PM833 Spindle Upgrade - AC Servo vs Inverter Duty Motor

I recently converted a PM833-TV to CNC. I am now working on a spindle motor upgrade. My motivation for the upgrade is twofold. 1. The original motor runs almost 7500 rpm to achieve a spindle speed of 3200 rpm. This means it is very loud. 2. I machine mostly aluminum, so I would like to increase the speed to 5000-6000 rpm. I also wanted to use a one speed pulley system.

Full disclaimer: I have degrees in both mechanical and electrical engineering, however I never professionally practiced either. I am retired and love to learn and tinker with technical stuff. I welcome all feedback.

I watched several videos with folks using 2hp, 3hp and even 5hp motor on similar mills. My first impression was that seems like a lot of weight to put on these benchtop mills. More weight means potentially larger ball screws, larger servos/steppers, and more wear & tear on the gibs. Furthermore, more weight means more inertia and more force/wear on the ball screws and ball nuts every time the z axis accelerates or decelerates.

I also viewed a few videos and read posts with AC servos being used for spindle motors. They claimed the servos provided enough torque to approach the limits of the rigidity of the mills.

Here is what I have learned. Both AC servos and Inverter duty motors share similar constant torque curves. This means they provide their full torque from about zero rpm up to their rated rpm. Above the rated rpm, torque falls off quickly. Torque is defined as twisting force, so these motors provide their full twisting force up to the rated rpm. This is ideal for a milling machine because you need max twisting force at both high and low rpm.

I began looking at 1.8KW and 2.6KW low-cost Chinese AC servo motors. You can find these all over the internet from the mid four hundred dollar up to very low six-hundred-dollar range. The 1.8 KW (about 2.4 hp) options put out 6 Nm of torque and weight just under 15 lbs!
The 2.6KW (about 3.5 hp) option puts out 10 Nm of torque and weighs about 27 lbs! It can produce as much as 25Nm for up to 3 seconds.

By contrast, a 1.5KW, 2HP Black Max Inverter Duty motor is rated at 8.13 Nm of torque. The locked rotor torque is 29.5Nm. I am not sure how long it can be overloaded. This motor weighs a whopping 68 lbs. This motor along with a VFD and resistor cost about $1500 + a bunch of freight.

The OEM motor on this mill weighs about 35 lbs. If you are like me and add a power drawbar, the weight of the power drawbar + 2.6KW AC servo weigh about the same as the original motor. If you use the 1.8KW servo, it will weigh less. Personally, I think the weight really matters.

One challenge with servos is their max rpm. The 1.8KW option has a max rpm of 3000-4000. The 2.6KW option is 3000-3500. For timing pulleys and belts I am working with For literally a few dollars they will bore out stock pulleys and cut custom keyways. I ordered a 72 tooth 5M HTD pulley for the motor and 40 tooth pulley for the spindle. These were the largest pulleys they sell to fit in the mill and smallest pulley possible with a 40mm bore for the spindle. Belt, pulleys, custom boring and keyways cost $175.00 delivered from the UK. These guys are fantastic. Ted in tech support calculated that this pulley system can deliver 7.5KW.

This will give me a 1.8:1 pulley ratio. This translates to a max spindle speed of 5400 – 6300 rpm. The spindle torque will be about 5.5 Nm from 0 to 5400 rpm. The Black Max motor would have a pulley ratio of 1:1, so it will produce 8.3Nm of torque at it’s synchronous speed of 1800 rpm but will have significantly less at speeds over 1800 rpm.

To put these numbers in perspective, an engineer (Uwe on the Centroid Forum aka Suntravel) with 40+ years of experience working on milling machines in Germany, converted a small mill using a 1.5KW JMC AC servo with a 1.68:1 pulley ratio. He stated “1.5 kw Servo will be ok to get the mill to its limit of stiffness. I have mine with 1:1.68 for higher spindle speed, buts still ok to tap M16x1.5 with 800rpm in tool steel.

I think either of these AC servos are great options for spindle upgrades. I chose the 2.6KW JMC servo from the available Chinese options because the JMC manual was excellent, the frame and shaft are larger than comparable Chinese offerings, and Uwe provided me with a schematic and the programming necessary to interface with my Centroid Acorn controller.

I am a happy to provide any drawings and any information for anyone interested. I purchased the 2.6 KW AC servo motor from TOPCNC (338) on ebay. It was not available anywhere I looked, I messaged them, and they located it for me. They were great to work with. Servo motor and servo controller cost $633 delivered.

