cmcdonald Posted February 17, 2011 Posted February 17, 2011 I am working on a high speed laser cutting application using linear motors. We are controlling X X' & Y gantry and we are using the Linear and circle commands. With this application we have a lot of small moves mixed in with larger moves. These small moves need a Low TA TS for the smaller moves to complete quicker. When we use a the same TA TS for the larger moves we get a lot of jerk in the system. I was wondering if anybody has some experience the dynamically changing the TA/TS to eliminate Jerk based the type of move. The goal is to optimize the cycle time with the least jerk and following error. Thanks Chris
Guest majidy Posted February 17, 2011 Posted February 17, 2011 I am not sure what exactly you want to achieve. For your longer moves you have to increase your TA/TS to avoid the high jerk. This is usually done by a fixed number calculated by user. You can also define some variables (like P variables) and calculate these P variables in a PLC and then implement them in your TA and TS like TA(P1). If you clarify more what you want to achieve, we might be able to help you better.
steve.milici Posted February 17, 2011 Posted February 17, 2011 Using lookahead might be a better solution to this issue. Your corner sizes will be defined by TA and TS but the CS will not violate Ixx16 and Ixx17.
cmcdonald Posted February 17, 2011 Author Posted February 17, 2011 Using lookahead might be a better solution to this issue. Your corner sizes will be defined by TA and TS but the CS will not violate Ixx16 and Ixx17. I am already using the look ahead feature. I'm using I5113=5 and I5120=50. And because it is important to hold the cutting path we are using DWELL0 on most of the LINEAR moves. Chris
cmcdonald Posted February 17, 2011 Author Posted February 17, 2011 I am not sure what exactly you want to achieve. For your longer moves you have to increase your TA/TS to avoid the high jerk. This is usually done by a fixed number calculated by user. You can also define some variables (like P variables) and calculate these P variables in a PLC and then implement them in your TA and TS like TA(P1). If you clarify more what you want to achieve, we might be able to help you better. We are doing this manually and it works. I was wondering if there was some formulas out there to calculate how much to increase TA/TS. Some of our customers have 1000's of different shapes and sizes so we want this done automatically. Thanks Chris
Guest majidy Posted February 17, 2011 Posted February 17, 2011 Remember that Lookahead will not change your trajectory, so you still can cut the same trajectory but limit your Velocity and Acceleration within Ixx16 and Ixx17. Your question of How much TA/TS need to be increased depends on what you want to achieve by that. If your blending in on, TA/TS will change your trajectory and has to be set differently for different shapes. The formula for 90 degree corners is TA=2*Radius/Feedrate, If that is what you are looking for.
cmcdonald Posted February 18, 2011 Author Posted February 18, 2011 Remember that Lookahead will not change your trajectory, so you still can cut the same trajectory but limit your Velocity and Acceleration within Ixx16 and Ixx17. Your question of How much TA/TS need to be increased depends on what you want to achieve by that. If your blending in on, TA/TS will change your trajectory and has to be set differently for different shapes. The formula for 90 degree corners is TA=2*Radius/Feedrate, If that is what you are looking for. Since we have tight tolerances to hold to, its best we don't change the trajectory of the path at all. With my experimentation the TA/TS value to get the fastest cycle time with lowest following error shows that A low TA/TS (20/10) and A feedrate of 500 mm/s works well with laser cutting a 1mmx1mm square but if I run a 100mmx100mm square at the same TA/TS and feedrate i get alot of 'jerk' or knocking in the corners and my following error goes up. SO I need to rincrease the TA/TS of the larger squares to get the fastest cycle time. I need a way to do this automatically or is there a feature in the controller to help this? Is there a way to factor in the length of travel in to this formula TA=2*Radius/Feedrate? It seems i would get the same results for a 1mm square or a 100mm square. I've also experimented with turning the blending on and use the following settings so that we blend certain LINEAR moves that are over a certain angle. Does this work well enough? Isx82 Coordinate System ‘x’ Blend Disable Dwell Time Isx83 Coordinate System ‘x’ Corner Blend Break Point Isx85 Coordinate System ‘x’ Corner Dwell Break Point Thanks Chris
Omron Forums Support Posted February 18, 2011 Posted February 18, 2011 I recommend using lookahead for your tangential moves (i.e., the linear moves entering the corners) but also incorporating usage of these I-Variables to limit the accelerations on corners that use Circle1/Circle2 move modes: Isx96 Coordinate System x Circle Error LimitIsx78 Coordinate System ‘x’ Maximum Circle Acceleration These variables help you to improve the tolerances on your corners greatly. There is a formula given on p. 161 (pdf page 179) of the Turbo SRM that describes how to set Isx78 based on the desired tolerances and corner speeds.
Guest majidy Posted February 18, 2011 Posted February 18, 2011 Lookahead will make sure you are not violating Ixx16 and Ixx17 and I think this is what you want. If you are cutting Sharp corners like your 100mmx100mm or 1mmx1mm, you need to disable your Blending with Isx92=1. Lookahead is still active and will make sure you are not violating your Ixx16 and Ixx17. Feel free to call me at 818-444-5218 if you have more questions.
Recommended Posts