Thread: Auto-Levelling

Sometimes the simple option can be the best, could you affix a thin sheet of MDF to the bed and use the machine to level it, skimming as small amount as needed to get it flat?

My machine was poorly built and I did that with a 10mm MDF sheet, it stayed there for about 7 years without issue, although I admit I wasn't moving it around a lot.

Russell.

Water Level or a Dumpy Level. A water level is about as simple and quick as it gets, you just need to make sure there are no bubbles in the water.

Ideally you want a very rigid metal frame properly bolted to a solid floor with appropriate shims under the feet to physically level the machine. Then you have an MDF spoilboard (we used 16mm) which you have the machine skim in 0.5mm deep passes until it has taken all of the skin off the top surface. That will leave a surface about as flat as your machine is capable of achieving. You skim the sheet to set up each time the machine is moved and setup again. For the commercial machine that I used to drive, the skim bit was a 75mm diamond tipped 'claw' bit that could happily skim a 2400 x 1800 surface in about 2 minutes go to whoa, then reset the surface height variable in the machine to the finished height. (All tool heights were then referred to that height)

I accept that your machine lacks the whistles and bells of the commercial ones, and has to be mobile as well, but having a spoilboard and surfacing it before each session is the way to go. In terms of doing the job, decide upon a bit that the machine can happily handle, then set up a path that runs end to end, moves across about 95% of the bit coverage, and moves back to the starting end, steps across again and moves to the opposite end again. This is repeated as often as needed to cover the surface.

Its worth noting that for many projects you don't actually need to skim or flatten the entire top surface. For example, if your machine has a work area of 600x900 but your workpiece is only 180x300 there really isn't a lot of point in getting the entire surface flat. Occasionally it is useful to mount your workpiece in a jig (e.g. engraving brass plates), so you just flatten the work area you need in the jig.

With regard to autolevelling:
I've not tried it, but I think chillipeppr relies on input from a touch probe. I think the probe is connected as one of the limit switches (z min probably), and then a series of homing to z min cycles are run at various points around your work surface. The more points the more accurate the height compensation will be, but the longer it will take to do. The controller hardware should be fairly irrelevant, as long as it can support homing and at least one limit switch input. I believe grbl meets these criteria.
This might be useful, even if it is 2 years old now: https://www.youtube.com/watch?v=6WNE3E1ZZYY

If I recall the information in John's other thread, his machine is too small / slow to skim a new platen level every time its used.

I don't think a water level will deliver the accuracy needed for a CNC machine.

perhaps John's best option is to build a very solid table or trolley for the machine -- one with jack down feet -- and ensure the table returns to the same position each time the machine is used. Then carefully level the corners of the table each time. I think there's an off the shelf laser level with the accuracy required to get each corner of the support table level to within less than ±0.05 mm (less than 2 thou).

Ian is quite correct, we have a very poor situation and we can do very little about it.

We painted around the legs when we last did a platen surface, and we always try to position the machine in the same location, with varying degrees of success with varying amounts of dirt under the legs.

A couple of our members are competent welders and we have discussed manufacture of a 3-wheeled rigid support frame.
There are plenty of carpenters levels available to ensure the machine is level enough to use.

As foobillious mentioned, Grbl does have a probing Gcode that initiates a search for zero, but what one does with the value obtained is still unknown.

Research at the moment is a bit slow, but there will be some time in the next few days to download and install test applications and so, hopefully, all will be revealed.
We now have a licenced copy of CamBam installed, so one line of investigation will be to find out if it can load a surface map for the generation of the Gcode file.

Delving into uncharted waters is such fun,
John.

John
if you can arrange it, I suggest you install three steel pins in the shed floor for the support frame to sit on.
The jack down feet of the frame should have corresponding cones to engage with the pins.
You'd probably want to use 3/4 or 25mm pins in the floor and 3/4 fine thread bolts for the feet.

I also suggest that a machinist's level is a better choice for leveling the frame.

One advantage of using the chillipeppr software is that it automatically adjusts your g-code file to compensate for the height differences.

If you can swing the space I think the dedicated cnc workstation on wheels is a better option - that way your gantry and bed stay static in relation to each other. You might be able to sell it on the basis that all the associated tools and tooling, plus dust extractor etc all move together.

Another option might be a folding shelf, similar to Alain Vaillancourt's set up:
.

Another challenging day on the router at the Men's Shed.
The attached photo shows an attempt at cutting out multiple coasters.

Lessons about levels learned:
- level over the area is about 0.2 mm,
- entire area must have proper support to avoid bowing,
- Pause/Resume in UGS does work,
- during Pause Z can be adjusted by hand,
- Z scaling is very close to 0.01 mm per 'click' as stepper forced by hand.

Other lessons learned:
- zero (0,0,0) can be reset mid-job.
- lost steps affect entire job from that point on,
- best to avoid multiple repeat jobs,
- job can be quickly changed in CamBam to recover from a crash.

We discussed the suggestions made by the forum contributors, thanks for those and they will be kept in hand for future developments.

For now, we intend to leave router as it is and investigate survey of level using a simple manual method and noting the values obtained on paper.
The job will be paused, Z adjusted by hand, and job resumed as appropriate.

First manual survey method envisaged as a small PCB target with attached leads and battery/light.
Manual control of Z at each test location, and record value when light illuminates.

Next attempt at coasters is designed to avoid level problems:
- route out the body of multiple coasters with cutting tool,
- mount corner fences on platen,
- mount each body in turn against fences and engrave shed logo and name,
- stock coasters sufficient for all shed members and visitors on open day,
- mount each blank coaster in turn and engrave member/visitor name.

Router Coasters 1 Compr.JPG

John.