I am currently waiting for the pulleys to arrive. Below are images of Dr. D’s modified motor mount. Being a bit nuts for a quiet machine, I took of the cover to the servo controller, sketched it in Fusion 360 and modified it so I could mount a larger 80mm quiet Noctua fan on the exterior and removed the internal fan. Yes… I am a bit nuts… but I like to listen to music while setting up parts on my mill and hate the sound of whirring fans.

When comparing AC servo vs. Inverter Duty motors, I view the AC servo as a sports car engine with a tremendous power to weight ratio. It is light, high powered and will provide tons of torque. It is pefect for intermittent loads that change all the time. Just like a car.

The inverter duty motor on the other hand is like a diesel engine. It is designed to output up to 115% of its rated power all day long and will most likely last for years doing so. It is a continuous duty motor. It is similar to a boat engine with constant load. If I were boring large holes in hardened steel all day long…every day…I would choose the inverter duty motor. I would not however choose the inverter duty motor for high speed milling as this motor at 5600 rpm cannot provide much torque.

If I were using the mill for a small shop with intermittent loads, I would choose the AC servo. It will most likely last for many years.

Servo Driver With Modified Cover:

Servo Motor and Motor Mount:

if you are going to work with aluminum all the time the faster the spindle the better off you will be. if you are planning on doing high speed machining and 3D tool paths weight will factor in quite a bit into your acceleration which is a must for these kinds of machining strategies if you want to do them quickly. the biggest set back is going to be the gibed ways of the machine for this type of machining style. proper linear rails would be a huge improvement over a gibed way.
high speed machining is more about high spindle speed and fast feed rates. the cutting forces involved are not very high at all so torque and power are not as important for this kind of work.
i am working on a desk top machine with these goals in mind. this machine will be tested with a 300 watt spindle. but it is small light and runs at 10,000 rpm. the new frame i am building is like a tank compared to the junk the original machine came with.

Very interesting. The machine you are working on sounds like quite a nice project. Please post your progress. I would love to see it! I fully agree with your thoughts on speed, acceleration and weight.

I am new to milling and also plan on working with some steels that require low speed and high torque in the future. For this benchtop mill, the gibed ways are certainly a limiting factor, but I think with this 2.6 KW spindle, lots of torque and a 6000 or so max rpm, it will prove to be a good mill for machining a wide range of materials.

Thanks again… Richard

6000 rpm is not a bad range to be in and probably is fast enough for the gibed ways since i think that will be your weak point.

as for machining i started doing it when i was 16 years old and will be turning 51 this year so i have some practice :slight_smile:

Definitely following this post. Thank you for your analysis on induction vs servo motors for spindle drive. It is getting harder (if not impossible) to recommend inverter duty motors due to their price increase.

When I build/convert my next CNC mill, I will likely opt for a ClearPath MC

Thanks Dr. D. I have been watching these ClearPath MC servos as well. James Clough on the YouTube channel Clough42 did a nice review on these as well.

Just an FYI… I spoke with Technic today and found out that the Clear Path MC servos would not be a good fit for a spindle motor because they generate max torque at about 2000 rpm. A little disappointing! Here is the JMC servo selection guide for those who may be interested. This is the brand I am using.

Best… Richard

Boo… but much appreciate the heads up!

I got the customs bored pulleys and mounted the whole assembly today, but ran into a few hiccups:

  1. The bottom part of the motor mount was milled wrong. I messed up the location of the slots. I fixed the drawing and 3D printed the parts out of PLA. My plan is to mount the motor using the PLA printed parts, and mill the correct parts out of aluminum taking gentle cuts. Here is the motor and new pulleys sitting on the PLA legs:

  1. The slot in the spindle pulley was ordered to accept a 6mm keyway, but the key sits a few thousandths too high, and the rounded ends of the keyway do not fit exactly. I did not want to modify the original key so I ordered a set of 6m x 6m x 20mm keyways on Amazon for $9.95. Twenty minutes of wet sanding with 180, 320 & 400 paper solved this problem. The sanded key dropped right in.

