Build in Progress in Maryland

Hi All,

Thought I would just drop in to mention that I’ve begun a MPCNC build. My hope is that it will be able to cut aluminum mounting plates (with holes, pockets, etc) up to 1/2 inch thick – we’ll see if I’m able to achieve that or not… If it can handle 1/4 inch plates I’ll consider it a win, but 1/2 inch is the goal. To that end I’m going to keep the build area pretty small (12" x 18" x 3"). I would be happy with 12x12, but I like the idea of swapping out 2’x2’ spoiler boards (available pre-cut from home depot) so I’m ending up with an 18" axis. I figure if that axis ends up causing me grief I can always make it smaller later and just deal with having to cut down the spoiler boards.

My plan is to use the Dewalt 611 router - That seems to be used successfully to cut aluminum in the OpenBuilds platform, so I’m hoping it’s a good choice for this machine. BTW – I was looking hard at the OpenBuilds C-Beam machines as well, but the cost advantage of the MPCNC eventually won me over.

I’m printing my own parts, but I’ve had very bad experiences with PLA cracking over time if it’s under any ‘squeezing’ load, so I’ve opted to go for PETG for any parts that are under this kind of load (the feet are a good example). At first I was going to go with PETG for everything, but my first roller failed about half-way through the print and the resulting part exhibited very noticeable flex just under hand pressure. So I’ve decided to ‘risk’ PLA for it’s stiffness in most of the parts. Guess we’ll see how that goes as well…

I have a question about the table – it seems that most folks make a table just large enough to hold the router itself. I would have thought that having a bit of working surface to one side would be nice. I’ve planned my table to be about 18" wider than the router itself to allow room for a control laptop (or just general work area). Am I missing something here? It seems obvious to me, but when no one else seems to be doing it I have to ask – why not?

Thanks…

That is pretty big actually, smaller would be easier with aluminum that thick.

The MPCNC was built for the 660, you will get no gain from adding such a large heavy router. Think of it this way, would a 10 HP router help anything or make it worse? I have never stalled the 660 and I have to test things. Lots of people are using 500W import spindles and doing great and I am convinced teh 660 is much more powerful.

By far the best route, extra infill helps as well.

Not at all, I have mine on a large toolbox.

 

I am not trying to be negative but most people fall into this trap and I am trying to get you the goal you want, super thick aluminum. A little extra here, easier precut wood, a littel extra there. Trust me, short and as small as possible, especially the Z like I have written in a few places. You can move the table if you need it but nothing is needed over 2". An extra inch is a lot on any axis. If you are making 4"x 4" x 1/4" bracket build it for that and that only.

Apologies in advance if this is double (or triple) posted – tried cleaning up the quotes and the reply went away???

I plan on taking some deflection measurements after getting everything mocked-up. If I’m seeing more than 0.002" deflection I’m going to reconsider size…

Well, the 661 has two things going for it that I can see: 1) variable spindle speed built in, and 2) I have one. ;)
[quote="vicious1,post:2,topic:11851"]
I am not trying to be negative but most people fall into this trap and I am trying to get you the goal you want, super thick aluminum. A little extra here, easier precut wood, a littel extra there. Trust me, short and as small as possible, especially the Z like I have written in a few places. You can move the table if you need it but nothing is needed over 2″. An extra inch is a lot on any axis. If you are making 4″x 4″ x 1/4″ bracket build it for that and that only.
[/quote]
What is the minimum Z height? For some reason I thought it was 2.5"
Also, one of things I haven't got a handle on yet is exactly how much room is available for tool changes? So how short is too short to be able to switch bits?
BTW - thanks for taking the time to do all you do... It must be a labor of love on your part, because I know from looking at direct sourcing a lot of the parts you make available in your store that you're certainly not making a killing on sales! If I hadn't had a bunch of the necessary parts already lying around your kit would have been a very good deal. Thanks.

Have you ever tried aluminum composite material (ACM)? I would think it has a lot of advantages for custom plates. I haven’t tried it, but I would guess it has most of the strength and would be easier to mill.

