Yes. It is a question of thermodynamics for me. Xcarve is spending money on marketing, training, HR, giving discounts, especially to larger customers, that is all coming out of the pocket of the average consumer. Plus, I don’t think Ryan charges enough, and he can’t charge what xcarve does, because it is all easy to source, so many people would just source it themselves.
2 Likes
I would also like the MPCNC to stay in its current “niche”. So no ballscrews, no NEMA 23, etc.
Zip ties are a different story. For a couple of bolts and a little bit more plastic, you could have adjustable and easily removable belt tensioners. This is nice functionality to have.
Also, there are many after-market upgrades, but it’s hard to know which ones are relevant and which ones work best. A relatively easy way to make the MPCNC better is to choose one particular design for each kind of upgrade, and either incorporate it into the standard parts (e.g. belt tensioners) or maybe offer it as a V1 engineering supported/recommended upgrade (e.g. a dust shoe). For the latter, it may help to make some small changes to the standard parts so that the upgrade parts can more easily be attached to them, e.g. better mounting points on the gantry for a dust shoe.
1 Like
Yup, this has been kicked around since literally day one. There are problems with both systems. I am trying to come up with something just so I do not have to talk about them any more. Some people understand how to use the zip ties others just make giant loops and complain about them.
I don’t think that is true. Other than a PID controller I do not think there are any upgrades. There are mods to let you do more specific things but if they made it a better machine for the average user we would be using it if I got permission. This is tricky. Since you mention a dust shoe, this is the hardest for me. I know everyone wants one, but it is by far the largest source of error you can add, next would be Z axis mounted drag chain. This is not the most rigid machine in the world, so if I make an official dust shoe, first it has to be awesome…and fit lots of hoses, second I have to be comfortable that too much accuracy is not lost and at this time I can not promise that. I use my machine as well and having a dust collection would be cool but it does not let you watch you cuts closely, encourages you to leave your machine alone, and it kills accuracy, and they simply take up too much space. A small vac hose is almost as big as a 660, and some guys insist on using 4" vac hoses. The new parts have more screw mounting holes but I still do not have a dust shoe I am comfortable with. I added it to the LR just because those would be very long jobs and there was space. Dust shoe is on the list but not a priority for the MPCNC. If anything I think an air blast and a dust slot on the side is more appropriate. For the work we do and the beginners that should be studying their CAM.
2 Likes
So I really want to include everyone in the new design but with all the hate I now get and all the threats of re releasing all my stuff CC0 I unfortunately feel I need to keep things more under wraps than normal. The regulars around here I am sure know exactly what I am working on as I talk about it in all sorts of threads but I do not have it all in one spot like normal. I hope to include some in a private beta…that hurts my heart I have never done anything like that in private as I do not want to exclude anyone but I assure you there are some nasty people out there and I would be devastated if they messed with my new stuff.
6 Likes
The other day I noticed that thingiverse can recognize variables in openscad designs… so maybe a future design could be based on a variable tube size and then users can select the tubing size that is available to them (or they feel they need) for thier build. To be honest though, openscad is not for the faint of heart, BUT it could resolve the imperial vs metric problem in tubing sizes and allow users to experiment with different sizes.
1 Like
Releasing the design files creates massive headaches for Ryan. I had contemplated the possibility of a parametric stl generator similar to the thingiverse functionality, but with the difference that the design files are kept private. So watermarks are not so trivial to remove for example. In the end I think it still creates problems with too many unofficial copies reflecting bad choices and then people have problems. That discussion is here.
3 Likes
A dust shoe doesn’t need to be all things to all people, but you should be able to stand behind something you endorse. If there is no design that you’re comfortable with, then I agree that it’s best not to endorse anything.
3 Likes
Not necessarily any more. The new TMC5160 drivers have changed that radically: It is a cheap and capable driver chip with all the back EMF measurement goodies but it uses external mosfets to pump current through the motor. So in theory the sky is the limit with these chips (well in practice it’s not that simple but I’ll not dive into that). The fancy new Duet 3 boards have six of these with big enough mosfets to support 4A RMS each. This is well in NEMA 23 territory and the entire control board costs $230 at the moment which starts to drift into the “cheap” realm already, especially if compared to a setup with dedicated driver “boxes”. I’ve also seen some cheap “oversized” stepsticks appearing. It’s only a matter of time one can get a board with 5x4A capability for less than $100. You do want at least 24V though but due to LED lights being super popular nowadays, these power supplies are dirt cheap. To be honest, even with NEMA 17-s, I’d rather run the system at 24V to give me more headroom at higher speeds.
