Real world testing of 12v vs 24v

I would love some suggestions here. Looking to run at least these three tests.
Stipulations, hold the current the same (so the heat dissipated is the same, I have a flir to check), the rambo is limited to 24v but maybe run a 36v test anyway.

1-Holding capacity
2-Dynamic capacity

I figure for the capacity tests I just need to fix some variable weight to the endmill and rout the line over a pulley to use gravity instead of trying to rig up some sort of dynamic load meter.
Speed seems easy enough, I can just hold the acceleration constant and vary the speed.

This sound good, any other tests or gotchas I need to think about?


Clough 42 has a very cool test rig, What's the best NEMA23 Stepper/Servo? | Electronic Leadscrew Part 24 - YouTube

I really hope you run the test on both the Rambo and the SKR Pro. I’ve read some posts that indicate the 32-bit processor with the higher clock rate has a significant impact on speed, but no hard data.

That is even harder because the TMC drivers we use have a lot of settings related to speed and current. They can turn themselves up and down. I feel like I should start with the Rambo as that is probably a 1/3 less variables. That would test the power supply itself.

The SKR is gunna be special…maybe.

I was thinking a dynamometer, but that requires measuring acceleration, not really relevant for steppers.

Holding torque is easy to measure, and easy to prove that voltage is irrelevant, it’s all about current. Increase current, increase holding torque.

I have a bookmarked source for interpreting the graphs here: Speed - Torque Curves for Stepper Motors

Looks like real measurement is one of thise things that is more complicated than it should be.

1 Like

That testing rig is really neat. Time for a mostly printed stepper motor tester?

The speed and torque are coupled (in my mental model). So you would need to test with a constant speed and a variable force, or a constant force and a variable speed. You also need a way to immediately determine what the value was when the steps skipped.

That test rig gets up to a constant speed, and then applies variable torque. The force meter (scale) seems to have a max setting. So you don’t have to do anything special to catch it.

Coming from the other direction, if you had a constant weight on a line hanging off the side, attached to the bit, you could ramp up the speed until you skipped steps. But you need to know exactly what speed you were at when you skipped steps.

You could do separate tests, 5 load values and 5-10 max speed values. You would end up doing 25-50 tests. If a load of 1kg didn’t skip steps at 20mm/s, but did at 40mm/s that would give you some info. Then if the 24V did not skip at 40mm/s, then you have interesting info.

But to do every 5mm/s up to 200mm/s or every 20g difference in load (you can just use a water bottle), it would end up in a week of labor. So I am trying to think of some better alternative. But that may be over optimizing.

It would also be interesting to know if the constant weight test came up with similar results to the constant speed tests. I am not sure it would tell you anything besides the quality of your tests.

1 Like

It might be possible to mount a pen and make a sequence of marks in between the speed tests, where each speed test is slightly faster. The marks would be nicely spaced like a ruler if no steps are lost and it should be obvious if steps are lost between the 19th and 20th mark, which would indicate the speed limit for the given load. And so there is minimal penalty in terms of labor to have small increments in speed, because you just set up the job once for each load and read out the failure point.

The area where I might worry is if the acceleration would shift the load values and make the weight appear to be heavier. Maybe it’s not a big deal. I suppose as long as the acceleration is low relative to gravity then it wouldn’t be significant, or it could even be modeled and built into an adjusted weight value. For example if the acceleration is 1 m/s^2 then you know that the weight will appear to be 10% (1 / 9.8) heavier than it really is.


The 10% would be added to every speed test though. So it might be a scale on the whole thing. But you can’t avoid accelerations IRL either.

I don’t think it needs to be ultra precise. Just need ballpark numbers. Since every build will be different, actual numbers don’t matter as much as the broad result. Just want to know if higher voltage allows for 2% more speed or 20%, or any significant extra power in the range we typically use. As much as I want to actually classify the steppers I have, I rather understand this power supply debate more.

The other part is if people really want to have speed 20T pulley will shift the power curve / RPM over 20%.

So I did a little testing with 12V and 24V.

In a nutchell, nothing in the typical use case for a V1 CNC machine shows any difference between 12V and 24V. The one thing that I do where there might be a difference is that my Z rapids are faster than 12V can do at full torque.


Rough estimate, how much faster? 30mm/s to 35mm/s?

I keep wanting to run those tests but I keep sneaking away to get some jetski time in. It is very therapeutic for me.


My Z rapids go up to 50mm/s. Never even tested that at 12V lifting the gantry.

1 Like

Okay I finally started farting around with this.

Mind blown :exploding_head: for sure.

