Nice! You don't say which motor you are running... looks like the big $80 Turnigy Aerodrive SK3 - 6374-149KV is that right? And you mentioned gearing in a comment, but I can't see it in the video... are you directly connected to the pulleys?
Ah, that's the $56 motor. Great! So, with the direct drive, how can you hold against spindle side load and vibration, or is this only for e.g. 3D printing / laser cutting / plasma or other things with no side load?
So on the Y axis there is a shaft coupler to transmit the torque without loading the motor bearings sideways. I would recommend that, the bearings are not really rated for side loads. In this test design, the x axis is direct, but particular machine isn't meant for milling as you say.
Oh... no... that absolutely isn't what side load from a spindle is about. When you are cutting, the spindle is repeatedly slamming the sharp leading edge of the bit into the material being cut. This generates a "side load" which pushes back against both the axis of movement and also against the axis 90' out. E.g. you aren't just moving the spindle through space, you must PUSH it into the material and it is pushing BACK. With little or no gearing between the motor and the spindle, that back force, the side load, will shove your motors off position. Of course, the servo will sense the error and compensate, but any servo system must have some position error before it can react. As a result, there is a slight positional error in the cut. This is why most CNC systems use lead screws: The spindle can't transmit (much) force back through the lead screw to effect the motor. It can transmit a little, but far less than if it is direct connected. One of the advantages of the sort of motors you are using is extreme speed, so gearing shouldn't hurt you. And you can mount your encoders directly to the axis rather than on the motor side of the gearing so that any slop or error in the gearing system is compensated. Hope that helps.
Oh I see what you mean. Yes for a milling machine I would certainly recommend screws. That said, the control loop is running at 10kHz, so if the encoder is high resolution, the servo bandwidth could be fairly high and hopefully keep it on track even during disturbances. No-one has shown me results of this on a CNC machine yet, I'm still eagerly waiting ;D
Speed is quite impressive. Is the software ready for general consumption? Say Mach software compatibility? In other words can I buy an assembled controller and fire it off in one evening on my laptop without extra hardware, coding and tinkering around?
No worries, keep it up. You are doing a very impressive work and I am following it for some time now. Just keep us posted, especially when you will be ready to take my money. :)
how the hell does it work! how can it detect steps accurately?! but I feel like the real challenge is making a frame/etc that can handle such speeds without ghosting etc xD
Linear rails with roller ball carts eliminate any dynamic load vibrations or resonance. Using thicker extrusion like 40/40 will further increase the rigidness and table weight which will also vastly decrease instability as well. The way it detects "steps" is through an encoder that uses a hall sensor. This is how 99% of all servos motors work. The Odriver uses hall sensors and/or PWM input / Encoders with Index pulse. They are working on making them compatible via a firmware update to even run absolute encoders as well. This means resolutions of 40,000 steps plus could be seen with the right setup. Even most RC car speed controls these days utilize "sensored" 3 phase brushless motors for efficiency and smoothing increases that can take a load of amps from high voltage current Lipos that can deliver super high amp output.
Very imprssive - have you tested the precision of placements on this liteplacer machine? Since speed doesn't help, when the precision out of tolerance.
Very impressive ! Do you think it should be possible to make a less-powerful version for projects where there is not a lot of room for a board ? I already designed a board with the same DRVxx IC as what I saw on your schematics (this is where I discovered this IC) but without any microcontroller for the moment. I did'nt test it yet. If I use the same microcontroller, it guess it should not take a lot of code changes to make my board compatible with your awesome software ? I'd be glad to contribute to your driver :D
Yes, I am actually working on a smaller version right now. I will be posting on the forum about it when I have more to show. Keep an eye there ;D. discourse.odriverobotics.com/
Hey man , odrive has been designed for limited low voltages. I m wondering that to build a drive for some serious voltages and powers. I already have some experiences before from some serious ones. I would love to mixture one of my design with odrive project. Would you interested in ? Can odrive be used for such applications ?
That is fucking beautiful I am going to upgrade my CR10 3d printer with ODrive that will enable high speed printing how ever I will have to make a solid stand and bolt it to the floor to help prevent Ghosting What do you think?
Here the gearing is 40 mm/rev, and the motors used give around 1.4Nm, so that gives us about 200N of force. You can play around with different motor and gearing configuration on this spreadsheet: docs.google.com/spreadsheets/d/12vzz7XVEK6YNIOqH0jAz51F5VUpc-lJEs3mmkWP1H4Y/edit#gid=0
