LR4 isn't cutting accurately

LowRider CNC Troubleshooting
1

Hello all,

Over the past several months I’ve put together a LR4. Cut dimensions are 1250 y and 610 x. I try not to play the comparison game too much but I see other users reaching impressive speeds and/or accuracy and I am struggling to do either at this point. Some parts come out +/- 0.25mm, which is perfectly fine, but I have zero consistency. Just last night I attempted to cut several 100mm circles experimenting with depth of cut and climb vs conventional. I will attach the code I was running and some pictures as soon as I can get them but looking for some advice because I believe this is a very capable machine, but I’m about ready to throw in the towel.

I started the machine with all PLA parts, but in the process of chasing repeatable accuracy I have changed the core and the strut pieces to Annealed PET-CF. My uprights are still PLA. The reason for the change (in my mind anyway) is the machine is in an uninsulated garage and the temperatures can get north of 110 deg F in the summer. My thought was perhaps this is causing the PLA to soften and add flex, but there doesn’t seem to be any major change as a result.

I originally used the cheapest 9mm belt I could find on Amazon, I think it was the one linked in the BOM, but have since upgraded to a Gates GT2 9mm belt and that made no difference.

I started with a Carbide ER11 router that I got on sale, and have since changed to a Makita RT0701C thinking maybe the router was the problem. This I think helped a bit but hard to definitively say.

I’ve used a variety of bits from 1/8-1/4 most of them fairly inexpensive, but I’ve had the best luck with 1/4 and 1/8 bits from CIC and I am currently running a Whiteside 1/8 2 flute upcut bit. This bit has been able to get me within 0.25mm accuracy, but my testing yesterday was giving me ±1.5mm accuracy.

On to my observations from yesterday. The first test was a DOC cut. I arranged 6 100mm circles to be cut from 12mm Baltic birch using 1mm doc to 6mm doc at 500mm/min XY and 200mm/min Z. These were cut using Conventional milling direction. The 1mm DOC was the closest to accurate but was still 0.5mm too small in both X and Y. Each successive cutout was worse by about 0.25mm in both directions, ending up with the 6mm DOC being 1.5mm too small in the X and just shy of 1mm short in the Y. X has characteristically been worse than y.

The second test was 2 100mm circles cut in 12mm baltic birch at 3mm doc and 1000mm/min XY using a simple trace cut from Fusion. One circle was Conventional cut, the other was climb cut. Other than that both circles were identical in terms of setup. The conventional cut was 99.5mm in both directions and the climb was 101.5mm in both directions.

My main observation is where the bit plunges into the material I can see on the finished piece where the bit deflected, if you will. On the conventional cut pieces, where the bit goes in is raised (ie the bit flexes into the piece) and on the climb cut piece, the bit walks out.

I took the suggestion of these forums and tried to find loose bolts, but everything appears to be tight. I did notice yesterday that I can move the core in both directions about 1mm or more with out causing the stepper to skip. I don’t believe this is loose bolts as it almost seems to be related to belt stretch (I think I can see the belts stretch).

Sorry if this is a repeat question, I’ve spent considerable time looking for similar problems with no luck. Curious if anyone has any suggestions or solutions as this is getting extremely frustrating for me. I also apologize if this was a mess of info and if I left something out just let me know.

Thanks,

Ben

2

I’m wondering if this is part of your problem. The annealing process is known to make parts not straight anymore. I have sold several PET-CF builds and not one single problem with no annealing.

3

This may be a dumb question but how would the core being slightly out of alignment, or struts for that matter, cause that level of inaccuracy? I would think those pieces, while not perfectly straight are consistent at least? The X axis moves pretty freely on the rails.

4

Is the grubscrew in the pulley on the steppers shaft lose? That tends to give this was exaggerated backlash/play

5

I haven’t specifically checked all of them, but the x axis one was tight when I changed out the stepper last night. Changed from a stepper online nema 17 55ncm to 59ncm thinking that would make a difference. It didn’t. And to further add confusion when I cut the strut plates they were either massively oversized or under (these were the first things I cut with the machine). I ended up have to fabricate my own plates using a table saw and drill press.

6

Did you use a finishing pass? Can’t see it in your text.

1 Like
7

I didn’t last night, but have in the past and it yields better but still off results. I suppose ultimately, the parts were set to cut 0.4mm over with a 0.4mm finishing pass. I messed up the tabs holding them down so the parts were loose after the roughing pass, but even still if it should have cut a 100.8mm circle and it cut a 99.5, 99.25, 99, 98.75 and 98.5 depending on depth of cut.

8

One problem I had with the Carbide ER11 router was that I didn’t know how to use an ER collet. The collect snaps into the nut before you mount it. For me this looked like runout and cuts wider than they should be. Obviously this doesn’t account for everything you’re describing and chances are you were using the collet correctly.

3 Likes
9

Yep I ran into that when I first started using it. Never ran it without the collet clipped in though.

10

Sounds like possible deflection of the cutting tool. Without a finishing pass (which you said you didn’t complete due to holding tab issues), the bit will deflect under a full slot cut. The deeper the DOC, the more deflection The finishing pass has less load, so lower deflection, and closer tolerances.

