I have been planning out my CNC conversion process, and have come to the motor selection process.
There are lots of choices between different kinds of steppers and servos. I’ve decided to go with Teknic ClearPath servos, which are considered to be drop-in stepper replacements with built-in drive modules.
A lot of what I talk about here can relate to other motor types.
Fair warning – some of this is written in “thinking things through format,” as that’s what I did.
The ball screw conversion kit for the mill I bought (Precision Matthews PM-728VT) comes with NEMA 23-size mounts for X and Y axes, and a 34-size mount for the Z axis.
NEMA 34 motors are larger than NEMA 23 motors, and can (usually) handle higher axial and radial loads.
The Proper Way to Size a Motor
The proper way to do things is to calculate the load on the axes, resistive forces, and other motion characteristics, and determine the load inertia. Then, a servo or stepper motor can be selected based on its rotor inertia.
Oriental Motor has a post about this, where they say:
For AC constant speed motors, AC speed control motors, and brushless speed control motors, you will need to look at the permissible load inertia values. For stepper motors or servo motors, you will need to know the allowable inertia ratio each type of motor can handle.
For stepper motors, the general guideline is to keep the inertia ratio (load inertia or reflected load inertia divided by rotor inertia) under 10:1, and 5:1 for faster motion profiles or smaller frame sizes than NEMA 17.
For closed-loop stepper motors, up to a 30:1 inertia ratio is recommended.
For auto-tuned servo motors, the inertia ratio increases to 50:1. For manual-tuned servo motors, it can increase to 100:1.
But, I’ve found it quite difficult to determine all that. Plus, the load is going to change depending on workholding equipment and the size and weight of the work materials. A 30 pound vise, for instance, is going to change things.
According to online forums, for the mill I will be converting to CNC control, Teknic’s CPM-SDSK-2321S-ELS is a suitable choice for X and Y axes, and CPM-SDSK-3421P-ELS for the Z axis.
The Centroid Acorn controller I plan to use can take advantage of the enhanced (E) resolution option, and sealed shafts (S) are a good idea given the environment the motors will be used in.
For the motors, CPM denotes “ClearPath motor family,” SDSK denotes the “step & direction Stepper Killer” series and model motors. The first two numbers are the motor size, the next denotes the body length, and the last is for torque-speed characteristics.
After that, S is for series-wye for higher torque and lower speed, D is for parallel-delta for lower torque and higher speed, and P is for parallel-wye for a combination of characteristics.
For the NEMA 23 motor, the 2321S servo delivers 492 oz-in peak torque, 98 oz-in continuous torque, and 3,170 RPM max. These value depend on the input voltage, more on that in a bit.
For the NEMA 34 motor, the 3421P servo delivers 547 oz-in peak torque, 164 oz-in continuous torque, and operates at up to 3,070 RPM max.
Others who have converted their PM-728VT machine have shown these to work well, and a Teknic tech has also confirmed these recommendations online.
The ball screws in my conversion hardware kit have a 5mm lead, as per its spec sheet, which is around 0.200″ per turn, or 5 rotations per inch. This means that at 3,000 RPM, servos will move the axes at a rate of ~600 IPM (inches per minute).
From Teknic’s torque and speed diagrams, the torque capabilities drop off at higher RPM. This is typical for stepper motors as well.
Here’s Teknic’s chart for their 2321S servo motor.
Here’s the chart for the 3421P servo motor.
Teknic has selection tables and comparison graphs on their website, and engineers who can answer questions over the phone.
What do the Motor Torque Charts Mean?
Generally, these motors have two torque ratings – peak and continuous (RMS). Consider how strong you are. How far can you carry a 5 lb load? 50 lbs? 100 lbs? How far can you carry such loads when you’re running compared to when standing still or walking slowly?
Heat dissipation is also going to be a factor. Motor torque specs are usually given for specific operating environments and will be derated for higher temperatures.
The difficulty in selecting a motor is in matching its speed and torque capabilities to a machine’s size and the user’s preferences and requirements.
Do I need 3000 RPM on the Z-axis? Perhaps not, but I do need torque to be able to lift and lower the heavy motor head.
The Teknic tech remarked in an online forum that these servos, for the PM-728VT and Precision Matthews’ conversion kit, can deliver 600 IPM rapids and 1 G of acceleration on X and Y axes, and 500 IPM and 1/3 G acceleration on the Z axis.
It’s not clear to me how Teknic calculates their G acceleration estimates.
The max IPM travel speeds are calculated from the RPM and number of inches of travel per ball screw rotation.
It’s worth looking at the shape of the torque curve at different speed ranges.
