OptiMill MB4 CNC conversion (part 1)

Bigger, better, faster milling machine

After years of pushing my little X2 Mini-Mill to the limits, and trying to cut aluminium on my large router table, it became clear that I need a bigger machine, that can handle metal with decent precision. And of course it has to be CNC, and it should be a capable beast. Fast, precise, versatile.

Selecting the machine

After lots of searching on the web, and talking to a friend at the local makerspace, the Optimum MB4 came to my attention. And of course the recommendations of Stephan Gotteswinter add a lot of weight. After some good experiences with my Optimum lathe, I went to Stadelmann (swiss Optimum dealership) to have a look, and to compare it to the other OptiMills (BF30, etc.). Of course I packed my bag with dial indicators for rigidity testing.

Here it is in the shop:

Compared to the BF30, which is roughly the same size, the MB4 definitely is the heftier, sturdier machine in any way. The deflection from spindle to bed (all ways locked down) with as much muscle I could muster was 0.02-0.03 mm, about half of the BF30, in this admittedly unscientific test. Just looking at the size of the Z-column explains this nicely though, it’s much fatter. The MB4 has a 6-speed gearbox, and a 2-speed selector for the 1.5kW motor, for a speed range from 80 RPM to 3200 RPM. While I like the vario drive on my lathe, and the BF30 offers Vario, I think having the super-lowspeed option is nice for tapping, and the gearbox will offer far more torque. The maximum speed of 3200RPM is a bit low for aluminium and small tools, but more on that later.

I also liked the 25mm lead screws, as this would allow me to put in some hefty 25mm ballscrews. The big advantage of the BF30 is that there is an official CNC conversion kit available. However, it is quite expensive, particularly when also buying the optional ballscrews (and a CNC mill without ballscrews is hardly worth doing). So I decided to go with the bigger machine, and do the conversion myself. The MB4 is also the less expensive machine, another plus.

Delivery and setup

The machine was delivered fairly quickly after ordering. Time to take it apart! 🙂

     

The parts looked pretty good. Following GTWR’s advice, a quick deburr and thorough cleaning is always a good idea. Next: taking measurements, creating a CAD model, and designing the parts for the conversion. While that is happening, I also welded a solid base for it. I didn’t like the base offered with the mill – not much storage space, and you can’t get a pallet jack underneath. So I built my own:

(continued in part 2)

8 Replies to “OptiMill MB4 CNC conversion (part 1)”

    1. I really should finish the write-up of this… unfortunately I’m quite busy with work right now. For the CNC drives, I use NEMA34 hybrid servos from AliExpress (closed loop with encoder). They work really well, rapids up to 6000mm/min. I also replaced the leadscrews with 2505 ballscrews. I hope this helps, and hopefully I’ll get around to write up the rest soon…

  1. Hello,

    I am looking for a heavy mill to work with steel and my favorite is the MB4. I want to convert the machine to CNC and would really interested in your write up.

    Did you have to modify the bed to install the ballscrews or could they are mounted without milling the bed? I have only a CNC mill out of aluminum and can’t really milling the heavy bed of the MB4 and hope ball screws could be mounted without heavy modifications.

    Do you have pics from the conversation?

    Best regards,
    Jan

    1. Hi,
      thanks for your interest! Good timing – I was actually just writing part 2, with CAD models, BOM and lots of pictures of the build. It should be up on the site now.
      I only did some very minor modifications on the MB4 itself, all of them on a tiny X2 mini mill. I flattened one spot of the Y axis to allow the ball screw to sit a bit better, but I think it would have been possible to do it without that, by changing the ballscrew mounting piece a bit or grinding off the flange on the ball nut a bit more (you’ll see, I did have to do a bit of grinding to fit 2505 screws). I hope part 2 will answer your questions.
      I’m very happy so far with the MB4. It’s a solid machine. I did have a little bit of stick-slip on the Z axis (about 0.05-0.1mm), which I think was due to a badly matched gib on my machine (slightly wrong angle in the taper). It is probably also an issue caused by the big front-heavy head. Some people reported to have fixed it with gas springs or counter weights. Of course it’s not an issue for manual machining. For now I fixed it by adding backlash compensation in the software, and it doesn’t bother me that much for what I do. I haven’t measured in a while if it’s still there after running in. The X and Y axis are very good. I also changed the grease in the main spindle bearings, as they got a bit too hot for my taste. I think with these machines there can be minor issues here and there, that some people have and others don’t. But all in all I think the MB4 is great, and at a good price. Looking at the BF30 in the dealership, the MB4 is definitely a lot sturdier and more rigid.

  2. Brilliant ! Thank you for the detailed documentation of your project. I’m keen to convert my MB4 to ball screws, possibly with a CNC conversion in the future, but more for what I perceive as smoothness and better feel as a manual mill. Have you tried while building this, to crank it with the handles as a manual machine ? I’m curious if there is an improvement in feel, ease ? Is there a loss of fine movement, as I think the ball screws are moving 5mm per rotation v’s 3mm for the std lead screws ?

    What did you do about the end/support bearings ? From your pics it looks like you used the original thrust bearings and mountings with no radial bearings ? I see with many of the ball screw suppliers, they have bearing kit/block that you’d have to mount in there.

    Thank you 🙂

    David

    1. Hi David,
      thanks for the nice feedback!
      Yes, I did try to turn the handles with ball screws mounted, and it’s very smooth as one would expect. I can actually still put a hand wheel on the right side of X (motor is on the left), but the motor adds a lot of resistance so it’s not very smooth like that. I was originally planning to keep the option to use it as a manual machine, but for the Y and Z axis it would have been too complicated. I often use it “semi-manually” by jogging and positioning it through LinuxCNC, which then also has the advantage of offering a “digital readout” and consistent motor feeds.
      I would say, using ballscrews is probably a bit smoother than the original leadscrews, but that depends a lot on the quality of the original screws as well, and how tightly they are adjusted. For a manual machine, one can leave them a bit loose and accept a bit of backlash, which will reduce friction. Then there’s probably not much of a difference. The main reason for ballscrews is to eliminate backlash – this is also nice for manual machining, but with DROs maybe less important now. One major difference maybe is that ballscrews can be back-driven. This could be a good or a bad thing – for machining fine features with small tools, maybe it will give better feedback. But with big tools, I’m wondering now if it would be counter-productive if the handwheel “kicks back” when the cutter engages.
      5mm vs. 3mm – I suppose you’d lose a bit of resolution, but it probably doesn’t make that much of a difference. The diameter of the handwheel, and how steady your hands are, also come into play. I actually have a small Mini-mill with 1605 screws and steppers, and often used it manually, but then I also never really did manual precision work on that one so I can’t say.

      The support bearings are all still the original ones, so it’s a pair of thrust bearings, and I think just a bushing on the opposite side. It seems to work just fine. The thrust bearings do have a bit of a groove, and when loaded will also provide some amount of radial support. And as I’m driving them with a motor, there shouldn’t really be any radial loads on the screw anyway. The bearing blocks from suppliers have bearings that look like radial bearings – I assume they’re deep-groove bearings. For thrust, I would imagine that they’re not as good as proper thrust bearings, or angular contact bearings, but I’m not an expert for that. There’s probably also a big difference if the blocks are cheap chinese no-name, or high-quality brands for 10x the cost.

Leave a Reply

Your email address will not be published. Required fields are marked *