Marc Bodmer wrote: > Dear EMC2 users > I had to set a pretty high Following Error for high speeds. I've > set 5mm for FERROR. (max speed is 18000 mm/min).
18000mm/min = 300 mm/sec 5mm error = 0.0167 seconds of movement Since the PID loops runs every 0.001 seconds, you will never get below 300mm/sec * 0.001 sec = 0.3mm. Realistically, you should expect 2-3 times that amount, perhaps 1mm. Do you see the following error near 5mm during the steady high speed portion of a move? Or are the errors largest during accel and decel, and much smaller during the steady portion? If the former, try some I gain, and/or FF1. If the latter, try FF2. Unfortunately, all I can do is suggest trying some things, tuning is one of those subjects that is very hard to explain. > The actual problem now is that it needs very high cutting speeds > for plasma cutting (5000 mm/min). But at this speed the corners > are not sharp anymore (about r=3mm..5mm). Sharpness of corners at high speed depends on acceleration. If you are moving along X at 6000mm/min (100mm/sec), and you want to make a 90 degree turn and go zooming off along Y at the same speed, X needs to decelerate to a stop and Y needs to accelerate to 100mm/sec. The distance the machine takes to do that determines the radius of the curve at the corner. From high school physics: d = (1/2)*A*t^2 t = V/A substitute V/A for t, and do some algebra: d = (1/2)*A*(V/A)*(V/A) d = V^2/(2*A) If you are cutting at 100mm/sec, and you have the machine configured for a maximum acceleration of 1000mm/sec^2, then V is 100 and A is 1000. Plug in and solve for d: d = (100*100/(2*1000) d = 10000/2000 d = 5 So the turning radius will be about 5mm. To find out what acceleration is needed for a tighter turn, you can solve for A: A = V^2/(2*d) So, if V is 100mm/sec, and you want the radius of the curve to be 1mm (d = 1mm), then you get: A = (100*100)/(2*1) A = 5000mm/sec^2 Note that the decel and accel times are quite short: t = V/A t = 100/5000 t = 0.020 seconds, to go from zero speed to 6 meters/minute. If your motors can't accelerate that fast, or can't be tuned to that level of performance, then you won't get 1mm turns. > => Q5. What is the influence of smaller or higher values of the > MAX_ACCELERATION value. How does one determine the needed > acceleration value on an axis? As described above, acceleration lets you make sharper corners at higher speeds. More high school physics: F = ma Force = mass times acceleration If you have a 50kG gantry and you want to accelerate it at 5m/s^2 (5000mm/sec^2), then you can calculate the needed force: F = 50kG * 5m/s^2 F = 250 newtons Can your motors and drives deliver that? If so, its just a matter of tuning them, and setting EMC's acceleration limit to 5000mm/sec^2. If not, figure out what they CAN deliver, calculate how much acceleration that will give you, subtract 15-30% for friction and other losses, and use that value for MAX_ACCELERATION, at least to start. You can use motor and drive data along with machine information like weights of parts to calculate approximate performance using the formulas I've provided. But that only gets you close, since there are many factors that can only be estimated. To find the real performance of your machine you need to do tuning and testing. Regards, John Kasunich ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys-and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users