry router and am currently using Mach3 as planer and to drive my servo drives. While having some mysterious problems I decided to give EMC a try. One cannot have enough alternatives. In Mach3 I am using a hardware to generate steps and direction signals to the servo drives. I read that there is a piece of Hardware doing the same thing for EMC. If someone could give me a shopping list, that would be fantastic. I have a rather high resolution on the axis and thus need higher pulse rates. it owuld make me feel better if I have a reliable source for those pulses than my PC... I ran the latency test and my Dell Optiplex GX280 is having a feisty processor but 27-30 is to slow The problem I am having with Mach3 is that I have flat's on my circles. And no clue where they can come from. I have no measurable backlash and different toolpath shows no change either. So I suspect Mach3 doing some strange stuff with the settings I applied. EMC to the rescue haha -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Debugging kinematics modules (was Re: 5axhydrotelkins)
Gentlemen, I think I have it figured out. I can now see both the forward and inverse variables after moving the machine to a position with mdi. I will be playing with this much more and report what I find. I will document how I did this when I get it figured out and repeatable. :) thanks Stuart -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
On Mon, 2 Feb 2009, Jim Fleig - CNC Services wrote: Date: Mon, 2 Feb 2009 14:56:40 -0500 From: Jim Fleig - CNC Services j...@cncservices.ws Reply-To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Subject: [Emc-users] PID questions In conversation with a supplier of motion control products for EMC and Mach3 the discussion of PID loops arose. The discussion was in the context of servo driven systems. Here is some information that I would welcome others review and comment. There are no flames in the following and there is not intent to start any flaming. An apples to apples comparison is being sought. A system has to be chosen for to retrofit a mill that will be cutting 3D surfaces. Curve intersecting curve intersecting flat on an angle, etc. I have retrofit similar mills with Centroid systems and the users regularly program at 100 ipm and observe that the mills average about 60 ipm (as low as 10 - 15 ipm in very curvy detail or tight corners and up to 95 ipm on almost flat curves and wide open corners). Does anyone have experience with EMC or Mach3 achieving the same level of performance as the Centroid system? If yes, what configuration was used? For emc, the PID loops are in the PC software, for Mach3, the PID loops are in the motion control hardware. EMC's PID loop has a cycle rate of 1000 times per second. Is this a fact? If yes, are there any options to get the PID loop to run faster? There is motion control hardware available for Mach3 that is capable of a PID loop with a cycle rate of 5000 times per second. 1 KHZ is just the default EMC servo loop rate, with the right hardware, 10KHz and above is possible. SNIP-- Have a good day, Jim -- Peter Wallace Mesa Electronics (\__/) (='.'=) This is Bunny. Copy and paste bunny into your ()_() signature to help him gain world domination. -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Jim Fleig - CNC Services wrote: In conversation with a supplier of motion control products for EMC and Mach3 the discussion of PID loops arose. The discussion was in the context of servo driven systems. Here is some information that I would welcome others review and comment. There are no flames in the following and there is not intent to start any flaming. An apples to apples comparison is being sought. In the same vein ... A system has to be chosen for to retrofit a mill that will be cutting 3D surfaces. Curve intersecting curve intersecting flat on an angle, etc. I have retrofit similar mills with Centroid systems and the users regularly program at 100 ipm and observe that the mills average about 60 ipm (as low as 10 - 15 ipm in very curvy detail or tight corners and up to 95 ipm on almost flat curves and wide open corners). Does anyone have experience with EMC or Mach3 achieving the same level of performance as the Centroid system? If yes, what configuration was used? It should be possible, depending on the acceleration of the motors. If it's a block processing speed issue, then it may depend on the size of the individual moves - 1 moves of 0.0001 inch each are a higher CPU load than one move of 1 inch. EMC2 has a Naive CAM detector which will aggregate multiple short moves into a single longer move, keeping within a certain tolerance. This is activated when you use G64Pxx, where xx is the tolerance you want. You can load some example of the code in the EMC2 simulator (or using a sim