I’ve never tried (or even considered) it. Just did a little research and didn’t find much of anything on it’s mechanical properties… I would suspect that since the aluminum skin is thin (very thin) that it would behave more like the sheet of the low density polyethylene filler than a solid aluminum sheet of similar thickness.

It does seem like it could have an application for replacing non-structural 1/8" aluminum panels (basically non-load bearing applications), but I doubt it would be applicable as a 1/4" or 1/2" plate replacement.

I found a site online that allowed ordering of samples, so I submitted a request. We’ll see if they show up.

I just checked my super build (red and black build linked at the bottom of the specs page) with unreleased parts and best I have gotten is 0.005" at 1" with 2kg load, that is what I would consider a very top tier goal to try and get. This is a CNC under $200k…So you will need to master the finishing pass and find an endmill that can handle that sort of depth. I can tell you 1/2 AL is going to be a big stretch at any size. The amount of perfectness in your build is going to difficult to achieve.

 

Protomold is a good place for a few parts and it will give you an idea of cost, value, complexity.

 

Hi All - progress is being made…

I completed printing all the parts and have the machine assembled, wired and cutting foam! Woohoo!

With everything assembled, I’m measuring a .002" deflection at close to the center of the long carriage support tube, when the carriage is in the middle of the bed and I add a 5lb weight directly on top of the carriage. A 10lb weight gives just under .005" deflection. I’m seeing about half of that movement at the outer rails - so it’s possible I could reduce the deflection somewhat by installing mid-rail supports, but since the center tubes can’t be supported, it’s just as likely that I wouldn’t see any improvement. It should be easy to verify, but I didn’t have anything handy to use as rail supports, so haven’t done that yet… In any case, I’m actually pretty pleased with the rigidity I’m seeing right now. My only real concern here is lack of rigidity causing the tool to grab and try to run away with the work-piece when cutting aluminum – not the precision of the parts being cut out…

First test movements were done with a pen. A pic of the crown and a test pattern I threw together:

Scale, precision and squareness are all spot-on (within the measurement accuracy of a 6-inch scale). So quite happy with that!

I mounted the router and did some test cuts in styrofoam:

Gotta say I’m pretty pleased… Although that ragged area on the pic is from starting the program without starting the router - Doh!

So that brings up a question – I clicked “kill” (on Repetier) almost immediately, but it had already sent a fairly large number of gcode steps, so it just kept going for quite a while. Is there a better way to quickly abort a print gone bad other than hitting a master power switch?

BTW - the samples I requested of the composite materials showed up. I have to say that I’m actually pretty impressed with them. I would say that for small plates they would likely be just as stiff (possibly even more so) as an equivalent sized plate out of 1/8" aluminum. I haven’t tried cutting them yet, but according to the manufacturer they should be easily cut with a router. So @jeffeb3, thanks for that suggestion!

 

2 Likes

Nope, have to kill power to stop it immediately. If it’s doing a bunch of small fiddly stuff it will stop faster, but if it’s doing long straight cuts, it could take a bit to stop. Has to clear the cached movements.

1 Like

Which manufacturer? IDK if I’ll ever get to a project like that, but I am curious.

I saw them on the seckit printer. Unfortunately, the designer hasn’t released the composite parts dxf files.

I got the samples from alpolic.com
Haven’e done anything with them yet (and may not for a while)… I did use the new machine to cut some hold-downs out of 1/4" corian-like material I had in the scrap pile and it worked great. Also have done a couple of test carves in oak - also seemed to perform well for those.