Now that said, 0.59Nm NEMA 17-s are more than enough for MPCNC.
Here’s a video of my MPCNC lifting about 12lb of weight with its X axis (the big lead block is 10lb, the small is 2lb), no problem. Cutting forces like these are in the “big boy” router territory. If I were to apply a force like that to the end of the cutter, MPCNC middle gantry would bend like a reed in the wind.
4 Likes
I did some research when this blew up but since it wasn’t rigorous enough I didn’t post it. I’m posting what I have now, maybe it will help just take it with a grain of salt.
-
I put zip ties on my X and printed blocks on my Y axis. I measured the “stretch” by loading the gantry in both directions and I was not able to find a difference that would be lower than my repeatability error. I don’t remember the exact numbers any more but it looks like the belt stretches more than zip-ties if you have tight loops.
-
Belt stretch is fairly linear (I recall a thread in V1 forums about this as well) so the exact tension doesn’t seem to matter much as long as it is within a range. Gates recommends about 6lbs, Shapeoko guys run their machines at 10-15lbs. There’s a long thread here where they have contemplated a bunch of options. I’m willing to bet that there is actually a very little difference between the extremes with the belt possibly wearing out sooner at the higher end.
-
Shapeoko is designed to use the gravity (weight of the machine) to tension the belts without any eccentric hardware. It’s an interesting idea, although not directly applicable to MPCNC.
Personally I would be comfortable with a “prescribed pre-stretch” system where I attach a plastic block at the end of the belt, pull it taught by hand and then push the block to click in place at the corner. There will be an angled surface that moves the block back by let’s say 2mm before it can click into place and this movement puts final tension on the belt.Simple, easy to remove if I need to mess with the motors/belts, no additional hardware and almost impossible to over-tighten.
EDIT: Actually I don’t think we need different blocks. Pulling the belt taught by hand should be enough to compensate for longer belts stretching more.
3 Likes
The implicit tenets that lured me into building mine are the following:
- Total cost is low enough to risk it all
e.g. if I end up using it only once or buying a different machine in the end it’s no big deal. Shipping is a huge part of that equation and I think this is the single most important quality that makes MPCNC unique. - Mistakes are cheap
MPCNC is weak enough not to ruin any expensive parts of the machine if I mess up my CAM. Sure, it can break a few plastic parts, but parts are easy to re-print. My machine is made of PETG and it’s been surprisingly resilient to mishaps. I had an incident where the cutter was sucked into the material and bent the middle to a nasty angle as a result. I thought this surely must of broken something but no, the plastic just sprung back with no ill effect. I have not been able to break a 1/8" cutter yet either, I guess I have to pick up some steel to pull of that “feat”
- Accuracy and part quality is good enough or better
One can achieve amazing results with this machine. DW660 is so far the least accurate part of the entire system. I’m currently researching a way to improve this because 0.1-0.16mm of runout is a tad too much to my taste.
Now all this means we have to compromise elsewhere. We can’t make it tall, we have low material removal rate etc, but for hobbyist the most important aspect is an ability to produce that part we want without spending a fortune and this is where the current offering ticks most of the boxes. The only thing I can wish for is a more rigid L-shaped version (similar to traditional vertical mills) that could get to approximately 5"^3 work envelope with good precision and rigidity. The kind of semi micro-machining application (using 1/8" or smaller cutters) that could be used to produce the occasional odd metal part.
5 Likes
That first point is something we don’t talk about enough. I never would have spent $400 on my first printer, because I didn’t think I would use it much. I went 50-50 on one that was $350 and I have spent much more than that in the last 4 years that I had it. When I made my MPCNC, I was sure I would never spend $1000 on CNC. But printing my own parts, using some spare parts, and cheap EMT and I made it for less than $250. Not needing to “prove” that it was worth it to anybody made it 10x easier to jump in.
The level of “proof” for each person might be much different. $250 is a big investment for some people, and other don’t blink at $1k.