config) and see what happens to effective feed rates. For emc, the PID loops are in the PC software, for Mach3, the PID loops are in the motion control hardware. Yes. You have to be careful here. Motion control hardware could mean anything from a Galil DSP-based smart servo control card to a step-to-servo motor drive, such as a Gecko G320/G340. With EMC2, the position control loop is in the PC, but the velocity and/or torque loops are usually in hardware, in the motor drive. That's not strictly required, there has been some success directly driving H-bridges from EMC2, but it's not common to do this. EMC's PID loop has a cycle rate of 1000 times per second. Is this a fact? If yes, are there any options to get the PID loop to run faster? There is motion control hardware available for Mach3 that is capable of a PID loop with a cycle rate of 5000 times per second. The servo loop rate is set in the ini file. The default is 1000 times per second, but depending on your hardware (both motion control hardware and the PC CPU), you should be able to go faster. Some PCI controller cards, coupled with a fast-ish PC, should be able to get to the 10 kHz range. Although EMC receives feedback in realtime it does not adjust commands to the axes if the axes are getting closer to exceeding the following error limits. I do not know this to be true. This is a statement from the supplier of motion control hardware for EMC and Mach3. This is false. PID by definition changes the command output as the error changes (unless you explicitly set all the coefficients to 0). The PID in EMC2 is actually PIDFF - there are feedforward terms, which can pass changes in control input directly through. This improves response significantly, since no error is required to get an output. PID coefficients are applied to the position error, so normally there can be no output until some time after a command to change position has arrived. FF coefficients are applied to the command input, so they generate some output immediately. Mach3 is an open loop system sending commands that are managed by the motion control hardware which will shut down if the following error limits are exceeded and then tell the Mach3 system that it has shut down. Yes. If that's enough for you, then you can use EMC2 in the same way. There is nothing that Mach can do with hardware that EMC2 can't, except that EMC2 expects to use dumb hardware, and Mach must use smart hardware. (note: step-to-servo drives are smart in this context, because they get position commands (in the form of up/down steps) from the controller, and they manage how to track those commands) - Steve -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
Check out pico-systems.com How are you measuring your backlash? Ed. Thanks Ed. I use a regular magnetic base and a electronic dial indicator. Due to the use of the timing belts and no screws, there is no backlash. No hesitation whatsoever and all correlates with the intended pulses. R -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Peter C. Wallace wrote: On Mon, 2 Feb 2009, Stephen Wille Padnos wrote: Although EMC receives feedback in realtime it does not adjust commands to the axes if the axes are getting closer to exceeding the following error limits. I do not know this to be true. This is a statement from the supplier of motion control hardware for EMC and Mach3. This is false. PID by definition changes the command output as the error changes (unless you explicitly set all the coefficients to 0). The PID in EMC2 is actually PIDFF - there are feedforward terms, which can pass changes in control input directly through. This improves response significantly, since no error is required to get an output. PID coefficients are applied to the position error, so normally there can be no output until some time after a command to change position has arrived. FF coefficients are applied to the command input, so they generate some output immediately. Hi SWP! Hi Peter! :) I was going to answer the same but I think what is being asked is a little different. I think what was being asked is whether EMC could lower the feedrate based on The magnitude of the following error being some_limit Oh, you may be right. EMC2 can do this, but I don't know if anyone has yet. The PID component now outputs error on a pin, and also outputs a saturated signal when output has been too high for a period of time. These can be used, along with adaptive feed override, to reduce the overall feed rate. - Steve -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] EMC2 Compatible with Gecko G540
Hello Folks, I am designing my first CNC machine, I would like to use EMC2. I am looking into the Gecko Drive G540 controller. Has anyone used EMC2 with a G540? Thanks, Jeff -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] EMC2 Compatible with Gecko G540