The one thing I quickly realized is that some form of automated dust collection is going to be very high on the priority list, so I’ve been working on something for the last couple of days. I really like where it seems to be heading, but have only printed bits and pieces of it to prototype with so far…

2 Likes

I think I’m gonna call my dust boot complete. Still need to track down some lightweight vacuum hose, currently just sticking the shopvac hose on the exhaust tube and holding it up manually during the cut…

I made adapters which mount to the XY bearing blocks which hold 8mm rods which extend down towards the work surface. The dust boot itself attaches to these rods. Here’s a pic of the right side:

The left side also provides support for the exhaust tube (and is difficult to see when it’s all assembled):

This approach keeps any stresses introduced on the XY assembly and away from the Z-axis.
The “window” over the top of the boot is held on my a couple of magnets, and the original plan was to cut this out of some thin clear plastic to allow visual access to the bit when running. What’s on there now is simply is just a thin 3d printed (opaque - black) window.

I’m happy enough with the results - it’s definitely not perfect, but the amount of dust that escapes is significantly reduced:

So - with that done I started back on making some practice cuts, and I’ve run into a bit of an issue that hopefully someone can offer some advice on… I’m doing a sign. Used Fusion360 to generate a SVG of the text. Imported that into Etlscam to generate the tool paths using carve and a V-bit (including huge pics here so you can see the details):

Here is what it looks like cut into foam (Which I think looks reasonable):

And this is what I’m getting in oak (scrap hardwood flooring):

As far as I can tell the piece of oak has a reasonably flat surface, so I don’t think that’s the issue. It’s certainly not out of flat enough to account for what I’m seeing at the top of the 5.

I suspect the issue is that the bit isn’t cutting to full depth all the time (I don’t think it should be trying to take more than a 3mm doc?) After seeing this I did some push/pull tests on the gantry. With the motors powered I can’t generate any movement on either X or Y (with reasonable force), but I can ‘lift’ the Z axis a small amount (almost like a rocking motion). I took a look at the bearings holding the z-axis tubes and noticed the following:

  • On the top set of bearings the innermost bearing (on both sides) is touching the tube, but loose enough to spin by hand.
  • On the bottom set of bearings all ‘left’ bearings (looking towards the tool mount) are snug against the tube.
  • On the bottom set of bearings the ‘right’, innermost, bearing is very loose - not touching the tube.

I’m suspecting this may be the source of the Z-axis movement (and of course I’m suspecting this is the cause of the poor carving results)…

Thoughts? Suggestions on what to tighten to bring all Z bearings into contact (assuming that’s a valid desire)? Anything else I should check on?

Thanks in advance for any pointers!

1 Like

I would replace that vac tube with a flex hose and anchor it to the gantry. Many many times we see all the errors coming from the torque multiplier hidden in the long vactube getting tugged on by the hose.

1 Like

I’d say that this looks like backlash from the Z axis. It doesn’t seem to go down enough on some spots.
Normally the Z axis should compensate the backlash by itself because of gravity, but maybe something is preventing it to do so, could be some bearings or something rubbing against something else. V carving bits are not really good at plunging so it is possible these efforts together with more friction are enough to overcome the gravity pulling the gantry down.
To quickly confirm this theory without having to check too many other stuff, maybe you could redo this cut, without the shop vac attached to the tube and adding some weight on the Z axis.

1 Like

It looks like in the foam, it was actually cutting too deep. The tips are rounded over. I think you need a finishing pass or configuring the depth steps to take one more cut. Maybe even just run the gcode again. When the bit reaches the tips of the letters, it looks more accurate, probably because there is less load pushing the bit up. However, you get almost the same effect on the E, so maybe not…

1 Like

First off - thanks to all for the suggestions.

I tried tightening “Tension Bolts C” to bring all 3 bearings on each of the XYZ plates into contact with the Z tubes… This appears to have had zero effect. I haven’t loosened those up yet, so all of the following tests were done with those bolts tightened. I do plan on loosening them back up.

I’m pretty sure that I’ve tracked the real issue down to the v-bit I was using being a countersink bit, not a bit meant for routing.


For my original cuts I had eyeballed the angle of the V at about 75 degrees. I later took a protractor to it and got about 80 degrees. I now suspect it was actually supposed to be an 82 degree cutter, but further testing, as described in this video:

leads me think it’s probably closer to 78 degrees.

