[RogerClark]

Hi,

I bought a Chinese made CNC 3020 last week and have been running some tests, and I have noticed that I have a calibration error in X

I noticed when drilling a set of holes for a PCB (printed circuit board), as the connector that is supposed to fit into the holes doesn't quite fit.

I'd not calibrated the machine myself, apart from putting in the 400 "steps per" as indicated in the single sheet of paper that came with the machine.

So I use direct input of GCode to G00 X00 Y00, then drilled a 1mm hole, then moved G00 C150 and drilled another hole.
Then I used my digital micrometer to measure the distance between the holes, and to within about 0.1mm I think the centres of the holes are 150.56 mm apart, not 150mm.

I also measured the Y axis, but it didn't seem to be quite as far out. In fact over 100mm it seem to be spot on.


It seems odd that I'd need to adjust the X axis to some strange number, i.e around 399 or 398 to fix this error, but I can't quite see how this could be caused by anything other than perhaps the lead screw not being exactly the correct pitch.

If it was dropping steps, I thought the number would be lower now higher, i.e its going further than it should in X.


I know these machines are cheap, and perhaps its just a build quality issue, or I'm trying to use it beyond its accuracy capability, because of hysteresis etc in the lead screw.

But it would be good to get some feedback about whether I'm doing something wrong in my calibration method etc



BTW.
I'm new to CNC milling, but not new to CNC in general, as I also have a 3D printer (MendleMax) which I built from a kit and have calibrated so that is quite accurate.


Thanks in advance

Roger

Edit. I'm using a demo copy of Mach3 that came with the machine. Its an old copy but seems fine. My PC is running XP and is relative fast (1.3ghz)

Edit 2.

I have taken another look at the Y calibration, and I've realised that its not very good either.

So it looks like rather than 400 steps per, it needs to be about 398 steps per in both X and Y.

I don't know about Z, As there is much less travel in Z its going to be hard to get an accurate calibration.

[w7bcw]

I have a CNC 3020 that i had problems with as your seeing now. It turns out the driver electronics for those Chinese machines are garbage.
What your seeing are missing steps and what i did at first was what was mentioned in this forum on adding bypass caps etc but that did not
help as it did for others.
The ultimate repair was to get rid of the garbage controller and replace it with a Gecko G540. After i did that the machine works perfectly no missing steps
and the holes are aligned with the pads perfectly. For motor tuning I got 12,700 and I am using inches not mm's.

[RogerClark]

Thanks for the reply.

I've had a look inside the controller unit (several times), and it has one board with 3 sets of stepper drivers (Toshiba TB6560), and a load of opto isolators.

Its got a fairly substantial toroidal transformer as well, but practically no smoothing. Well there are some caps on the stepper driver board, but probably not enough to do very much.


With you machine was it under running or over running? Mine was overrunning by about 0.3% in both X and Y.

I changed the calibration (steps per) to correct this, and it seems to have improved things a bit, but I need to do some more repetition tests to see if the fix is repeatable, or whether it just seemed to fix it for the one time I re-measured the distances.

When you say its missing steps, I would have thought that it would under run not over run, which is why I hadn't considered that missing steps would be a possibility.

Apart from replacing the stepper driver board, I also have a spare Arduino Mega 2560 + RAMPS board, which has stepper drivers on it, but I'm not sure they are rated to handle the motors on the CNC as they are larger than on the 3D printer.

I also have on order a chinese made 4 axis controller board, which replaces Mach3, in that it creates the signals that Mach3 does, but the board goes inside the stepper driver / psu box. I have now found out that these chinese made controller boards are based on an old open source version of someone's control board, and the original author is not happy with the chinese manufacturing his designs, - but this is a perennial problem as soon as you make any design public / open source.


It would be good to know precisely why the stepper driver board in my unit would skip steps, e.g. insufficient smoothing, cheap optos, etc, as the main Toshiba motor drivers look genuine and overall the PCB and the PSU as a whole look relatively well made (I've seen a lot worse)

[w7bcw]

Yes, the driver chips and opto's may be genuine but it is the way the board was designed. If you look at the data sheet for the driver chips whom ever in China designed
the board did not use the suggested circuit design and left out many components that would prevent missing steps and noise (could cause extra steps) on the signal
lines can cause all sorts of step issues and not just missing steps as I experienced. I now avoid all Chinese made driver pcbs since they take short cuts in the design
that can cause issues/headaches for the buyer.

[w7bcw]

Try using search terms of "missing steps" on this sub forum and see what others have experienced.