jeff.tomerlin wrote: Hello Folks, I am designing my first CNC machine, I would like to use EMC2. I am looking into the Gecko Drive G540 controller. Has anyone used EMC2 with a G540? Sort of. I have made a G540 config, but I haven't published it yet. I need to ask Mariss a question or two, but soon it will be available from the Geckodrive site, and should be a sample config with EMC2 as well. - Steve -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive,highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
On Mon, 2 Feb 2009, Stephen Wille Padnos wrote: Date: Mon, 02 Feb 2009 15:41:46 -0500 From: Stephen Wille Padnos spad...@sover.net Reply-To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Subject: Re: [Emc-users] PID questions Peter C. Wallace wrote: SNIP- Oh, you may be right. EMC2 can do this, but I don't know if anyone has yet. The PID component now outputs error on a pin, and also outputs a saturated signal when output has been too high for a period of time. These can be used, along with adaptive feed override, to reduce the overall feed rate. - Steve Making such a system stable is not something I would like to try however... Peter Wallace Mesa Electronics (\__/) (='.'=) This is Bunny. Copy and paste bunny into your ()_() signature to help him gain world domination. -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] MDF-milling videos
For a model-yacht project, we are making positive plugs in MDF over which glass/carbon-fiber moulds will be laminated. Jari has posted two new videos, they're in my blog: http://www.anderswallin.net/2009/02/milling-mdf/ enjoy, Anders -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
Rainer Schmidt wrote: Check out pico-systems.com How are you measuring your backlash? Ed. Thanks Ed. I use a regular magnetic base and a electronic dial indicator. Due to the use of the timing belts and no screws, there is no backlash. No hesitation whatsoever and all correlates with the intended pulses. R The problem with timing belts is that they and their support systems are not as rigid as a screw. As an experiment set your indicator to the spindle and gently push and pull (a few pounds) while watching the indicator, a little deflection can make quite a difference in the roundness of a hole. What kind of tolerances are you trying to maintain? Ed. -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Peter C. Wallace wrote: EMC2 can do this, but I don't know if anyone has yet. The PID component now outputs error on a pin, and also outputs a saturated signal when output has been too high for a period of time. These can be used, along with adaptive feed override, to reduce the overall feed rate. - Steve Making such a system stable is not something I would like to try however... Indeed. I imagine you'd need pretty low frequency filters on the error, and only allow AFO changes after some saturated period. Of course, that may make the technique ineffective, since responding to impending error conditions probably needs to be pretty quick. It also won't help for spindle-synchronized moves (unless you also modify the spindle speed by the AFO factor). - Steve -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Bill Volna
Peter - yes it's the same Bill Volna - I don't know him well at all, but did meet him a couple years ago at a CNC conference. He eventually sold his lens to a guy in Canada who as of a couple years ago was almost done with this monster telescope. Unbeknownst to Bill, the bottom side of his lens had some damage and the new owner set about polishing out the imperfections. Hopefully Bill is on this list and can fill in more of the details. -- Original message from Peter blodow p.blo...@dreki.de: -- Hello Gentlemen, looking at the Photographs in search of EMC2 features at http://fenn.freeshell.org/retrofest/default.html I found a picture of Bill Volna, expert on kinematics . I remembered that name because in June 1991, I wrote a letter to him as an answer to his article in Sky and Telescope about the classroom observatory. I don't know if his obs has been realized, but mine, which I planned at the time and described to him in that letter, is standing and has been in use since 1996. 