  2. I questioned if the two M8 set screws would be adequate to hold 4.25" diameter HTD timing pulley on the AC servo. I learned from a few seasoned machine builders on the Centroid Forum that it is standard practice to use Loctite 620 or 648 (Retaining compounds) to secure timing pulleys on servos. It was explained that the forces due to acceleration/deceleration can cause the pulley to walk (axially) off the shaft. I decided to use Loctite 620. To ensure proper pulley alignment, I first assembled the pulley on the shaft with no Loctite, adjusted its position to track with the spindle pulley, then removed the motor/pulley from the mill and 3D printed a gauge block of sorts to allow me to assemble the pulley on the shaft with the Loctite and be certain that it was in the correct place. I guess you can say I was a bit nervous of glueing the pulley in the wrong postion. This stuff only comes off with heat at about 400F.

Today I milled the two bottom pieces of the motor mount that I previously messed up. These were milled with the new spindle motor & PLA printed motor brackets. I was a bit concerned running the motor on PLA brackets but the speeds and feeds calculator indicated I was only drawing 1/8 HP. The motor ran amazing! It is extremely quiet. They came out great. My first impression is that this AC serov motor is a perfect fit for this machine. Now I need to take this thing apart one more time to install the new pieces.


I have one last hurdle to finalize my spindle upgrade, proper spindle temperature. I purchased the mill with no previous milling experience. Right out of the box I thought the spindle ran a bit warm. After running at 3000 rpm for a short while, I could barely touch the TTS tools. The folks at Precision Mathews indicated that this spindle can run a bit warm and that was normal. They offered no info regarding an actual safe temperature.

There were two posting on this forum (Dr. D and Todd Cande) who replaced the oil seal with one with a slightly larger diameter. I ordered this seal, but by the time I got it, I had more hours on the mill, and the spindle was running warm, but not as hot as originally. Todd also mentioned that he could not be sure that the seal was the solution. I also learned that it was difficult to remove this seal, so I have not replaced it yet.

With the new motor installed, and the ability to reach 6000 rpm, I need to revisit spindle temperature. I briefly brought the spindle up to 5500 rpm and all looked good, but after running for just a few minutes at 3500, the spindle was too hot to touch for more than a second or two.

I have read some info regarding breaking in the spindle bearings starting at 10% of the max speed, running for 30 minutes, increase 10%, run for 30 minutes, and keep going to max speed. If the temp exceeds 140F at any point, stop, cool to room temp, go back to the previous temp range and start again.

I have no idea as to how much this will help. I welcome any and all thoughts regarding figuring out what is a safe spindle temp for tapered roller bearings and how the oil seal plays into this.

Best… Richard

one of the first things that i would watch is the PLA printed parts melting if things running hot is an issue. if you have to print temporary parts, a media with a higher melting point would be my first thought. PLA has a low melting point.

breaking in the bearings might help with the heat but i don’t own a P.M. so i can’t help on that topic. spindles in general have some heat generated and safe operating temp. should be able to be found i would think? 140F is getting close to the melting point of PLA so i would bear that in mind.

Thanks Jay. Not sure why I said PLA… used PETG. Just replaced the PETG with the correct aluminum parts this morning.

Feeling a bit more confident now!

I also got the following advice from a seasoned machinist on the Centroid Forum:

Depends on the bearing grease, I use Isoflex NBU 15, thats good from -40 to 130C.
My spindle is reaching 45-50C fast and stays there even if I run jobs with 6000 rpm 16h a day.
If your spindle temp keeps rising and will nor equal out at some point, I would stop at 80C, dismantle, Clean and fill with NBU 15 max 30% of the space in the bearings.
Too much grease and too much preload are main reasons for a too high temp.

Today a started running for 30 minutes starting at 1500 RPM, increasing 500 rpm up to 3000 rpm. By the time I hit 3000 rpm, the temperature measured on the surface of the quill was 140+F.

I stopped the experiment at this point for now and hoping to get more info before proceeding.

Thanks… Richard

That particular grease is of good quality but it is a bit expensive. it might help but the preload of the bearings might be your problem. if memory serves me those are tapered roller bearings and they could need adjusted on the preload. that type of bearing is pretty strong as far as loads go so i would think a light preload to take any slop out would be fine. over tighten and you get added friction and unwanted heat as a result. So i would lean towards the bearing preload as being the heat problem.

Thanks Jay. When I first bough the mill, I released some preload. It did not do anything, but maybe I need go a bit further. On the Dr. D video, he suspected that the oil seal was the culprit. I actually have the larger size oil seal. As a start, I may replace the oil seal and back off on the preload.

Thanks again… Richard

Hey RJS,

I threw the kitchen sink at the spindle heat problem.

