Holy cow, if you have a 3.93" long bit you will not have good results. Even running a 4" long 3/8" diameter endmill in a WAY more rigid machine (Haas TM-2) you would get chatter.
Sounds like a drill bit?
I donât mean to criticize the use of these 500W spindles, I have seen a lot of activity on the forums asking about or using them, and people have been able to do some fantastic work with these cheap tools. They are a quiet and economical option for light cutting, and are well suited to many applications.
However, I have had a number of issues with mine (chatter, vibration, poor tolerance and finishes, etc), and as a researcher by trade I feel compelled to understand why it has been performing so poorly for me. I have seen documentation of other low power DC spindles (like in the Carbide3D Nomad that is only 70 Watts!) performing so much better, so I know that the power alone is not the culprit. I have also used the same MPCNC frame with the much larger Makita at 10K rpm, with no problems with no vibration, so I know that the frame should be able to handle cuts at those lower speeds. I tweaked CAM profiles for months and designed a set of extremely low-profile mounts to get the spindle as close as possible to the z-axis, but I would still run into issues with cuts that should not have posed any problems. Something has to be going on with the spindle itself, and I want to find out.
A comment in another thread mentioned that the bearings in these spindles might be seated in rubber. This would allow it to feel stable and keep the runout low while idle, but as soon as you get cutting or the tool heats up everything can fall apart quickly (especially if you have a long tool or a less than rigid machine). I found a video where a guy takes apart and critiques a 300V Quiet-cut spindle (https://www.youtube.com/watch?v=7P5THNNjafU). He finds these rubber gaskets, as well as some other issues that he thinks explain the EMI issues. This was a slightly different model than the 500W spindles everyone seems to be using now, so I figured I would take mine apart and see what was in there.
[attachment file=âDissassembled Spindle.jpgâ]
After opening it up it was pretty clear this was just a slightly scaled up version of the spindle in that video, no significant improvements that I could see, and a couple of issues that werenât mentioned.
The bearings are the common 608zâs that we all used during he build process. These are not ideal for this application (they are not designed for simultaneous axial and radial loading), but I donât think they are the major problem here. On definite plus is that we all probably have some of these bearings on hand if you ever need to swap them out.
Unfortunately, the rubber seats mentioned in the comment and the video are present, and were much thicker and looser than I had anticipated. The bearings and rubber seats could easily slide out of the top and bottom plates, indicating that the rubber is not compressed in its natural state. The rubber sides are over 0.5mm thick, and I could easily compress it with my finger. The radial forces involved in cutting could definitely deform this rubber, and any deflection at the point of these bearings will be multiplied at the collet and even more at the tip of a long endmill. This will allow the cutter to be push away from or be pulled into the material, causing the violent chatter that has been reported.
[attachment file=âShaft with upper and lower plates.jpgâ]
[attachment file=âRubber Thickness.jpgâ]
I also noticed these wave springs between the rubber seat and the front bearing. I believe the purpose of these is to provide an axial pre-load on the bearings. Ideally, the screws that connect the top and bottom plates would provide this preload, but the plates instead clamp the body of the spindle. The manufacturers needed some way to put some pre-load on the shaft, and I think this was there quick and easy solution. This seems questionable, because any tension on the shaft (e.g. cutting with an upcut endmill) can cause the entire shaft to be pulled down relative to the body of the spindle. This would cause deeper than expected cuts, exacerbating any other problems with vibration or chatter.
[attachment file=âWave spring.jpgâ]
I am not an electrical engineer, so I canât comment on the design of the electronics inside, but if you are interested in more details I am happy to provide more images or measurements of those components.
Again, I donât want to say that you cannot have goods results with these spindles, clearly people can and have, but I think it is important for anybody considering them or currently using them to be aware of their capabilities and limitations.
2 Likes
I understand that the longer the bit the more chatter. However it worked fine in a fine milling operation even on a 500W spindle, I cannot complain.
Itâs R2xD6x60x100L from Aliexpress.
Eric H, thatâs a valuable insight, thank you for that!
Hi. Today I tried to do a full cut with 1.5kw water cooled spindle. Which is around 3kg on itâs own plus some water mass running through.
The piece gives 24000 RPM and I expected it to cut better than 500W air cooled one.
However, I noticed that it:
- makes a lot more burrs. Sometimes even chips off small parts on roughing 3D job
- is very hard to move, MPCNC's mena 17s move in jerked manner trying to move it
- when RAMPS is turned off, the tool slides all the way down - resting friction cannot hold it in place
- High mass makes it hard to stop and reverse on X-Y directions - little vibration happens there
On the upside I noticed that it spins perfectly without any clue of vibration on itâs own - nice to look and hear. The only benefit - to sit and watch it like a TV 
I will use it when I grow into havier router design.
1.5kW is a very strong spindle, we run a 600W and I have seen no reason to go larger. I bet it is a beauty though, maybe one day I will make a design for something that size!
1 Like
Sorry for being off topic.
You really went above and beyond to figure this thing out. This is good info, good on you for sharing, maybe update your original post about the spindles or make a separate one to make it easier to find if someone is looking into it in the future.
1 Like
Hi,
Today working with 500W again I noticed that machineâs steppers skip steps when it changes direction and has a lot of material to cut at the same time. By âa lotâ I mean 6mm flat bit, 2.5mm stepdown, 900 mm/min feed and the bit touches material from 2 sides (like performing kind of round movement of very small radius fast).
Skipping definitely comes from trying to change direction and accelerate fast still having to do some job on material.
I have been thinking about possibility to gradually reduce/increase speed during changes of spindle movement direction. Is it possible to do in marlin itself or should I search for such possibility in artcam/estlcam?
Thanks.
I donât know about artcam/estlcam but maybe look for a setting called âtrochoidal millingâ in one of those softwares. It is a strategy designed to avoid spikes in tool engagement like youâre talking about.
Surely you mean 900mm/min also?
Yes 900 )
Thanks, I 'll check that.
â
Can it (the one that I was thinking of) be #define DEFAULT_ACCELERATION in Marlin?
Skipping steps is a sign of something drastically wrong, it is not a common problem, never has been.
Are your spindle wires parallel to your stepper wires at any point? The higher current in the spindle wires can induce current in nearby conductors and interfere with step signals. I had some problems with skipped steps when I was using a 500w spindle due to what I believe was this issue. I had the wires powering the spindle parallel to some of my stepper wires, and every time I would have high engagement I would lose steps. I thought it was mechanical at first, but when I tried rerouting the cables the problem went away.
Also, regarding slowing down at areas of high engagement. Troichoidal or Adaptive clearing are definitely great strategies for consistency. However, if you still want to do simple contour paths (slotting, finishing, etc.) there is a optional feature in Fusion 360 called âfeed optimizationâ where the feedrate is reduced at corners like you want.
And - yes, it is.
You can set such acceleration by DEFAULT_ACCELERATION or on the fly sending command M204 P200 where 200 is acceleration.
I found 200 to work well (previously was 400). And you can slow it down even further thus making it move super smooth but it will greatly affect milling time.
Eric, thanks for the info. I will check out.
And no, spindle wires are completely apart from steppers wiring. I knew that people have problems with it so didnât want to take risks.
If anyone cares or is having similar troubles - a lot depends on the bit itself. When I discussed this here I was working with 3 flutes bit made for aluminium, from Aliexpress. I had nothing to compare with so I thought that the bit is not applicable for blaming. When I received 2-flutes bit made for wood I realized that half of success is within the bitâs responsibility.