Checking Flex in DIY CNC & Calibrating / Adjusting Steps

In this video I’m going to check the flexibility of my spindle, and I’m also going to do any final adjustments to the steps of the XY and Z axis. And just to mention – I’ve already squared the x axis to the y axis as best I can, trammed the spindle – which made a lot easier using the pivot plate method, surfaced the wasteboard and finally checked the tram of the spindle by measuring the bits squareness against the wasteboard.  Anyway back to the main content of the video.

I really need a pendent. 0.02, so very good. I’ll start by pushing the top of the machine. So it doesn’t want to tilt that way – its mainly the nod and it’s 0.1 of a mm. That’s not bad considering what I was getting from the previous machines. I am quite close to one side of the x axis, so I think if I was in the centre there may be a little bit more twist but I’ll check that later.

It’s the backlash in the machine. So I think maybe the coupler or the locking collar has got loose on this side and that’s creating that backlash there. It looks like it’s this section here – loosen up a bit. It could be something slipping on the other end. It could be the motor coupler changing it’s tension over time, and it could be the grub screw on the locking collar just slipping. It’s a lot better but  if you can get split locking collars the clamping pressure would be across the entire collar and not a single grub screw.

Ok that’s gone down to about 0.1 of mm. Unfortunately with these type of lead-screws and the locking collar it’s kind as good as your going to get it. I mean in terms of movement and the steps, it seems to be okay on this axis but the backlash in the issue.

So like I said the steps looks correct but there’s still some backlash present because of the mechanism and the nutblock and that’s 0.1mm. and I think I just have to live with that.

Ok so the backlash on this is massive. Where’s that coming from? So I start by moving the lead screw, it’s not great so it’s in this. Ok that’s a lot better. I’m not pushing the spindle along the x axis to see what the flex is.

So the thing you might not notice when checking the z, is that the weight of the spindle is going to keep it down on the top edge of the thread. So if I lift you can see that’s 1mm there. It could be any number of things. It could be again the locking collar slipping because there’s only a single tiny grub screw holding that in place. In fact it’s definitely that so I think we’ll definitely adjust that first.

There’s not much room to adjust that without taking the spindle out. So the backlash on this is 0.06mm.

I’m now going to cut out a series of shapes so I can test the steps of each of the axes. I’m going to do a couple squares which should be parallel to the x and y axes, and then I’m going to cut out two shapes where the squares are rotated at 45 degrees and then I’ll do a couple circles. And each one I’ll do an inner cut and an outer cut.

I’m just doing a probe here, you can’t see it. I’m going to go over how I used the probe feature and set up my jobs in a separate video so stay tune to that.

I ran into a few problems with missed steps. What shapes I did cut seemed more or less the right size but the movement between them was incorrect. At first I thought it might have been the motor shaft coupler slipping, but that was fine. I double checked the stepper drivers and it turned out I hadn’t set the voltage correctly – it as a little low. That in conjunction with my acceleration speed being too high – meant when the spindle was moving between shapes (larger distances than the circumference of the shapes themselves) and coming to speed without the required voltage and subsequent torque.

my stepper motors run at 3 amps so after reading the manual for my M542t driver, I changed switches 1-3 from 2.8A to 3.2A on the peak current. But there is a note that states:  Due to motor inductance, the actual current in the coil may be smaller than the dynamic current setting, particularly under high speed conditions. This made me wonder whether I could change the current from 3.2 to 4.2 amps, on the peak current because that setting had a reference current of 3.0 amp. That and because I’m using switch 4 to reduce the current while the steppers stand still, that could mean I wouldn’t ever reach he peak current of 4.2A for long enough to cause problems. Tentatively I raised the current to 3.2.

For the second job I stopped at 4mm into the material – partly to save time in case something went wrong.

Inside, inside, inside, outside, outside, outside.

And things looked better, but when I began part 2 of the job, to complete cutting through the shapes, I noticed the cuts were not lining up and had shifted along the x axis.

This is interesting but could be the result of a couple factors. It could be the acceleration still effecting the feed-rates. The shapes are small, and the spindle doesn’t reach the max feed-rate speed while following those vectors because it’s still acceleration, and each time it completes a cut and drops down it starts again. But the distance travelling between shapes and to the origin position is large enough to reach full speed and miss steps. Also I have two stepper motors on the Y axis and a single out on the X.