8 tons all steel and aluminum construction, 4 by 4 meters octagonal dome, 6 by 6 m to walk around it, and all at 12 m above ground. I'm now thinking of using EMC2 to make the dome rotate according to the telescope fork position (Meade 16 altaz mount). Give Bill Volna - if it really is him, William M. Volna of (then) Minneapolis Minn. Coolidge St., my best regards. Funny thing to run across a man after all those 18 years from this distance half around the world, isn't it? Greetings Peter Blodow -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Hi Peter, Hi Stephen, Thank you very much for your answers. The goal is to retrofit a knee mill (Hurco KM3P) in very good mechanical condition with the original servo amplifiers (analog signal, + / - 10 vdc) and then cut 3D gcode output from a MicroSoft (please keep the booing in the backround) based CAM software. The user must be able to connect to the MicroSoft network (Centroid is currently Linux based and I connected the customer's machine to the MicroSoft network by following Centroid's instructions) to download files (most preferable) or load the gcode onto a thumb drive and walk the program to the EMC control, load, setup part zero and tools and run. This application is very blocks per second intensive. Spending money on a faster CPU or dual CPU's would be weighed against the benefits of the ability to mill faster. The Hurco, prior to the control dying, would average 25 ipm. If after retrofitting it would average 50 - 60 ipm this would be a substantial increase. I really don't think it will do much better than that because it is a dove tail saddle. Consistently higher feedrates would require linear ways. The Centroid retrofitted mill that averages 60 ipm has a dovetail saddle. Any recommendations for a fast motherboard? Fast motion control board? Approach to configuring the system? If this goal can be achieved on a par with a Centroid system (the best I have seen so far for 3D milling on knee mills but I haven't seen everything) then I have a customer ready to give it a try. I look forward to your reply. Have a good day, Jim - Original Message - From: Stephen Wille Padnos spad...@sover.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Sent: Monday, February 02, 2009 3:59 PM Subject: Re: [Emc-users] PID questions Peter C. Wallace wrote: EMC2 can do this, but I don't know if anyone has yet. The PID component now outputs error on a pin, and also outputs a saturated signal when output has been too high for a period of time. These can be used, along with adaptive feed override, to reduce the overall feed rate. - Steve Making such a system stable is not something I would like to try however... Indeed. I imagine you'd need pretty low frequency filters on the error, and only allow AFO changes after some saturated period. Of course, that may make the technique ineffective, since responding to impending error conditions probably needs to be pretty quick. It also won't help for spindle-synchronized moves (unless you also modify the spindle speed by the AFO factor). - Steve -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
Hi Ed, FWIW Many, if not most, of the ballscrews on the machines I service are coupled to the motors by timing belts (various tooth types). I laser calibrate and do lead screw compensation adjustment on these machines and am amazed at how accurately they repeat. Different tooth types will provide varying levels of repeatability. See link: http://www.gates.com/brochure.cfm?brochure=7916location_id=11536. Standard disclaimer: I have no commercial interest in Gates. Properly tensioned, a belt can be as rigid as a screw for the system it is driving especially when the belt length is reasonably short. On the machines I calibrate I tension the timing belt slightly more than contact to the pulley and get excellent results that perform well for my customers for long periods of time. I have never found it necessary to tighten timing belts like V belts but have often found them that tight from the manufacturer or from other technicians. Just sharing some real world experience without the conclusion that it is the answer for every situation. Have a good day, Jim - Original Message - From: Ed ate...@mwt.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Sent: Monday, February 02, 2009 3:23 PM Subject: Re: [Emc-users] Hi All Rainer Schmidt wrote: Check out pico-systems.com How are you measuring your backlash? Ed. Thanks Ed. I use a regular magnetic base and a electronic dial indicator. Due to the use of the timing belts and no screws, there is no backlash. No hesitation whatsoever and all correlates with the intended pulses. R The problem with timing belts is that they and their support systems are not as rigid as a screw. As an experiment set your indicator to the spindle and gently push and pull (a few pounds) while watching the indicator, a little deflection can make quite a difference in the roundness of a hole. What kind of tolerances are you trying to maintain? Ed. -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
