Machine tools are a little different from 3D printers because 1) Printers are not built nearly are solidly are milling machines. Milling machines do not flex very much 2) the printer head moves in air and there is no resistance to movement. While an end mill has to have force applied to cut metal 3) if anything causes the mill’s spindle to vibrate it is the teeth of the rotating end mill, not “spring” in the machine's structure
That said, I had planned to use accelerometers in the feet on my next robot project. Robot legs are very much like 3D printers or milling machines. Each leg has three or more axes and needs to be precisely moved and while moving, the foot might unexpectedly collide with some object. We can compute the expected acceleration on the foot and compare to the actual and find the difference. The robot operates in the uncontrolled real-world and I expect many “surprises”. The 3D printer is “bendy" and flexible so I would expect a difference between expected and actual. But the milling machine is so sturdy, I’d expect little difference. But still, it would be a very easy experiment. Decent accelerometers cost less than $20 and you could simply place one in the jacobs chuck and then jog the mill around and see what the accelerometer says. No software changes to LCNC would be needed, just put the little PCB in the chuck and look at some plots. Here is one that I have https://www.amazon.com/HiLetgo-Gyroscope-Acceleration-Accelerator-Magnetometer/dp/B01I1J0Z7Y/ref=sr_1_1_sspa HiLetgo MPU9250 GY-9250 9-Axis 9 DOF 16 Bit Gyroscope Acceleration Magnetic Sensor 9-Axis Attitude +Gyro+Accelerator+Magnetometer Sensor Module IIC/SPI amazon.com > On Dec 17, 2023, at 8:52 AM, gene heskett <ghesk...@shentel.net> wrote: > > One of the not so quiet revolutions taking place in the 3d printer landscape > over the last couple years is called input shaping, something probably done > best in the cards like Peter Wallace makes. > > A simplified explanation is fitting an accelerometer to to tool head, the > exciting each axis it turn withe an audio sweep of a small amplitude, > sweeping from 10 Hz to about 250 Hz while measuring the resulting movement > with an adxl345 recording the data it outputs for each of the axises tested. > Since a 3d printer works in slices, the z axis is generally much slower on > the 3d printer, so is left out in some versions. > > This data is then run thru a forrier or butterfly transform to develop a > compensating acceleration curve that does not effect the overall speed, but > does reduce the fine detail or boost it, to essentially cancel the machines > natural vibrations. On a 3d printer, given enough heat to supply the head > with continuous hot plastic, the use of this compensation has taken the 3d > printer from 50 or 60mm/second maximum speeds to 200 mm/second over the last > 2 years. I can now buy a printer with this built into its OS for under $1000. > There is one obstacle, most 3d printers do not have an interpreter that knows > about G2/G3 and its ilk, so most slicer's converts those to tiny straight > line moves that look like a circle in plastic. > > This is done in the default interpreter, Marlin, but is done better by > klipper in the better printers but can be reflashed into 99% of the > controller cards out there just like we can do with Peters cards. > This interpreter runs on the controller cards, often stm32 based cards that > sell for, in the ones BTT makes for $59 for a low end octopus card, which can > drive up to 8 motors, 5 fans, 3 heaters and all our limit/home switches and > probing gizmo's. Some of these even include the G2/G3 stuffs. These cards > are nearly all designed to handle nema-17 motors at 24 vols and maybe 3 amp > max motors, but one line of the octopus family of cards has a separate motor > supply input that assumes 60 volt rated drivers so even those tiny motors can > be moved at amazing speeds. > > Top that with signal stealing plugins that fit the driver socket of these > boards I'm rebuilding 2 bigger printers with nema-17 versions of the closed > loop servo/steppers with optical encoders that use drivers like we use with > linuxcnc, 2m542 sized stuff, but now rated for 90 volts and up so they can be > driven at unreal speeds w/o losing home. The PID is in the driver, linuxcnc > just tells them what to do and they do it. And if they can't do it, they will > tell linuxcnc, stopping it by linking that signal into the F2 of linuxcnc. > Doing the stop quick enough on my Sheldon the I can position a chuck jaw in > the way, and jog a carbide chipped tool into that jaw at 20mm a second, it > hits the jaw, the driver shuts off the motor drive at the same time it tells > linuxcnc to stop, the release of the motor drive lets it spring back away > from the jaw by 10 thou or so. The carbide chip is not damaged and the chuck > jaw is not marked. Tested many times, but has never occurred wile running a > job on either machine where I've put those motors. No PID's in the lathe > config, no PID's in the 6040 config. Its all in the much smarter drivers. > They use the PID error to control the motor current so if conditions are low > load, very little current or motor heating, but it it hits something, it will > use every amp the supply has to prevent a step loss, so home is maintained > under downright abusive conditions. In short, they Just Work and work well. > I'm so sold I have 3 more motors, drivers, and higher voltage power supplies > to convert the G0704's other 3 main axises to them, the A axis already is. > With the PID's in the drivers, there will only be one PID left in the G0704 > when I'm done, in the spindle circuit. > > It is the "input shaper" thing I think could improve the finish linuxcnc can > do, effectively driving the machine to cancel its vibrations as it changes > speed and directions. Doing it at ridiculous speeds. And that pays the bills > in a for profit shop. > > Something to investigate, for 3.1 maybe? > > Cheers, Gene Heskett. > -- > "There are four boxes to be used in defense of liberty: > soap, ballot, jury, and ammo. Please use in that order." > -Ed Howdershelt (Author, 1940) > If we desire respect for the law, we must first make the law respectable. > - Louis D. Brandeis > > > _______________________________________________ > Emc-users mailing list > Emc-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/emc-users _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