Straying from Internet Recommendations
What if I add cooling mounts and other accessories to the Z axis in the future, or something a big heavier such as a powered drawbar?
Torque drops off at max RPM, but the machine won’t (usually?) be performing cutting operations at higher travel speeds.
I think I want greater torque reserve, for the sake of being more future-proof.
I should add that Teknic’s SDHP (high power) ClearPath servos are similar to the SDSK models I’ve been looking at, but with more linear peak torque curves at higher costs.
In case you were wondering, here’s what the torque and speed profile looks like for SDSK (Stepper Killer) vs SDHP (High Power) motors. The HP motors have a more linear peak torque vs speed relationship.
The peak torques line up for lower RPMs, and the continuous torque specs line up most of the way, with the SDSK motor torque starting to droop sooner than for the SDHP.
The price difference for this is substantial; in this example the SDHP motor costs 45% more than the matching SDSK motor.
The benefits look to disappear at 48V. Meaning, if you’re not using Teknic’s 75V power supply, and aim to run the motors using a 48V power supply, there’s no apparent benefit to the “high power” series.
I’m sticking with the less expensive SDSK series of motors.
Here’s a comparison between Teknic’s ClearPath 3421P and 3432P motors (for the Z axis). The 3432P motor is longer, heavier, more expensive, and consumes more power. It delivers greater peak and continuous torque, but has a lower max speed.
To simplify things, I have been looking at the torque for specific speed ranges, such as 0 to 500 RPM and at around 1000 RPM.
Here’s their comparison chart for Teknic’s 2321S and 2331S motors, which I’m looking at for the X and Y axes for my machine.
The peak torque starts to dip at around 400 RPM, which for my machine would be around 80 IPM. This is important for operations such as plunge cutting. If the feed rate is increased to say 750 RPM or 150 IPM, the peak torque will have been reduced by nearly 50%. The same is true for both motors.
There’s less benefit in going with the beefier motor even between 400 and 500 RPM (~80-100 IPM).
To be frank, I don’t yet know what the ramifications of this will be, but it’s something I will try to remain aware of.
What if I Choose Wrong?
Teknic has a 90-day exchange policy for under/over-sized motors. I’m a little hesitant, as I read that some of the motor faces need to be slightly modified (drilled through) to fit the machine mounts. But, Teknic looks to build their servo motors to spec for orders of 5 or less, and so they might take be able to swap a modified mounting plate to a new motor.
In this case, motor selection can be guessed, but requires testing to verify. ClearPath motors can/should be tuned via software, and that will tell me what their torque output looks like for different loads. But, this still requires choosing a trio of motors to start off with.
ClearPath servos are not inexpensive – they’re a couple of hundred dollars EACH – and so I’d like to get things right.
This isn’t a CNC router with travel distances that can be measured in feet, and so I might choose higher torque motors that give up a little on speed.
S vs P Torque Profiles
For the NEMA 23-sized servos, I’m sticking with the ClearPath “S” torque profiles, as they provide high low-end torque and reasonably high max speeds. For the NEMA 34-sized servos, I’m sticking with “P” torque profiles, as they provide high torque with relatively high speeds.
The “S” torque profiles in the 34-sized servo motors sacrifice too much speed. Maybe? Let’s think about that.
This is how the 3421P and 3421S motors differ with respect to torque and speed output. With the S configuration, you get around double the max torque but around half the max speed compared to the P configuration.
Peak torque is about the same between 500 and 1,000 RPM, and so the question is whether I will need higher low-speed torque or a higher max RPM ceiling.
Add in the 3432P motor, which is slightly larger, heavier, and more expensive, and you can see that it delivers more torque than the 3421P model, and without sacrificing as much speed as the 3421S configuration.
The 3232S motor sacrifices too much speed for me to consider it. But besides that, look at the range from 0 to around 250 RPM. Above that, or above around 50 IPM, the 3232S motor’s torque drops down rather quickly. In comparison, the 3221P and 3232P motors maintain their peak torque ratings until around 500 RPM, which would be around 100 IPM for my machine.
Peak torque comes into play when moving a stationary axis, or when doing things such as reversing from moving in one direction to the other. The continuous torque is what to look at for most machining operations, where the cutting action involves forces that resist movement.
The speed is important for machining and rapid travel motions, but its importance for rapid travel – in my opinion – depends on the size of the machine.
What do I want greater reserve or headroom of – torque, or speed?
1000 RPM would still provide for roughly 200 IPM travel speed with a directly-driven 5mm lead ball screw.
I’m concerned about material removal rates.