Jim Fleig - CNC Services wrote: Hi Ed, FWIW Many, if not most, of the ballscrews on the machines I service are coupled to the motors by timing belts (various tooth types). I laser calibrate and do lead screw compensation adjustment on these machines and am amazed at how accurately they repeat. Different tooth types will provide varying levels of repeatability. See link: http://www.gates.com/brochure.cfm?brochure=7916location_id=11536. Standard disclaimer: I have no commercial interest in Gates. Properly tensioned, a belt can be as rigid as a screw for the system it is driving especially when the belt length is reasonably short. On the machines I calibrate I tension the timing belt slightly more than contact to the pulley and get excellent results that perform well for my customers for long periods of time. I have never found it necessary to tighten timing belts like V belts but have often found them that tight from the manufacturer or from other technicians. I _think_ (it's not entirely clear) that the machine in question doesn't have screws at all - either the carriage is clamped to a belt that runs between pulleys at the extremes of travel, or the belt is clamped at both ends and runs over a pulley on the carriage. Either way, the belt is quite long, and it doesn't benefit from the mechanical advantage of a screw. This is completely different than the traditional belt and two pulleys used to couple a motor to a screw. Using a long belt to replace a screw is significantly less rigid, but can be much faster and less expensive. It is usually done on large, fast, but non-precise machines like wood routers or plasma cutters. Regards, John Kasunich -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Jim Fleig - CNC Services wrote: Hi Peter, Hi Stephen, Thank you very much for your answers. Sure (from me anyway :) ) The goal is to retrofit a knee mill (Hurco KM3P) in very good mechanical condition with the original servo amplifiers (analog signal, + / - 10 vdc) and then cut 3D gcode output from a MicroSoft (please keep the booing in the backround) based CAM software. The user must be able to connect to the MicroSoft network (Centroid is currently Linux based and I connected the customer's machine to the MicroSoft network by following Centroid's instructions) to download files (most preferable) or load the gcode onto a thumb drive and walk the program to the EMC control, load, setup part zero and tools and run. Either networking or sneakernet will work fine. There is also the ability to remotely monitor (or operate) the machine if you want it. This application is very blocks per second intensive. Spending money on a faster CPU or dual CPU's would be weighed against the benefits of the ability to mill faster. The Hurco, prior to the control dying, would average 25 ipm. If after retrofitting it would average 50 - 60 ipm this would be a substantial increase. I really don't think it will do much better than that because it is a dove tail saddle. Consistently higher feedrates would require linear ways. The Centroid retrofitted mill that averages 60 ipm has a dovetail saddle. You should do some throughput tests with any candidate software and evaluate the results as needed. You can try EMC2 by downloading the liveCD and running from it, or for a better test you can install it to a hard drive. You should modify one of the stock configs (in the sim directory, since those don't need any special hardware to be present) so it has the correct acceleration and velocity limits for the Hurco. This will give you a feel for how fast EMC2 can do block processing. If the tolerances for these parts are reasonable, you can also set G64Ptolerance to combine multiple blocks into single moves. I believe that Jon Elson did some block processing speed experiments, and came up with some few hundred blocks/second issue rate. I don't know if that was with G64P- or not. If you can do say 500 blocks/second, that's 3/minute, so wach only needs to be 0.0015 or so to get to the speed range you're looking for. This is assuming that block processing speed is the limiting factor, not servo acceleration or something else. Any recommendations for a fast motherboard? Fast motion control board? Approach to configuring the system? A couple of things: 1) You will need a PCI board for high servo cycle rates. There are several parallel-port-connected devices, but the I/O is too slow to support very fast servo cycles. At the moment, I'd recommend a Mesa card http://www.mesanet.com. 2) The realtime kernel we ship by default does not support multiple CPUs (which includes multiple CPU cores). There is someone working on making an SMP RT kernel right now, and there is an older experimental one that has worked for me on another system. 