Look, a typical stepper motor has 50 complete cycles/periods from electrical point of view. We can think of the two phases as sine and cosine functions. So, we have four peaks at which the position will be very firm and exact, due to mechanical properties of a stepper motor. You can see that is is very easy to make real sine and cosine functions of the control current and feed it to the stepper motor. But at those positions the position will not be guaranteed due to the motor construction, it will be somewhere there as we think in theory where it should be.
So, if We microstep the electrical phase into 200 steps, you get 200*50 positions per revolution (10000), very nice. What comes into question is that the maximum torque and thus position stiffness varies during the electrical phase and cyclically repeats during motor revolution. If we have such motor connected to ball screw with 10mm/revolution, we get good resolution of 1 micrometer per step, but we have far, far less accuracy than that. If you have a tool standing still against a rigid block and you move it in abovementioned way, say 6 micrometers, in reality it will not move a bit, but due to limited rigidity of the CNC machine and especially the small phase effect change will just cause increased pressure of the tool agains the target. Nice point is that you are able to change it slowly at low speeds (as this effect helps in machining to touch the tool to the material more gently), but at hight speeds, forget. The accuracy is due to the large phase skew (and rotor+machine inertia too) way off. (this is the reason why powerful servo systems contain encoder and DC motor, it is so much simpler to design! +only 1 current to control, directly, linearily proportional to the torque, yum!) Things get even worse as motor speeds up. Bear in mind that pouring 2kHz at many amperes trough a motor windings is not easy. Motor is a REACTIVE device, it not only takes current, it also puts it out! And not only that, the will of the motor to take current and other things is heavily affected by the magnetic field position (the electrical phase) and applied load torque. Plus... ANY MOTOR which takes in AC current will have a phase skew, otherwise it could be NOT rotating and the current would NOT be AC! Any motional system will have a phase skew. Therefore we buy servomotors and such which will take care of it and will not bother us with the complicated theory. Problem is that such servomotors cost much. (plus the theory: not only feedback, but also feed-forward, we not only want to be where we should have been a while ago - there you will have a problem how to meet the positional goal and not to overshoot it with ever changing load and inertia conditions, or how to get to that position just in the right time) I thought that you already knew about these properties of stepper motors, or feedback position motor control. Mario. On Dec 19, 2007 9:06 PM, Anton Buzinski <[EMAIL PROTECTED]> wrote: > > hi Paul: > > yes we do plan to provide access to jtag pins of the fpga. the code in fpga > is not open source, though. but you can install your own code. > > > Mario: > > the board will look like servo device to emc as it will feedback the > pulse-counter as the actual position. > > as far as cal. lookup table, it seems like you need encoder feedback so you > can avoid table lookup. is this a normal thing with stepper drivers that > they sag? If a table lookup is really needed, it can supported in the > driver code. > > since we are targetting this board for emc, what can be better than the emc > mailing list. > > regards, > > Anton B. > www.vitalsystem.com > > > > > > ----- Original Message ---- > From: Mario. <[EMAIL PROTECTED]> > To: EMC developers <emc-developers@lists.sourceforge.net> > > Sent: Wednesday, December 19, 2007 12:21:52 PM > Subject: Re: [Emc-developers] Stepper Controller > > It would be nice if you also implemented a callibration table, as my > microstepping unit is able of extreme amounts of microsteps, but they > are not exact, so that you could run one axis at a time at various > speeds and remember what the sagging behind the assumed position was > and compensate for that. > Ideal controller would be one which would be a servo drive, so that > the EMC would only command position and controller would catch /do > whatever necessary to be in that position in time specified etc.. > > Why don't you make a mail-list where you would send mail to people > interested as you make changes to the project. > > On Dec 19, 2007 8:02 PM, Anton Buzinski <[EMAIL PROTECTED]> wrote: > > > > > > > > hi Mario, > > > > we are hoping to have the pci board in beta testing in the next couple of > months. will know the pricing then. keep in touch as we will be looking > for potential beta testers. > > > > thank you > > Anton > > > > > > > > > > > > > > > > ----- Original Message ---- > > From: Mario. <[EMAIL PROTECTED]> > > To: EMC developers <emc-developers@lists.sourceforge.net> > > Sent: Wednesday, December 19, 2007 11:45:29 AM > > Subject: Re: [Emc-developers] Stepper Controller > > > > Currently the step resolution and encoder resolution are independent (but > stepper use with encoder is not that used much). > > When you are going to make the board, could you please post the pricing > and availability too? The other guy who made similar board for PCI (USB is > useless due to unpredictable latencies) was not able to post about pricing > or sale for months after the announcement that they have it ready :(. > > > > > > On Dec 19, 2007 7:41 PM, Anton Buzinski <[EMAIL PROTECTED]> wrote: > > > > > > > > > > > > > > hello EMC team, > > > > > > we are working on a stepper controller board in pci and usb format that > will provide high step rates (upto 5mhz or so). we are trying to figure out > how the driver might work. could you tell us what api or global variables > we can use to findout the axis command position for the current cycle and > the configured number of pulses per unit? > > > > > > if we use an encoder interface for feedback to emc (instead of just a > step pulse counter), the encoder resolution may or may not match the motor > steps per rev. how do we deal with this discrepency? will emc allow user > to enter both encoder resolution as well as motor step resolution? > > > > > > thank you > > > Anton B. > > > www.vitalsystem.com > > > > > > > > > > ------------------------------------------------------------------------- > > > SF.Net email is sponsored by: > > > Check out the new SourceForge.net Marketplace. > > > It's the best place to buy or sell services > > > for just about anything Open Source. > > > > http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace > > > _______________________________________________ > > > Emc-developers mailing list > > > Emc-developers@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/emc-developers > > > > > > > > > > > > > > > > ------------------------------------------------------------------------- > > SF.Net email is sponsored by: > > Check out the new SourceForge.net Marketplace. > > It's the best place to buy or sell services > > for just about anything Open Source. > > > http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace > > _______________________________________________ > > Emc-developers mailing list > > Emc-developers@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-developers > > > > > > ------------------------------------------------------------------------- > SF.Net email is sponsored by: > Check out the new SourceForge.net Marketplace. > It's the best place to buy or sell services > for just about anything Open Source. > http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > > > ------------------------------------------------------------------------- > SF.Net email is sponsored by: > Check out the new SourceForge.net Marketplace. > It's the best place to buy or sell services > for just about anything Open Source. > http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace > _______________________________________________ > Emc-developers mailing list > Emc-developers@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-developers > > ------------------------------------------------------------------------- SF.Net email is sponsored by: Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace _______________________________________________ Emc-developers mailing list Emc-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-developers
