I’ve been working on this for a while, but I think it’s time for a check around this group and others to see if this is really something we need and if it would be helpful.
Along with another company, I’ve designed for a fully integrated, NEMA17 brushless DC servo that should cost $60-$70, in volume. The standard NEMA17 motor that I will use has a small PCB with a 14-bit magnetic rotary encoder, a 32 bit microcontroller, and a 17A peak current motor controller circuit. The board, so far, can talk UART and receive regular step/dir commands, but also is capable of talking I2C.
The motor I have specified is a custom made motor. The company that produces them makes the parts in china and does the winding in the US.
Performance wise, the motors are expected to quieter and faster than stepper motors and should retain a torque similar to that of many medium stepper motors even at low speeds. With a 14-bit rotary encoder the motors can detect a 0.02 degree error, which is between a 64 microstep motor and a 128 microstep motor, assuming a 200 steps per revolution motor.
Before I continue development and testing, what I’d like to know is if there is any particular interest in something like this. please vote below.
The biggest problem is 17A peak current. For 2 motors required for printer plus heated bed and hotend it will require pretty powerful psu and depend on the voltage of your motor if it is not 12v this could became pretty costly. This will make sense only if you use 12v as because of the cheap and powerful pc psu available.
I am interested but I need to see close-up photos of actual prints made using the motors. 3D printing closed loop needs a very tight control loop. I had tried ClearPaths and I wasn’t happy with the print quality, they just couldn’t match the firmness of steppers needed for common 3dp mechanisms.
If you want this to be a drop-fit replacement for a NEMA 17 stepper, I think you’re going to need to answer two technical questions that most printer users don’t even realize they need to ask about servos:
How are you doing PID tuning?
Is the motor configuration a good inertia-match for the load?
Open-loop stepper drivetrains simply don’t have these problems, so 99.8% of printer designers don’t know about them. Servos suck if you don’t have good feedback loop tuning, and you can’t tune them well if the load and motor inertias mismatch too much. So your options are to educate users and provide the necessary design tools, or build a servomotor that is configured out of the box to handle typical 3D printer drivetrains.
Deltas are also a big issue for servo systems, because the reflected inertia varies through the range of arm angles.
Another angle is to promote this product in the industrial space as a setup/index axis. It you put an Ethernet chip on board and let it talk an industrial protocol you could sell it for nearly $1,000 a pop.
Wouldn’t a closed loop add-on to steppers be a first step… self correcting any loss in steps? It exists today, I have some from a kickstarter. But adding the cost to a printer is tough… it would need to be included in higher end more expensive printers to be justified.
Correct me if I’m wrong but non-stepper motors only turn in relatively large angle increments, so your motors will need gearing, and gearing will add slop and have an impact on efficiency.
@Brook_Drumm don’t you think that adding encoders for a printer is really useless? To lose a step with proper engineering mean physical obstruction. If you have this no encoders could save the day anyway. During normal operations stepper do NOT loose step. I believe better drivers would add much more value for the printers reducing overshooting and low/medium speed resonance. @Eric_Davies So far the things with new BLDC motors have been improved a lot. Smart controllers could do something like microstepping also for BLDC servos but they are not stable in their positions they always circulating between the encoder steps.
Hi Guys, totally forgot about the algorithm. The algorithm maps the torque ripple of the motor in calibration so that the imperfections of the motor are known to the driver board. The board can then compensate for those imperfections. This technique has been verified by researchers at UPenn. Not sure about anything else though… I need to do some more research on my part
@Ulrich_Baer That motor uses a similar concept but drives a stepper motor. This motor is using a brushless motor which, if done right, should be much quieter and smoother.
Well i am electronic engineer and i like to test and make stuff. I could test them. But because i am unemployed. I have to think twice before buying. is it worth 60$. I got 5 pieces 17HS19-2004S1 steppers that price. And the biggest Question is allways the price. I could maybe pay half the price. maybe little more if they have the torque. 17HS19-2004S1 have 5.9 KG. Servos correct them self automatically but even those break in some load. How much is it 50 Kg or more.
@David_Cutting running a stepper with Field Oriented Control (FOC) is perfectly quiet and smooth. That’s where you run the motor by controlling phase advance between the coil energization and rotor position… it’s a lot like the BLDC commutation scheme but more flexible and precise. Also provides higher power density than BLDC motors and much better behavior under stall conditions. Oh, and you can use them as dynamometers, dynamic brakes, constant-torque actuators, etc etc…
yeah, closed-loop FOC steppers are frickin’ awesome.
@Ryan_Carlyle the controllers are FOC… and can do dynamic braking. Glad to see this is what you’re looking for in a motor.
@Jukka-Pekka_Ylitalo these aren’t really intended for low cost builds. But compared to industrial controllers and motors they’re pretty competitive. You’re also forgetting the step drivers which, if using non-Chinese controllers can add $10 per motor. Same thing with the motors. Not cheaply made Chinese motors, and we can customize the windings. More of a pro option.
@David_Cutting ah so if i got this right you will use a geared bl motor instead of the stepper? as a stepper is already brushless motor, and a DC BL motor just has fewer poles. I wonder if you didn’t add backlash in this concept (sure there are also solutions but will come with a pricetag)