  • How much “stickout” is there?
  • Does using a 1/4" mill instead of 1/8" change the behaviour?
  • Does slowing down the feedrate (or increasing the RPM) affect the results?

If not tool deflection, then I would look for possible looseness in the core. If you grab the core and twist it back and forth in different directions, does it move/wobble on the rails? The core bearings shouldn’t move away from the rail when you grab and twist it. If it does, you may need to tighten things up.

You do mention that it moves “in both directions”: about 1mm or so. Do you mean X+ & X-, or do you mean X & Y? In either case, you probably shouldn’t have that much movement. Are the belts tight?

And then there is the old standby - grub screws

11

I’ll double check grub screws tonight. I don’t believe they’re loos but won’t hurt to check. As for belt tightness, I think they’re tight too. I made a tension tester, but have no idea how accurate it is. I say I don’t think its the grub screws because when I try to move the machine (±X and ±Y) I can see the belts physically droop, but the steppers are not skipping. When I say both directions I mean ±X and ±Y.

I’ll play some more with it tonight and try some trochoidal milling from Estlcam. The best cuts I have are from running a full 12mm doc with a 25% stepover at 2000mm/min from Estlcam. This all really came to a head when I tried bumping up my speed to 2200mm/min and X axis just skipped steps left and right and messed up a nice chunk of 12mm plywood I was using. So I went back to testing and now can’t even reliably use the minimal settings that are set out in the build docs (1mm doc, 480mm/min, etc).

12

Sorry missed some questions there too. I have tried playing with speeds, haven’t gone below 500mm/min yet. I’ve increased RPM before but was going of off guesstimated chiploads. Something like 0.001” chip load on a 2 flute 1/8 bit at 10000rpm gives me right aroung 500 mm/min. I believe this was on the smaller end of the acceptable chip load for a bit. I was noticing with a higher RPM (~15000+) I would get a lot of squealing.

I haven’t really played with the 1/4 since the chiploads on that make it such that the machine would have to run in the 1500mm/min area and I don’t know it can do that. I may also be completely misunderstanding chiploads…

the 1/8 bit is from Whiteside, 1/4 shank 1/2 cutting depth so probably about 7.8 stickout I would think?

13

I had mine running at 6mm DOC at 3000mm/min with a 6mm endmill. It does work, so something seems horribly off, if even the 1mm at 480mm/min is problematic.

Did you try to gently rock the core? Not manhandle it, just try to move it front to back, like @Bartman suggested?

1 Like
14

That should be too many flutes. A single flute 1/8” is far better for our typical rpm range.

You have been doing this all alone. It looks like you tries so many things, except asking us for some pointers. Let us help out.

if you are only using 1mm DOC you are wearing out the first 1mm of your endmill very very fast. The minimum is usually 1diameter depth up to about 3 diameters deep. (10-12mm is not out of the question for a single flute 1/8”)

Did you use a full depth finishing pass of about .3-.5mm stepover?

2 Likes
15

Another thing I think you might need to look at since you have changed out so many parts is router tram. Let’s see if you have a substantial tilt for some reason. I have an easy printed tool on printables if you do not have something already.

First and foremost, you need to get consistency. The numbers don’t matter if you can’t repeat them. So from all your test cuts (use squares not circle, or a square with a circle in the middle). If your machine is not cutting the exact same thing each time, that is the largest problem by far, everything else is tuning.

A good starting point is a single flute upcut bit (1/8, or 1/4”), 4mm DOC in wood, Router RPM near the lowest speed. (your two flute 1mm doc at ultra high RPM is probably dying in minutes)

2 Likes
16

Thank you all for the responses. Looks like I’ve got some more tests to run. I’ll try to run some of those tonight and see how they go.

Ryan - Currently I only have 2 flute 1/8 and 1/4 bits but will be ordering some single flutes here. The idea that my bit was horrifically dull has occurred to me but didn’t think it would be causing all the problems I’m seeing.

I have run pieces with a finishing pass - I typically have run about 0.2-0.3mm in the setup, no idea what it actually ends up being in actuality. I have noticed I get closer to actual values when using a finishing pass but still not anywhere near the numbers that other users have been reporting.

I’ll print the tramming guide here now and test that. On that note, is there a guide for tramming the machine? I’ve looked around but don’t see it? That maybe an obvious question but at this point almost nothing is intuitive to me about the machine.

Phillip - I have tried just rocking and nothing seems to be moving but I’ll try again tonight.

Thanks again everyone.

17

Always if you are testing accuracy.

If you see the plunge marks try a slightly larger finishing pass.

Just test it and shim the router mounts with some tape if you need to. A little goes a long way.

18

If you have made circle cuts without a finishing pass you will see stairstep on one quadrant of your parts if you are out of wack.

19

I have seen that, figured it was related to the machine cutting out of whack but I can see that being a tramming issue.

20

If you can see it your mount is way out of wack. I can be pretty far out and not see it.

Make sure to loosen up all 6 mount screws. Tighten the two on the same side first. Then the other two. Then clamp the tool. It should be very very close without shims, like 2mm or less over the radius of the tram tool. If you tighten the screws in a weird order it can give it a little twist.