So since the auto-leveling on our new Delta Go printer uses the nozzle itself as the probe to make contact with the aluminum bed, there is always the chance that the plastic at the tip will be in the way. We’re probably going to do auto-leveling before shipping the printer. It’s a set-and-forget feature unless you change nozzles. I don’t think shipping will throw off the auto-leveling because our printer is all steel/aluminum and we use thread-rolling screws with loctite in critical places so they don’t come loose during vibration. (but of course we’ll test our theory before production)
But anyway, in case the user needs to re-run the auto-leveling for some reason and the nozzle is covered in plastic and can’t make contact with the bed… I thought of making a secondary “backup” probe by threading a hole in the heatsink itself and using a screw as the contact. All you need to do is tighten the screw so it’s the same length as the nozzle before re-leveling and it can always be removed while printing. What are your thoughts on this backup method? Hoping to get some feedback.
Wouldn’t doing the calibration with a hot nozzle solve the problem? (which could be the way to go anyway to cancel any deviation from thermal expansion)…
I’ve noticed, no matter how straight we make printers, the end effector always rocks back and forth, front and back towards the edges of the machine. For the longest time I thought that maybe we were assembling the things wrong - but I haven’t found any indication of that.
Having the probe offset from the nozzle like that, on a Delta does strange things with auto leveling.
A zero-thickness layer of plastic won’t be much of an insulator. The problem occurs if you have a bit of plastic on the end that makes the nozzle stop before making contact with the bed, so a screw at the same height as the nozzle wouldn’t help.
I would make a second script for using a screw as a probe vs. using the nozzle, and give it a Z offset of 3-5mm (whatever gives you an appropriate screw size). The platform compensation algorithm really needs to know the X/Y offset of the probe from the nozzle anyway, so it would be a bad idea to use the same code with two different probes.
This is a dumb problem but I wish homing could detect if I left something on the bed. I screwed up the other day and the crash threw off calibration. Maybe if the bed could sense force or vibration.
@Matt_Harrington I think the easiest way to add the feature.
Would be to add a protection ring around nozzle and calibration sensor.
In its easiest form it would be two little rings, one with little smaller diameter so it fits inside the other. The outer ring must be little flexible mountet so if it’s crash inside a forgotten object it will be moved against the inner ring. Attach a wire to each ring and you have a wonderful crash sensor ;-).
How to add it to your printer is another question. But I think the dirtiest option would be attach the wires to the end stops ?
You can also use one ring and 3 switches each in 120 degrees around the ring.
@ThantiK
If none of the ball joints is binding, then that rocking of the end effector is likely due to at least one pair of pivot centers on the effector having different spacing from the spacing up on the vertical carriage.
The only other things I can think of is if the rods aren’t exactly the same length, or your carriages have some flex/rotation.
@Mark_Fuller , the DeltaMaker is 100% precision machined aluminum from the same people who produce makerslide. Arms are within a quarter of a mm of each other. The balls on the arms do reach their limits near the outside edge, but this happens even near the center to some extent.
How are you sensing that the tip has made contact with the bed?
Just thinking out loud … If you used a conductivity sensor - metal tip touching the aluminium bed completing the circuit, then the plastic bits hanging below the tip will be ignored… If the tip was hot during the autolevel it would simply cut through any excess plastic to make the circuit.
@ThantiK
Thanks for the info on that. How much tilt are you seeing?
Kinda hard to tell from the images, but what is the spacing betwee rod pairs?
If there’s any play left in the ball joints, you can pre-load them with springs like the ABB flexpicker robots.
@Mark_Fuller , the problem is that the tilt we’re seeing is almost entirely invisible to the naked eye. We’re talking a 1mm difference in height on either side of a 90mm end effector as it travels across a 200mm wide circle.
It’s enough to make any kind of offset probe (like we have) register higher on one side, and lower on the other.
We’ve gotten around it with an ugly hack, but I don’t trust delta geometry at all.
Wouldn’t it be easier to machine a cap that goes over the nozzle? Then you put it on, and use a pre-determined offset. More fun way is force gauge sensors.
you also need to think about residue on the bed, like whatever crap they put on to get prints to stick.
@Stephanie_A this is exactly why I want to use strain sensors as a hot end probe. Requires some more circuitry, but I think it could add some advantages.