At the back of my mind, I’m thinking about the balance between speed and feed rates. The spindle speed of my machine reaches around 4,000 RPM max. Smaller end mills in softer materials – such as aluminum (which I plan to mainly work with) – tend to require higher speeds. Certain tooling tends to require higher speeds as well.
Plus, toolpath have evolved towards faster movements with less engagement. “Adaptive clearing” and “high speed machining” are terms I didn’t see thrown around the DIY and hobby space when I first started looking into CNC machines and tech 10+ years ago.
I’ll need fast movement speeds, up to what the rigidity of my machine will allow, leading me to want at least some reserve speed.
This has me looking at the continuous torque line up until where it starts to drop down with speed. Basically, I’m focusing on the zone where 100% of the continuous torque rating can be achieved. Leaving a little extra on the table provides for a margin of error, and should help to ensure I’m not maxing out the servos.
Still, I want some speed in reserve – and usable speed.
Max Speeds of Commercial CNC Mills?
How fast can expertly-designed CNC machines operate?
- Tormach 770MX
- X/Y: 300 IPM
- Z: 250 IPM
- Tormach 770M
- X/Y: 135 IPM
- Z: 110 IPM
- Syril X5
- X/Y/Z: 393 IPM
- Langmuir MR-1 gantry router/mill
- X/Y: 100 IPM
- Z: 40 IPM
- Shapeoko HDM gantry/mill
- X/Y/Z: ~197 IPM
- Bantam Benchtop mill
- X/Y/Z: 250 IPM
- AVID Benchtop Pro
- X/Y/Z: 200 IPM cutting speed, 300 IPM rapid travel
- HAAS Mini Mill
- X/Y/Z: 500 IPM max cutting, 600 IPM max travel
As an aside, I find it strange that Syril describes their X5 machine as a “hobby CNC mini mill,” seeing as how it’s priced at $30,000.
If the motion motors used for these machines have power and speed curves anything like the ClearPath servos or typical steppers, I’d assume – or at least hope – that the max feed rates correspond to the point at or before the continuous torque drops off.
So, if I’m ignoring all of the guidance I saw online for CNC mill conversions, and starting from scratch, I suppose I would look at other small vertical mills and aim for max rapid feed or travel speeds of maybe 250 to 300 IPM – at most.
Even that seems extremely ambitious for anything aside from rapid travel (as opposed to cutting).
Figuring Out the Specs I Want
The mill I’ll be converting has a total weight of around 370 pounds, and so it’s nowhere as rigid as larger machines or any of the commercial models mentioned. Rapid travel speeds over 300 IPM will be useful in a CNC router. But for a benchtop mill?
Also, what’s my target IPM for cutting aluminum? I’m not sure. The max spindle speed (~4000 RPM) will be a limitation, and the toolpath can require very different speeds depending on traditional or high-speed techniques (if even possible with this spindle speed).
Everything I’ve read suggests feed rates of less than 50 IPM for aluminum.
100 IPM and under seems most realistic, which would correspond to 500 RPM for the ball screws I’ll be using.
This suggests that, roughly, I’m looking at 0 to 250 RPM for operational torque, maybe up to 500 RPM, and up to maybe 1250 to 1500 RPM for max rapid or feed speeds.
The PM-728VT is a small benchtop mill. Even so, throwing the weight of its table and spindle column around at a couple of hundred IPM is going to require a very sturdy table.
Back to the Charts
The 2321S motor is considered to be sufficient for the X and Y axes. A fixture plate and low-profile hold-downs won’t add much weight. But what about a vise. Three vises? A vacuum workholding plate?
I’ve been thinking about the work I’ll want to do on the mill, and might need a second vise to hold longer workpieces.
Stepping up to the 2331S from the 2321S is a $20 to $23 difference (depending on shaft seal and resolution options), or $40 to $46 for both X and Y axes. Power consumption would be a little higher – 193W vs 173W each.
This would provide an additional ~26% in continuous and max torque. There’s a sharp changeover at around 2400 RPM. 2000 RPM would be around 400 IPM, at which point, according to the chart, the 2331S would still provide a slight torque boost compared to the 2321S.
How likely am I to run the machine above 400 IPM?
The extra torque seems unnecessary for the mill, where I am likely to exceed the unpublished max weight limit of its cross slide table before the NEMA 23 fails to move it. But I think I want the extra reserve anyway.
Is there a downside to greater peak torque at under 500 RPM/100 IPM, and continuous torque across the speed range I’m looking at? I don’t think so.
I’m also thinking that lower utilization during regular operations will help with keeping the motors cooler for when higher torque is needed.