My build, fully loaded router and vac hose just as I use it. 12V

Zmax speed, 63.8mm/s
Xmax, 300mm/s
Y max, 285mm/s

See those numbers, that is nuts, fully loaded CNC. To be very clear, everyone’s CNC will not match these numbers. My firmware normally limits the XY to 50mm/s, Z to 15. Clearly I can raise that a bit.

24v…here is the catch Acceleration. RepRap Calculator - Original Prusa 3D Printers
Z max (stock 80mm/s/s) 75mm/s *accel limited. Plenty of room for more. Too much experimenting so I played with the X instead.
Xmax (stock 180mm/s/s) 330mm/s *accel limited. Okay hold your britches ----800mm/s/s 650mm/s and this number did not fail it just gets maxed out for the axis length I have.

let me repeat that one, 800mm/s/s 650mm/s

So lets discuss this a little bit. Clearly the 24v PS allows for faster rapids. Now in terms of actual real world use, rapids will help overall but 90% of the time is spent cutting, These numbers have no effect on this.

So to even take use of this we can not really speed up cutting accel, we could speed up travel accel though. I just don’t think it makes too much real world difference.

I also did check holding torque with a luggage scale, it is the same. Holding torque is current dependent, and current equals heat. This can’t be turned up until we move to metal stepper mounts. X axis was holding at more than 5kg (tmc 80% holding current), Z axis was more than 30kg holding!


Thats nuts.

My mental model says that the max force drops with speed (even though the current stays the same). Your speed test is finding the place where the force needed to accelerate the machine, plus the force needed to overcome friction is equal to the max force available that speed. Higher speed and the force drops below the force needed to move it.

Adding loads to it would add more to the needed force, which would bring down that top speed. My guess is that it would mean the difference between those max speeds would be even more different.


That was more or less my findings. I have mine set more conservatively than you do, but I just bumped up the rapids speed until I was satisfied.

I do have some projects that have a lot of small cuts in stuff, with larger spaces between, so the rapids does have some real world impact there.

I did not find any difference in the speeds that I am able to cut with 24V vs 12V. Most of my cutting is in birch plywood, some solid hardwoods like cherry, walnut and oak, and the unavoidable MDF cutting, of course. Of these, MDF is probably the slowest.


Unless you use multiflute end mills, any speed benefits from 24v are pretty much wasted while cutting. On 24v I did 100mm/s feed and 800mm/s/s accel for the cnc race with a 3 fl end mill and managed a hair over 80mm/s feed in aluminum feeling out the limits of my machine.

My feeling is that 12v is more than sufficient to get the job done. Only really need 24v+ if you really want to push the performance envelope to the absolute limit.

1 Like

I agree cutting speeds are not any better but what about Z rapids?

For X/Y rapids I wouldn’t think the higher top speed would help much since you won’t generally have enough long, fast travels, but maybe you could more easily hit the Z speed limit. If your job has lots of short segments and you are conservative on the clearance plane, the Z speed for rapids could matter, perhaps?

I’m thinking about the scenario where you’re making a cribbage board or maybe ascii art with a pen.

Maybe not a huge difference but still worth it.

Or if you are dead set on using a one-start Z screw and you use 24V, maybe you cut the Z speed in half instead of 1/4, and that could matter.

You are probably right about no need for 24V for common routing tasks, but there are a variety of applications where 24V, perhaps combined with a 32-bit board, might provide some benefit to address things like these.

  • Some LowRiders use 1-start lead screws to avoid the Z drop when the machine is unpowered. This requires a 4X lower max feedrate.

  • Before Covid, I was cutting foam cosplay props for my daughter. I was pushing max feedrates and would have gone higher if I could have.

  • On the Primo/Burly, the max Z axis feedrate is below the cutting feedrate for X and Y.

  • I have some pen projects planned that may benefit from higher feedrates.

  • Laser engraving might benefit from higher feedrates. Some of the better laser engravers report a max feedrate of 12,000 mm/min.

1 Like

I think the 15mm/s we run now is a too conservative. Mine doesn’tr cut out until after 60mm/s. I think if you found your machine’s max and dropped it 15-25mm/s you would be safe.

That could make the biggest difference. I have asked my supplier about some more 24v PS but the ones they currently have are red=negative and white=positive. That is going to cause too many issues.

Faster than the numbers I am getting with my 12V? The firmware is only limited so newcomers have a safety net on poor gcode. We can obvioulsy run much higher that we curerntly do.

Not have a hose and router on there I don’t see why you could not run much, much faster. This test showed good builds can fly around. 24v allows for more but if you see it move more than 100mm/s you might think twice about wanting more. At those speeds a mistake can break things.


Already able to run faster than that.