3) There has been a report that this motherboard http://www.newegg.com/Product/Product.aspx?Item=N82E16813157141 peforms quite well on the latency test provided with EMC2. You can get a dual-core CPU if kernel support becomes less of an issue. I don't know if 3 or more cores will help much, but two should. The EMC realtime layer will automatically use the highest core found if there is more than one, so there would be a split between the realtime and userspace code with two cores. I'm not sure how many user processes are involved in getting a command out of a file and down to the realtime system though, so more than two might not be of much benefit. This is a place where you may need to do some experimentation, if the performance of a relatively fast PC with one or two cores isn't fast enough for you. If this goal can be achieved on a par with a Centroid system (the best I have seen so far for 3D milling on knee mills but I haven't seen everything) then I have a customer ready to give it a try. Great. Let us know if you need help testing EMC2 for your application. - Steve -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Hi All
I _think_ (it's not entirely clear) that the machine in question doesn't have screws at all - either the carriage is clamped to a belt that runs between pulleys at the extremes of travel, or the belt is clamped at both ends and runs over a pulley on the carriage. Either way, the belt is quite long, and it doesn't benefit from the mechanical advantage of a screw. This is completely different than the traditional belt and two pulleys used to couple a motor to a screw. Using a long belt to replace a screw is significantly less rigid, but can be much faster and less expensive. It is usually done on large, fast, but non-precise machines like wood routers or plasma cutters. I saw a link from fenn on IRC the other day, it shows an interesting concept, using 2 belts. http://www.designnews.com/article/160365-A_Better_Belt_Drive.php?nid=2337rid=1696394 This probably works ok even for longer distances, as the actual belt distance is kept very short.. Regards, Alex -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Hi All
Hi there This is my first post. My name is Rainer Schmidt and I'm from Bloomfield, NJ. I build a 4x8 servo drivven gantry router and am currently using Mach3 as planer and to drive my servo drives. While having some mysterious problems I decided to give EMC a try. One cannot have enough alternatives. In Mach3 I am using a hardware to generate steps and direction signals to the servo drives. I read that there is a piece of Hardware doing the same thing for EMC. If someone could give me a shopping list, that would be fantastic. I have a rather high resolution on the axis and thus need higher pulse rates. it owuld make me feel better if I have a reliable source for those pulses than my PC... I ran the latency test and my Dell Optiplex GX280 is having a feisty processor but 27-30 is to slow The problem I am having with Mach3 is that I have flat's on my circles. And no clue where they can come from. I have no measurable backlash and different toolpath shows no change either. So I suspect Mach3 doing some strange stuff with the settings I applied. EMC to the rescue haha Cheers Rainer -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] PID questions
In conversation with a supplier of motion control products for EMC and Mach3 the discussion of PID loops arose. The discussion was in the context of servo driven systems. Here is some information that I would welcome others review and comment. There are no flames in the following and there is not intent to start any flaming. An apples to apples comparison is being sought. A system has to be chosen for to retrofit a mill that will be cutting 3D surfaces. Curve intersecting curve intersecting flat on an angle, etc. I have retrofit similar mills with Centroid systems and the users regularly program at 100 ipm and observe that the mills average about 60 ipm (as low as 10 - 15 ipm in very curvy detail or tight corners and up to 95 ipm on almost flat curves and wide open corners). Does anyone have experience with EMC or Mach3 achieving the same level of performance as the Centroid system? If yes, what configuration was used? For emc, the PID loops are in the PC software, for Mach3, the PID loops are in the motion control hardware. EMC's PID loop has a cycle rate of 1000 times per second. Is this a fact? If yes, are there any options to get the PID loop to run faster? There is motion control hardware available for Mach3 that is capable of a PID loop with a cycle rate of 5000 times per second. Although EMC receives feedback in realtime it does not adjust commands to the axes if the axes are getting closer to exceeding the following error limits. I do not know this to be true. This is a statement from the supplier of motion control hardware for EMC and Mach3. Mach3 is an open loop system sending commands that are managed by the motion control hardware which will shut down if the following error limits are exceeded and then tell the Mach3 system that it has shut down. I look forward to replies. Have a good day, Jim -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Debugging kinematics modules (was Re: 5axhydrotelkins)