However, I haven’t managed to solidly confirm that yet because I think I’ve discovered a flaw in the way I was putting together some of my susequent test patterns in estlcam. I was taking a part of the pattern (the 5 in this case), setting up an inside carve with one of the tools I had defined (let’s say with a 77 degree end configured), I would generate the gcode and run that test, then go back to estlcam and change the tool for that path to another tool (76 degrees for example), then regenerate the gcode and run that.
I now suspect that estlcam was not actually regenerating the tool path when the tool was changed (without deleting the original path and re-applying it) and since I was cutting with the same tool, I think I was seeing the same results in my cuts:

I’m planning to go back and loosen the C-bolts a bit, then do some more testing (probably focused around 78 degrees) with this cutter. But deleting the tool path and regenerating it in estlcam with the correct tool each time. I also plan to order a few actual carving v-bits.

It’s funny that carving wasn’t my original intention with this build, but it just looks like it would be a lot of fun once I get it dialed in…

Also, just a note - I’m pretty certain that the vacuum exhaust isn’t putting any stress at all on the Z axis, and there’s no way it would be putting enough stresses on the XY gantry to affect that. Eventual plan is to suspend the vacuum above the center of the machine to minimize any torquing.

Also, just another footnote, early on in the testing I switched over to an 1/8" straight bit to see if I was actually seeing any XY deflections and everything turned out exactly like I would have expected it. In pine:

and in oak (the top edge looks bowed, but that is just the shadows, it’s straight):

2 Likes

You’re going to be mad when you try a real v bit and see how much better it cuts.

1 Like

Ah yes, if you used a wrong bit then it explains a lot!
Try one purposely made for carving and it is very likely to solve your problems, the straight bit test seem to be very nice so that’s a good sign.

1 Like

I suspect you’re right - although I’ll probably be more happy than mad… :slight_smile:

I spent some time today figuring out that bit seems to actually be 77 degrees, and at this point it’s probably pretty dull. As it spent the day (mostly) burning through MDF. But eventually I got a decent looking “5” carved out. At this point I’ve decided to just table the carving until I get some real carving bits (just ordered an assortment from drillman1 on ebay - after finding him recommended in the archives)…

Had some success today cutting 1/4" aluminum plate. Just for grins I recorded the cut and uploaded it here:

Honestly though, I can’t in good conscious suggest anyone watch it, unless they’re really into shaky, blurry, screechy, and even sometimes out of focus video of what turns out to be a mundane cut. :wink:

Just because I’ve found value in others posting their tool configs when cutting aluminum I’ll include mine here as well:

This was also my first reasonable length cut with the dust shoe attached and vacuum running. It definitely didn’t catch everything, but overall I think it did a reasonable job containing most of the chips.

At this point I’d have to say that my build is performing well enough, still thinking that 1/2" plate shouldn’t be out of the realm of possibility…

EDIT - One thing I did notice is that the center hole is about .012" undersized. I had done a similar cut earlier (evidence of which can be seen on the plate) with a 2 flute bit and that hole was about .003" over. The earlier hole was also a helical drill, but it was done with with an incredibly slow feed rate and resulted in only the finest of shavings (not chips). Part of me wants to ‘blame’ the size of the single flute cutter not being exactly .125" dia, but that is difficult to verify.

I ran this part with a finish pass, but is there a way in estlcam to force multiple spring passes on the final cleanup? A setting for this didn’t jump out at me.

Also - just to be clear here - I think 12 thousandths is plenty good for what I’m looking to get out of this machine. I’ve got a real mill for dealing with critical parts and even on that I’m more than happy with .003 accuracy. Just curious if a couple spring passes would bring it tighter.

1 Like

I love the video and thank you for posting your details. Makes it great for follow up questions.

Solid cut, I’m proud!

The hole was undersized but the out dims were good? If that is the case try doing the finishing pass slower on the hole give the single flute a chance to hit it a little more. The other option is leave a little more or less for the finishing pass. (something I intended on testing at some point). There is a sweet spot, too much and it is not a finishing pass, not enough and it can work harden.

1 Like