Anyone Using NEMA 23 Stepper Motors in Robotic Applications?

[fred29]

Hey everyone,

I’m currently working on a mid-sized robotics project and considering using NEMA 23 stepper motors for the drive and arm movement systems. I’ve used NEMA 17s before for smaller CNC and 3D printer setups, but I need a bit more torque for this robot without going full servo.

The robot will weigh around 15–20 kg and have a 4-wheeled differential drive setup. I’m also planning a simple 3-DOF arm for light object manipulation (1–2 kg max payload).

I’m looking into NEMA 23s with around 2.8A and 1.2–1.9Nm of holding torque, possibly using them with digital drivers (e.g., TB6600 or DM542). Has anyone here used NEMA 23s in a similar context? How do they perform in terms of precision and speed for robotics?

A few things I’d love your input on:

Real-world performance – Are they reliable enough under load and continuous operation?

Heat and power consumption – Any overheating issues when run at higher currents?

Closed-loop options – Are closed-loop NEMA 23s (like ones with encoders) worth the investment in robotics?

Alternatives – Would you recommend switching to servos for better efficiency/control?

I’d really appreciate any experience, photos of your builds, or even wiring advice if you’ve worked with similar setups. Feel free to drop motor/driver combo recommendations too!

Thanks in advance!

 

[Huzurlu Adam]

If you want to run motors stably, use PID control systems.

 

[Minder]

One, steppers should always be ran at the rated current.
If going closed loop, you may as well go with servo's!

 

[Sunnysky]

Servo motors vs steppers of same size will limit your torque but reduce the noise. Go with larger NEMA 34 motors and higher V+ with 3A drivers for enough power. The motors may need miniature fans.

If you need torque to mill aluminum, you will need much more torque using full steps which causes excessive noise at max speed, so then large thin brass disks immersed in oil in thin disk shells will not add much mass but dampen steppers more effectively than examples found on Google. Parallel disks even better.

Then you want a g-code application that allows you to set the maximum acceleration ( to prevent step skipping ) but not limit maximum speed with no load. I used GRBL Panel (open source and very good) no updates needed. The code was run on an Arduino Uno with latest G-code. This combination expects you to use Return to Zero or Home for calibration with sensor switches. Then you can use any windows program to compile your g-code.

 

[hevans1944]

Using stepper motors in a closed-loop, negative feedback environment is different than doing a similar implementation with servo motors. I have designed and built and troubleshot systems using both, but steppers are tricky to use for closed-loop position control. A lot depends on your choice of position feedback device. Optical absolute position encoders don't play well with stepper motors. Stepper motors are ideal for open loop position control because, if your are careful about the motor drive so as to guarantee one step per pulse, you can just count pulses up and down to determine position. This works better with a direct drive (no gears, belts, etc.) because it does not consider backlash. It also is easier to design if you don't push the stepper to its maximum synchronous speed, which generally requires ramping the pulse rate up to start and ramping the pulse rate down to stop if you exceed a certain stepping rate. As for NEMA 23 versus NEMA 17, here is what Grok 4.1 has to say about that:

ME:
Someone on Electronics-Lab.com asked about NEMA 23 motors. Can you refresh my memory on what that means?
NEMA 23 is simply a standardized physical size for stepper motors (and sometimes servo or other types).

GROK 4.1:

• NEMA = National Electrical Manufacturers Association
• 23 = the faceplate is 2.3 inches × 2.3 inches square (actually about 57 mm × 57 mm, usually listed as 57 mm in metric countries)

That’s it — nothing about torque, voltage, or steps per revolution. Two NEMA 23 motors can look identical on the outside but one might be a weak 0.5 Nm hobby motor and the other a 3+ Nm industrial beast.Key things you’ll see on a real NEMA 23 spec sheet that actually matter:

• Body length (common: 40 mm, 56 mm, 76 mm, 100 mm, etc. — longer = more torque usually)
• Holding torque (e.g., 1.8 Nm, 2.5 Nm, 3.0 Nm)
• Rated current/phase (1.5 A to 4+ A typical)
• Shaft: usually ¼ inch (6.35 mm) or 8 mm diameter, single or double, with or without flat
• Wiring: 4-wire (bipolar), 6-wire (unipolar/bipolar), or 8-wire

In the maker/CNC/3D-printer world, NEMA 23 is the “big brother” to the ubiquitous NEMA 17. When someone says “I need a NEMA 23,” they usually mean they’ve outgrown NEMA 17 torque and need something that can push a heavier gantry or drive a leadscrew directly.So in short: NEMA 23 just tells you the motor will bolt straight into any mount made for a 57 mm square faceplate with four M5 holes on a 47.14 mm bolt circle. Everything else is in the detailed datasheet.

ME:

You need to be more specific in what you are trying to do. Nail down the specs first. Go look for a motor after. Maybe re-think how you plan to do the project based on what you discover. Lots of folks here that can help with that.

 

[Sunnysky]

Can you define anything for load force vs desired velocity then no load max acceleration with max mass? max position error per step?

You can use any 8mm belt pulleys to trade off speed and resolution for torque but don't under spec it and consider making vibration & noise specs. MRI scanners are horribly noisy but very precise. If designed well , all you need is an optical home sensor .