Gentlemen, 1. open two shells 2. type em2-trunk-sim/scripts/emc-environment in each shell 3. in the first shell start gdb gdb rtapi_app 4. at the gdb prompt b kinematicsForward 5. answer 'y' to the question 6. at the gdb promptrun 7. start emc2 in the second shell emc2-trunk-sim/scripts/emc emc starts and the model display starts - emc is frozen at this point 8. in the first shell gdb is showing messages and a prompt - the program has stopped at the break point 9. the gdb commands work ie next, step, etc - info locals shows the variables of the kinematicsForward function 10. at the gdb prompt type continue - gdb prompt does not return 11. emc becomes unfrozen and the axis display is now on the screen 12. in the axis display - estop inverts when pressed (all commands are very slow) 13. in the axis display - machine on inverts - the display shows the machine to be started - ready to move 14. the home button inverts when pressed - homing begins and completes as expected 15. switch to mdi. input motion commands - the machine moves as expected 16. hit control-c in the first shell - gdb prompt is returned 17. I am stuck at this point - I get no more motion from axis - I can type commands into gdb (I think the gdb response is logical) I have tried several things at this point - none successful ( the commands are effective in that I am finding ways it doesn't work) :) the axis display will respond to mouse clicks and keyboard input - machine does not respond in the gdb shell enable 2 - is accepted - gdb prompt is returned run - is accepted - the response is a question asking if I want to restart the program no is accepted - no response - the gdb prompt is returned info locals is accepted - the response is 'no variable defined' - the gdb prompt is returned yes is accepted - the response is a message 'Restarting program: ...' 'Error in resetting breakpoint 2' 'Function kinematicsForward not defined' after nine repeats of the messages - the gdb prompt is not returned 18. suggestions are welcome thanks Stuart -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Jim Fleig - CNC Services j...@... writes: snip... The goal is to retrofit a knee mill (Hurco KM3P) in very good mechanical condition with the original servo amplifiers (analog signal, + / - 10 vdc) and then cut 3D gcode output from a MicroSoft (please keep the booing in the backround) based CAM software. The user must be able to connect to the MicroSoft network (Centroid is currently Linux based and I connected the customer's machine to the MicroSoft network ... Hi Jim, I recently retrofitted a Kasuga knee mill that fits your description right down to the Microsoft networking capability. I have already documented my conversion in this forum, and you can search for Successful Emc2 conversion (happy dance...) or Kasuga or kestreltom to get a better idea of the particulars. (The Hurco KM3P was one mill that I was considering for retrofit.) Proposal: If you can find an EMC post for your Microsoft CAM software, why not pastebin or post a sample cut (.ngc) file that fits your target requirements? I would be willing to massage it to run on my machine (cutting air) and see what kind of feedrates I can achieve and report the findings here. That kind of information might be useful to a lot of people. What do you think? Tom -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] EMC2 Lathe
Is your spindle on your left hand side when you're standing next to the machine? to my knowledge that's how all lathes are usually set up, with headstock on left and tailstock on right. On Sat, Jan 31, 2009 at 12:30 AM, Len Shelton l...@probotix.com wrote: I'm pretty sure you will find it the other way around- X is the cross slide, Z longitudinal Cool - that clears up a lot. Now how can I rotate the Axis display and keys to match how I stand next to the machine? Len -- This SF.net email is sponsored by: SourcForge Community SourceForge wants to tell your story. http://p.sf.net/sfu/sf-spreadtheword ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] PID questions
