On Tue, Mar 1, 2011 at 8:30 PM, Jon Elson <el...@pico-systems.com> wrote:
> Igor Chudov wrote: > > Think about it this way: the way I worked with a MANUAL crank is, when I > > needed to move the knee, I would crank it, watch the dial, and stop where > > appropriate. > > > > After this, I let go of the crank. I do not obsessively hold the crank, > > trying to prevent any movevent (like EMC does), and specifically I do not > > keep pushing the crank to move that last thousandth, while not quite > pushing > > enough to really make the move. I just get it to where I am satisfied, > and > > let go of the handle. > > > > What EMC does instead is, it keeps applying the PID algorithm to the > motor, > > and tries to push it "the last thousandth", without actually geting > there, > > so the motor just heats up from current passing through it. > > > > The knee ACME screw is self locking. > > > > How can I make EMC do the same, just "let go" after it gets to the right > > place? > > > > > You really can't do this with an axis in EMC, as far as I know. If you > had a linear encoder on the knee, you could read that into EMC2 and use > it for whatever purpose you want. You could rig a switch to the servo > drive and set the following error tolerance so that EMC would accept and > display it as an axis, but not actually be able to control it. > But, if you can jog it or command it with G-code, then it needs to be > controlled all the time, unless you do something fairly different from > normal EMC use. > > I think the real problem is that the motor is severely underpowered for > the load, and is being forced into the peak torque range under > standstill conditions, which is an abusive situation, as you have > discovered. Either a stronger motor or a greater reduction ratio is > needed so the motor is safely within the continuous torque rating when > standing still. This may take a really big motor for such a heavy knee. > > Jon, thanks for a thoughtful reply. Your answer is definitely appreciated. I do think that the motor is underpowered for the axis, if I had to move the axis continuously. However, if I only need to use it occasionally to adjust for tool height changes, it is OK. What is the issue that you are pointing out, is that when the axis is "at rest", it is not really at rest from EMC2's standpoint, and its pid just keeps applying the same logic as during the moves. So, if I can find some way around that, I will be golden. I had a chance last night and spent about 1-1.5 hours working just on this issue. It took a long time and was educational. I realized that I had to increase my P and I parameters very substantially, and reduce max speed and acceleration, in order to deal with a very heavy axis. P = 260 I = 200.0 D = 5.0 After applying these adjustments, the moves are visibly much smoother, but still quite fast compared to manual cranking of the handle. My following error is only 0.001" outside of the deadband. With zero deadband, usually f-error is under 0.001 when the knee is under way. I did set the deadband to zero. The better news is that based on limited post-change testing that I did (I was very tired by then at midnight), these changed in P and I, do work better as far as bringing the ppmc.0.DAC.03.value down after a while. The motor does succeed at getting f-error very low (think 0.00007), and the output of the motor controller becomes several times lower than in the past. It settles down kind of slowly, over several seconds. I know that my drive is in torque mode (controls current), and the current is about 10 times ppmc.0.DAC.03.value, per ammeter. So I can watch ppmc.0.DAC.03.value and I will know what the current is. I am pretty sure that the motor will not overheat from, say, 1 amp. I will keep watching and experimenting, as I am not yet convinced that I am fully done with this. I want to be sure that what I did, worked, or else I will burn down the motor one day. I appreciate everyone's help and if you have something additional to say, I am all ears. i ------------------------------------------------------------------------------ Free Software Download: Index, Search & Analyze Logs and other IT data in Real-Time with Splunk. Collect, index and harness all the fast moving IT data generated by your applications, servers and devices whether physical, virtual or in the cloud. Deliver compliance at lower cost and gain new business insights. http://p.sf.net/sfu/splunk-dev2dev _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