The other reason is, that it was fun, on it’s own. Lego is the obvious example, and I feel a little silly spending money on lego for myself, but it is as entertaining to build as it is to play with. The MPCNC really hits this part of my brain. It is entertainment in a kit, not just a CNC in a kit.
6 Likes
One wise person once said: The only difference between men and boys is the price tag on their toys …
2 Likes
Thanks for that comment. I haven’t really looked at them in terms of what they can drive. Mainly thought of them as silent running and easier control. Another big data point in my quest to explore all the different hardware choices available.
Just watch out for some stepsticks claiming to be able to do 20A. Sure the mosfets may be rated to 20A but in a stepstick format they are hardly possible to cool well enough to sustain such current. Modern SMD mosfets are cooled via PCB so even a big heatsink on top will not help if there is no sufficient PCB area to carry away the heat from the junction. Somewhere in the fine print they normally claim a lot lower number but even then I would add a big asterisk next to it …
The other issue is that sense resistors are normally not set for such current. I could believe claims of 3A in stepstick format or maybe 4A if there is a blizzard blowing through the area all the time.
1 Like
Hi all,
I understand the challenge heavy/stiff/speed.
Because of what available to me and because I’d like to have a table not too small , 3’x3’ table.
I plan to mill small aluminium parts, wood and also to cut motorbike vinyl coverage)
I have access to stainless, OD 25mm but only 2 thickness
0.078I" and 0.049"
Would it be a good idea to use the thicker for the frame and X Y axis and the thinner for the Z axys ?
Thanks
Outside rails have only vertical loads, and you can support them mid span, so you could go thinner. But then the weight doesnt matter so it’s just the cost difference.
The z axis does not really deflect from bending of the tubes. It deflects from the center assembly. You could go thinner but again the weight is not huge so there is not much downside to going thicker.
The gantry rails are most important stiffness-wise so you sould want those to be thicker. Yes it adds weight but I think it would be worth it.
And in general I don’t think that weight matters that much if you are milling. For a 3d printer or pen plotter there is advantage to higher accelerations but milling it’s okay to haul around a little extra weight.
1 Like
Thanks Jamie for you help.
The thicker is available 6 meters or nothing, so I’ll use it on rails and Gantry.
nothing left because of the size I want so 'll go thinner (sold cut the size you need) for the Z
That’s basically what I did for my lowrider (still under construction): 0.083" wall tubing for the X rails and 0.049" wall for Z.
I was also considering 0.095" 1020 low carbon steel. It would of been almost 2x cheaper than stainless but I would have to protect these from corrosion somehow (probably smearing on a coat of inhibitors from time to time) so I decided the savings would not be worth the hassle.
I just got in some 1"/0.065" DOM from speedy metal. I could not test it with a full width since I only have 4’ lengths but it is certainly better than the SS I have. It is hard to say but a 46" span seems to be somewhere near 0.5mm deflection, vs the Stainless at about 1mm. Those numbers are not to be trusted too much but it is stiffer. The cold working really does seem to add a lot to it. On top of that it is extremely consistent 25.45-25.49mm round and smooth. It will certainly need some coating as we talked about in the other thread, paste wax or some other similar non silicon dry coating.
Less than half the price…should I even make a 23.5mm version of the new version? For that price a little wax is no big deal!
https://www.speedymetals.com/pc-3494-8242-1-od-x-065-wall-dom-steel-tube.aspx
https://www.speedymetals.com/pc-4494-8276-1-od-x-0065-wall-tube-304-stainless-steel-annealed.aspx
The 1010 steel pipe that I tested above was of the DOM variety. It is slightly stiffer than stainless ones I tested, pretty much as predicted by theory. Theory predicts 1mm deflection for a 48" span when material has 193GPa modulus, however an improvement of only 10% or 0.9mm is predicted at 210GPa.
That said, DOM or “seamless” as it’s called when applied to stainless pipes is a really nice process and that 1010 pipe was by far the most consistent and the straightest of the bunch that I tested. I went for seamless stainless because of a very good price I got from local metal yard, if I had to choose between welded stainless and DOM carbon steel, I’d probably pick DOM steel. Especially if we can find a good way to keep it from rusting. Would ACF-50 be a good fit?
1 Like