Accuracy of ANY kind

2 posts were split to a new topic: Difficulties learning Carbide Create

Yeah, those small grub screws just couldn’t handle it. My theory is some were pretty short and not making a lot of thread contact in the pulley and there was just enough force that it would rock back and forth breaking free. A longer regular grub screw would work too so it would get more purchase higher up in the pulley threads. But the clearance is not a problem and the socket head screws let you use a larger allen wrench to apply a little more force.

I also applied a dot of the low strength thread lock to the pulley shaft. Carbide actually recommends this in the instructions for the HDZ coupler, apply a small amount to both shafts and screws.

Lots of people will give you warnings about doing this with thread lock and larger screws. But ultimately you’re talking about a couple dollar pulley you can buy any where. If you strip it out, oh well, put a new one on - and really have to try pretty hard to do that.

When I switched to the HDZ, I reused the X motor, but had to pull off the pulley and change the orientation. The purple thread lock held for 2 years, but broke free with nominal force on the new set screws and the motor shaft. Cleaned it up, flipped the pulley, and put a tiny amount of thread lock back on the shaft and set screws again.

Just make sure you’re using the low strength (color varies by brand) meant for this application, not the high strength red!

2 Likes

What if you install a longer set screw/grub screw and ran a jam nut down on it. Could use 2 with a lock washer between.

Couple things on holding down thicker pieces I’m not a fan of 2 sided tape on my shapeoko 4 I run a 3/16 slot on each long edge leaving about 1/8 or so of meat left to clamp down on both sides of my piece if I have a flip program I do the same on the other opposite edge if not I do just one side I even run the grooves on 3/4 inch pieces using my table saw and it works great . One other thing if all possible set your x y z from the center of you piece for better accuracy and when setting it up use a 1/32 bit to pin point center that helps me alot then insert bit you need to use I have the bit setter so it always rereads bit height or you can raise up router insert bit then lower back down to set z

1 Like

Set screws are successfully used in all sorts of applications including heavy duty robotics. It’s probably more important to use high quality grub screws over cheap a shcs. On my 2015 SO3, I don’t think I’ve checked the set screws since I upgraded to an XL (maybe 2017?) if I even did then. As long as one is reasonably tight on the flat, you should be good.
Correctly use the correct tool to tighten them and you shouldn’t have to worry about it.
Also, if you do switch to a socket head screw, use a high quality one. You don’t want that head shearing off or you’ll be grinding off the entire pulley.

7 Likes

There’s a lotta ways to skin the cat. I chose to go simple with a socket head, less parts trying to figure out lock washers and bolts like you mention. Ultimately, I think my machine (and maybe others as well) shipped with some bunk set screws. You can see in my original post some of my grub screws were really short. Not sure if that was the problem or not, but switching to the new screws has solved all the problems and heartache I was having over that with ruined jobs.

While I agree with Neil that changing the hardware here could cause problems, I think everyone way overthinks these screws and miss the problems they cause for some. If you are applying so much force that you snap a socket screw or sheer your threads, well, you might be over-doing it.

I actually had that problem with the original grub screws, I kept re-tightening because they were backing out, trying to crank down to get it to stay, ended up rounding one out with that tiny allen wrench. Luckily I was able to put a dot of super glue on the end of the allen wrench and was able to get it out.

So, YMMV, as with any forum suggested repair, but I feel pretty confident in recommending it if you’re dealing with this - it put an end to it for me.

1 Like

A lot of good tips on here. Thanks. So I guess, to quote someone privately on this thread, “the only person you should be mad at is yourself”

I ran some tests with 4x4 MDF scraps today and carved out some test text. Leaving the board in place and cutting it again, the machine was accurate in the recut. On another test I did the same carve, then reinitialized the machine, and recarved. Same result. I also did a 3rd test with a re-zero. That worked too. So I guess by those tests I must have screwed up my zero-ing and/or placement of the pieces.

I still have some accuracy issues with X distances – if I tell the machine to cut 1" from the left zero point it cuts about 1.05" instead. So I think I just need to do some calibration.

Anyway, if I ever decide to redo this project I’ll try a more rigid fence and the dowels. And definitely use the smooth cylinder endmill for the bitzero-ing

Thanks

8 Likes

Great detective work. :slight_smile: How are you setting your zero by chance?

I put the bitzero v2 on the bottom left corner and put in the 1/8 cylinder. Then use the corner probe feature in CM

1 Like

Seems like a good standard process. Do you hold the BitZero in place?

Yes I hold it down… but I’m not entirely sure I should be doing that. Not sure if the software accounts for that tiny bit of movement.

1 Like

Blockquote

“I still have some accuracy issues with X distances – if I tell the machine to cut 1” from the left zero point it cuts about 1.05" instead. So I think I just need to do some calibration"

Blockquote

One of the things I do and asks others to do when they ask for my help, is to ensure your machine is calibrated (and not just square, plum and level). For this task, you can design a 100mm x 100mm (or a 4"x4") square cut in some scrap wood. After cutting the square, measure it to see how close to 100mm each side is. It should be just about spot on at 100m. Remember if its out even 1mm at 100m that compounds the longer the distance. Once you get the amount that is out by, there is a simple calculation you can do to determine how much to adjust the step/in in the command line. Its not in front of me, but I think the $100 and $101 are the X and Y steps. You may have something like 44 as the steps per inch and it may need to be something like 44.02 as an example only. Its quite amazing just how accurate you can get these machines with some time.

I actually like to run this test over the length of the X and Y. I may cut a square 450mm x 850mm and measure the accuracy as I routinely cut wide boards on my machine. being out 1mm per 100m would translate to 4mm on the Y and almost double that on the X.

I mostly work in standard inches, so I am trying to do the math in my head…

I lightly push mine into the material from the sides and hold it down. I usually get repeatability to withing a few thou if I repeatedly re-run the probe cycle.

2 Likes

Yep,

Calibration is a worthwhile exercise, helps you understand the machine, the limits of repeatability and whether something is off mechanically.

I’d do belt tensions (tune the note on the belts and ensure the Y belt tensions match) first and then on the calibration, it can help greatly to do two things;

First

Eliminate backlash & slack by cutting the same side of your test piece(s) at two distances. If you cut left side and right side, then front and back you’re including the machine backlash in your calibration. The calibration is per inch / mm travelled whilst the backlash is constant. If you cut left side, move over 500mm and cut left side again backlash and slack are going to be much less of an issue. All machines have some amount of backlash and slack.

Second

Maximise the measurement baseline to get your calibration data from, this reduces requirement for measurement precision but also reduces the impact of small errors such as motor minimum step size or any uneveness in the belt pitch if you have slightly dodgy belts.

I use a V bit, a 600mm long steel rule I believe is reasonably accurate and my phone camera on 10x zoom. I clamp the steel rule to the spoilboard aligned in X, jog the v bit tip down close to the ruler, move to 0 on the ruler and inspect with the phone camera, this is good to <0.1mm. Set X zero on the Shapeoko, jog to X=600, read off the actual position on the ruler. Divide 600 by the actual, multiply your steps / mm by that number.

You can do a similar thing cutting a target, but I’d suggest boring four holes in four corners for gauge pins and then measuring between the pins (note the round bore has no defining ‘edge direction’ for backlash).

HTH

4 Likes

Are you using a cylinder to set your x and y? Or are you using an endmill?

That is a really important point that @scarecrow1120 mentions above, an erroneous (variable, based on endmill flute position) measurement here would cascade into x and y not being accurate for every subsequently re-bitset run, regardless of how calibrated the machine is…

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.