There is a little bit of flexibility in this. I can’t really see the nut block well enough to know if something’s wrong with that but it’s not moving and it feels nice and solid. Maybe there’s a bit of flex in the lead screw or I’m just travelling too fast so I’m going to try drop that down and see if that makes a difference.

So I’ve just changed these down from 40 to 30mm/sec, and I’m going to do one shape at a time and see what effect that has.

Right 28.74…. I decided to do a quick stress test by measuring the distance between the gantry plate and y plate after homing, and then moved the gantry repeatedly along the x axis back and forth, before re-homing and measuring that distance again. 28.72 0 seems fine.

I repeated the cuts using the same method of completing the cutting job at 4mm down, letting the machine return to the origin position, load the second part as a separate file, and cutting down to 18mm.

Ok that seems a lot better. The issue was the acceleration. So now what I’m going to do is start cutting the parts out so I can measure them… Making sure to mark a corner so I can keep track of the orientation they were cut – and to then take measurements using my caliper and engineers square, directly from the shapes.

I noticed that the squares which had edges parallel with the x and y axis seemed a lot more accurate than the ones rotated at 45 degrees. Even though these shapes are identical in the software – they are cut entirely differently. The first uses a single axis at a time. The y is stationary while the x is moving, and vice-versa to create the square. The squareness of the shape is determined by the squareness of the x axis to the y. While the rotated shapes use x and y axis at the same time to create the square shape. If the steps are not set correctly the shape will be elongated along either along the x or y axes. But the measurements I’m gathering from shape 1 are from the diagonal of the machine. Holding it up to a engineers square reveals that it was not cut square – and both shapes are still not identical.

I can’t see any gaps using the square when the piece was cut out parallel to the x and y axes. And obviously at an angle, there is a bit of gap.

Another way to illustrate this is to cut two circles – and to try fit one shape into the other. If it’s not too tight you will be able to fit the shape in at a particular orientation but it will pinch when rotated. Mine was very tight.

I’m now measuring from corner to corner of the square that was cut out at 45 degrees to the x and y axes. I got 62.20 along the y and 62.61 along the x. I compared this to the measurement on the CAD drawing and the y was pretty much bang on. I now need to work out the new steps value for the x axis so it produces the same measurement as the y axis.

So to work out the new steps, you need to take the current value from the axis in question, and times it by the distance travelled divided by the (cough) the actual distance that it cut. so 62.6. So the new steps according to this is 396.098. I put this number back into my controller firmware – via bCNC, and did another test cut to see if this made a difference.

The square should be 44mm square or 62.225 from opposite corners. I’m getting 62.20 along the x axis and  62.17 along the y and the square   is 44.14 and 44.22. That’s about as good as I can get it. And that looks square now.

That’s 59.98, and 60.05. The most important thing is that I regularly check that the nut blocks are still in tension, and that the motor couplers and locking collars don’t slip now. But it seems to be cutting really well now. I cut a circle and another square this time with the edges parallel with the x and y axis. I got 44.00 along the y axis and 44.03 along the x. So there’s still a difference in measurement when the shape is orientated differently and I think this variation is probably due to backlash in the leadscrew caused by the change in direction happening more frequently with the rotated shape. That and the x axis is ever so slightly off from the y. Now I’ve calibrated this as best I can. I’ve got no idea how to fine tune this anymore, and to be honest 0.2mm is good enough for how I want to use this machine.

I just wish I had done this before I went and cut the insert nuts mounting holes for the waste-board – which explains why the t-track doesn’t line up. Anyway that’s the end of another video, and in the next one I’ll talk about how I’ve set up my stationary tool levelling probe, how that relates to creating cutting jobs, and the settings and macros I use with my cnc software . And that leaves me with the last thing to say which is thanks again for watching and you’ll see me in the next.

oooh and I almost forgot to mention that the DXF plate files and part lists are available now to subscribers on my patreon page. So if you’re interested in selling your x carve and building this machine please check out the link in the description.

Leave a Reply