Certain operations might require the motor to operate between continuous and peak torque limits. Even if two motors can both deliver the same torque for a certain operation, the one with the higher peak torque ceiling rating should in theory run cooler and longer.
It seems there’s no wrong answer to motor selection, at least until I can test things via trial and error. Unfortunately, at several hundred dollars per servo, I’m not willing to buy multiple sizes to experiment.
For the Z axis, there’s the question as to how much speed I’m willing to sacrifice. From 0 to 250 RPM, there are clear advantages in going with the 3421S configuration.
The torque will peak when accelerating or reversing the heavy Z axis. Here, the motor has to work against gravity on top of friction and inertia.
Roughing work will usually be done along X and Y directions, with lighter passes taken along the Z axis even for complex contours. Let’s say the mill is shaping a ball. The roughing steps would likely create stepped diameters before smoothing over with a ball end mill or similar.
In other words, the mill isn’t going to hog out material in the Z axis, at least not usually. Narrow holes can be accomplished with drill bits maybe, and larger holes will have short plunges.
The 3421S looks to start dropping off at just over 1000 RPM. It has a significantly higher torque ceiling up to maybe 400 RPM or so, after which it quickly matches the profile of the 3421P servo.
At 500 RPM, which would be around 100 IPM for the ball screws I’ll be using, the 3421S still provides more continuous torque, but drops below the peak torque of the 3432P motor.
The 3421S looks to drop-off in torque too early.
There are a lot of Z axis movements as a machine switches paths.
If the 3421P is considered adequate for the benchtop mill I’m converting, the 3432P is the next-best option with respect to torque reserve. There’s a boost in continuous and peak torque, which will be beneficial if I add to the Z axis over time, such as with a power drawbar, accessories, or maybe even a higher RPM spindle motor.
Compared to the 3421S, the 3432P can reach higher speeds. The continuous and peak torque curves for the 3432P line up at around 1750 RPM, after which they drop off slightly. For my machine, that would mean the 3432P should – in theory – sustain its continuous rated torque up to around 350 IPM, compared to roughly 400 IPM for the X and Y axes if I go with the upgraded NEMA 23 motors there.
The 3432P delivers higher torque, and sustains its power in the speed ranges it matters. Or at least, that’s my thinking. Remember, I’m new to DIY CNC mill conversion.
The price difference here is $39 to $45 depending on seal and resolution options.
There’s no realistic way to know how much torque overhead there is until everything is purchased, installed, and tuned via the ClearPath software.
Are there Downsides to More Powerful Servos?
The way I see it, stepping up to the next-level up servo motors results in higher cost (less than $100 more for servos that cost <$1100 as-is). They’re also a little larger, and consume more peak power. The peak power consumption would come into play when the motors are pushed to their limits, in which case the upgrade would be justified.
I should add that a lot of the same considerations are made regardless of whether we’re talking about servo motors or steppers.
There could be added strain the the motor mounts, but not any more than comparable stepper motors.
I’d be giving up some speed, but it’s not full-power speed. As mentioned, I don’t anticipate wishing I could step up from say 400 IPM to 500 IPM. With the motors I’m looking at for X and Y axes, the faster motor starts to lose torque at around 2400 RPM, compared to around 2100 RPM for the more powerful step-up model.
Max acceleration is a factor too, but every estimation approach I’ve researched requires measurements and motion specs I don’t have. That this can be adjusted and experimented with via software is part of why I’ve decided to go with ClearPath servos.
Is your head spinning? Sorry, mine is too. A lot of the CNC conversion and DIY motor selection advice I’ve seen online and in videos aren’t rationalized.
I feel that motor selection would be easier for a scratch-built linear assembly, as opposed to mill that’s being converted for CNC, as factors such as weight and friction can be more easily modeled or drawn from specs.
I have seen examples of what works for others, but couldn’t shake the idea that I could benefit more from added reserve torque compared to speed. And, I’m paying attention to the lower speed end of things, where my machine will be running for most processes. For this reason, I’ve been looking at the torque vs speed curves and how flat they are before drop-off.
If the torque drops off at too low of a speed, that’s something that is likely to impact cutting performance, as well as rapid movements during step or between different toolpaths.
Do I go with the servos that maybe 2 people in an online forum mentioned having good results with, or step up to higher power that I might or might not need?
I have seen a couple of other builds with different ClearPath motor selection, and across a couple of other brands’ motor options.
That’s part of the challenge with CNC conversions and DIY projects, or the fun, depending on the person.
But, it’s not for naught. With this back-and-forth, I at least learned how to interpret servo and stepper motor spec charts.