Hi Jim, If you remember the booth at Cabin Fever, there was a slide show playing at the corner of my booth showing a Hurco conversion done by a friend of mine. While he has not yet tweaked his system, we should be able to send him some sample files and see what sort of results he gets. His system is using the Mesa hardware along with the original Hurco amps and motors. His Hurco had the high resolution option, so he has not upgraded his encoders. Remember our discussions about how EMC handles path deviation? You can control the allowable error in three ways, first by programming the G code to set a tolerance band that allows the motion planner to simplify the path by merging blocks that are not necessary to stay within the tolerance, second by allowing blending of moves in constant velocity mode to avoid decreasing feedrate in corners, and third by setting a following error that monitors how far from the commanded path the machine is in real time. These features give you the ability to make selected tradeoffs in performance and accuracy based on your requirements. Regards, Steve Stallings -Original Message- From: Jim Fleig - CNC Services [mailto:j...@cncservices.ws] Sent: Monday, February 02, 2009 4:41 PM To: Enhanced Machine Controller (EMC) Subject: Re: [Emc-users] PID questions Hi Peter, Hi Stephen, Thank you very much for your answers. The goal is to retrofit a knee mill (Hurco KM3P) in very good mechanical condition with the original servo amplifiers (analog signal, + / - 10 vdc) and then cut 3D gcode output from a MicroSoft (please keep the booing in the backround) based CAM software. The user must be able to connect to the MicroSoft network (Centroid is currently Linux based and I connected the customer's machine to the MicroSoft network by following Centroid's instructions) to download files (most preferable) or load the gcode onto a thumb drive and walk the program to the EMC control, load, setup part zero and tools and run. This application is very blocks per second intensive. Spending money on a faster CPU or dual CPU's would be weighed against the benefits of the ability to mill faster. The Hurco, prior to the control dying, would average 25 ipm. If after retrofitting it would average 50 - 60 ipm this would be a substantial increase. I really don't think it will do much better than that because it is a dove tail saddle. Consistently higher feedrates would require linear ways. The Centroid retrofitted mill that averages 60 ipm has a dovetail saddle. Any recommendations for a fast motherboard? Fast motion control board? Approach to configuring the system? If this goal can be achieved on a par with a Centroid system (the best I have seen so far for 3D milling on knee mills but I haven't seen everything) then I have a customer ready to give it a try. I look forward to your reply. Have a good day, Jim - Original Message - From: Stephen Wille Padnos spad...@sover.net To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Sent: Monday, February 02, 2009 3:59 PM Subject: Re: [Emc-users] PID questions Peter C. Wallace wrote: EMC2 can do this, but I don't know if anyone has yet. The PID component now outputs error on a pin, and also outputs a saturated signal when output has been too high for a period of time. These can be used, along with adaptive feed override, to reduce the overall feed rate. - Steve Making such a system stable is not something I would like to try however... Indeed. I imagine you'd need pretty low frequency filters on the error, and only allow AFO changes after some saturated period. Of course, that may make the technique ineffective, since responding to impending error conditions probably needs to be pretty quick. It also won't help for spindle-synchronized moves (unless you also modify the spindle speed by the AFO factor). - Steve -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications
Re: [Emc-users] PID questions
Currently, EMC2 doesn't have feedrate compensation, ie. bringing down the feedrate override when following error rises. It could easily be That not quite true. We implemented adaptive-feedrate override some time ago. Initially it was thought to be used for electro erosion machines, but it can be used for anything. It's a float pin going into the motion controller with a value range of 0..1. If it's set at 1 emc2 will move at regular reedrate (program feedrate multiplied by feedrate override, clamped to machine limits). If it's less then 1, feedrate will be scaled accordingly. implemented OUTSIDE of most of EMC, with a small HAL component that would watch following error and use the halui interface to change the feedrate override. That is a bad idea, as halui is not realtime. By the time it notices the input and sends a command to adjust feedrate override, many moons may have passed. Using a component to watch ferror and use that to control motion.adaptive-feed is the way to go. Regards, Alex -- Create and Deploy Rich Internet Apps outside the browser with Adobe(R)AIR(TM) software. With Adobe AIR, Ajax developers can use existing skills and code to build responsive, highly engaging applications that combine the power of local resources and data with the reach of the web. Download the Adobe AIR SDK and Ajax docs to start building applications today-http://p.sf.net/sfu/adobe-com ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