  • Replaced the oil seal
  • Re-greased the bearings with Isoflex
  • Used a dial indicator to set the preload at minimum amount to prevent slop

My spindle is cool to the touch on short cutting operations <30 minutes, but will continuously heat up. When it runs for several hours the spindle will be to touch. But when I have coolant on that usually helps.

I will probably upgrade to a BT30 spindle before taking the R8 spindle apart.

Hey Dr. D,

Are you able to run at 5000 or 6000 rpm? If so how hot does it get?

Thanks… Richard

These observations are from 4k RPM. I would assume higher RPM would proportionally speed up the heating.

I am doing some investigation now. I managed to remove the OEM seal (45 x 72 x 10). It had a grove for a spring, but there was no spring. I was immediately apparent that it was rubbing on the spindle. You can see where it was rubbing. Even after cleaning it up, you can see where it burnished the spindle. The bearing seems to have quite a bit of grease in it. Lastly, I do not think this seal is rubbing on the face of the inner race. There is a few mm of space above the top wear mark made by the seal, .

I am running now with no seal and testing for temperature. I think the bearing has too much grease.

I will keep you posted.


Here are today’s findings. With no oil seal installed and the original OEM grease. Here is what I recorded for temps (see below). Since I don’t know what grease is being used or the brand of bearings, I stopped at 4500 rpm.

I inspected the bearing carefully during this temp test. The grease always stayed thick and nothing ran out at all. When I removed the oil seal, I did see a small amount of grease that that had melted and was sitting in the oil seal. This would have happened while I was running the mill previously.

I reached out to SKF, Timken and NTN. They all said these bearings run hot due to their internal resistance. Timken and NTN have a max temp off 250F with SKF at 300F. The folks at Timken said most of the time, they run at 160F to 180F. SKF is the winner with a top operating speed of 8000 rpm. NTN and Timken were rated at 6300 and 4000 with grease. For some reason, SKF claimed the 8000 is good with grease or oil. Quite a big difference from brand to brand. So, I guess I would really have to know the manufacture to figure out the safe max rpm.

For now, I think I will not be running over 4000 to 4500 until I replace the grease with Isoflex NBU 15.

I think this spindle will always run hot due to the tapered roller bearings. All three bearing manufactures confirmed that with some applications, these bearings do run at 250F or 300F for the SKF. I think replacing the seal with the 47 x 72 x 10 is the best we can do. Even though the hot spindle is inconvenient and will sometime require the use of gloves to change tools, I’m not sure if this temperature will actually damage the cutting tools.

Just my thoughts for now. I will be re-lubing the bearings and replacing the seals.

Best… Richard

Screenshot 2024-02-21 122506.png


For more information on the actual spindle bearing replacement, check out this post:

Spindle Bearing Replacement - CNC Mill Conversions and Builds - Dr. D-Flo (

With respect to this the spindle upgrade, I ran the new spindle motor with the new bearings installed and milled a bunch of aluminum. All went great! As the bearings are not fully broken in, I ran at 4500 rpm for one hour. The spindle temp never exceeded around 84F! I verified little to no runout even when intentionally forcing the spindle laterally.

I am currently running with a 47 x 72 x 10 seal. My previous tests suggested that the tighter seals could add around 35 degrees to the spindle temp. I am going to experiment with the seals a bit more. When running a high helix aluminum specific 1/2" end mill, it was literally throwing chips vertically right at the seal. Ideally, I want as tight a seal as possible to prevent contaminants from entering the bearings. The seal is the only thing safeguarding against this. I 3d printed a jig to allow me to install the seal in such a way that the seal is not driven all the way in, but 3.5mm short. This will guarantee that the seal is not rubbing on the bearing race. My goal is to use the tightest seal possible and keep the temps in range.

Prior to the above test run, I ran a spindle break-in procedure. 500 rpm for 30 min, increasing by 500 rpm every 30 minutes until I reached 5000 rpms. As noted above, I then milled aluminum at 4500 rpm for about an hour. I then removed the oil seal to inspect. The image below shows a fine layer of grease coating the rollers. From what I have read about lubrication, the goal is to lightly coat the rollers with grease but to minimize any excess. Excess grease between the rollers causes heat. I believe that this is the key to keeping these bearing running on the cooler side. I used a syringe to inject Kluber Isoflex grease into the bearings. I then rotated the inner race with respect to the outer race of the bearing a number of times by hand. This expelled a bunch of excess grease out the other side of the bearing. I removed this excess grease with a small dental tool.

I am really happy with the AC servo spindle upgrade and would highly recommend it. Next stop… Rigid Tapping!

Best… Richard