I mean, sort of. Obviously a BLTouch would be ideal, but there are some pretty safe assumptions that can be made that allow it to work just fine. Like with the SM provided print surface, there’s a thin even sheet of plastic above the steel - assuming the print surface follows the steel under the plastic is a pretty safe assumption with small variance.
True but in that instance the build surface is permanently affixed to the steel. Not sure it would be as accurate if you were putting tinfoil on the bed then removing it before you print. I would also worry alot more about consistency. You would need to be sure to alway put the same kind of tinfoil on (as other brands may trigger the sensor at a different hight based on there inductance) on the same way, and make sure it was smooth with no creases or bumps… sounds like a big pain to me.
Regarding foil on the bed, someone tried that and it was a horrendous failure - it’s impossible to lay it down smooth enough.
I donno, I’m still a fan of my solution - magnet sheet on glass, use the same build platform, use autoleveling as factory intended with less compensation required.
I agree that seems like a good solution. Much better then trying to put somthing temporary on the glass that then needs to be removed before printing.
Just found this as a possible glass probe https://www.filastruder.com/products/mini-differential-ir-height-sensor
Great find, here’s the same thing(?) on amazon for cheaper? https://www.amazon.com/Differential-Height-Printer-Leveling-compatitable/dp/B07RKTQ53F
I looked up a similar model and someone reported repeatability of +/-0.0125mm. That’s comparable to the induction sensor, awesome. And because of the 3 wire with digital output it should basically be a drop in replacement, after double checking voltage supply requirements.
I bought a couple just to have around, they are cheap and seem useful! Can put one on my delta and simplify the calibration on it!
This appears to be the original:
The output is kind of an odd duck. It appears the microcontroller senses something about the connection to pick an operating mode.
Automatic selection of a 4-level analog output for printer electronics that support it (e.g Duet), or a digital output for other printer electronics.
It doesn’t seem to be an open-source product. Apparently there are clones, and I’m guessing they’ve simply read out a firmware image and not made any changes to the code. It seems to have problems with multiple reflections off of a surface, with highly reflective surfaces, and with white printing on black backgrounds, not too surprising since it seems to work by differential amplitude.
It would be lovely if:
- This sensor were implemented with a CAN-capable microcontroller.
- Each work head had a way of wiring other devices into its existing CANbus.
- The software had an API to allow integration of such external devices to avoid the need for custom firmware to integrate them.
Unfortunately, the situation is not lovely.
From that page
Well, they were prime and should show up this weekend, will let you know what I find. I’m expecting it’ll be easy and a drop in replacement for the 3-wire inductive sensor.
I hear that auto detect some times has issues, but by putting a resister between two of the pins you can force it to always be in digital mode.
Hey @Atom - this has come up before, and suddenly i’m interested in lasering non-flat surfaces.
Just spitballing ideas here that may not be ideal, but are cheap and quick.
Use that light sensor to map the surface of the wood / whatever, get that out to create a surface (F360?) Use Lightburn?? or a custom postprocessor??? to modify the Z height as the laser moves around.
Yeah that would work, but keep in mind that when probing like that your number of probe points will be dependent on the rate of change in surface hight. Otherwise I think it will turn out poorly.
Keep in mind that at that point your effectively 3d scanning the surface then lasering onto it. And there are alot of good 3d scanning hacks out there. I feel like structured light imaging would be a better choice for speed given the relatively low accuracy required (thats what the Xbox kineckt uses and they have hacked that to use it to create meshes)
Also im not sure you would need to do any processing with the point cloud. Couldn’t you just write up a script to throw that into the marlin bed leveling points and let marlin take care of the rest?
…Good point with the bed levelling. Just change the bed levelling grid temporarily to a 50x50 grid or something and load the mesh values in. It’s off by default for CNC and Laser, but it can be turned back on.
A depth camera is a good idea, actually new smartphones capture depth information - might be as simple as taking a picture and extracting the surface from the image. The Pixel 4 and new iPhones do that.
Being able to CNC a surface similar to this: and then laser engrave a pattern onto that surface would be so cool. The surface is known, too, which makes the mesh settings even easier.
Edit: didn’t realize the kinect is so cheap. Awesome!
Could work, I point out the Xbox kinect eye because it’s cheap (you can get them for like $30) all the programing to get a point cloud from the camera is already done. And they were designed to get rough 3d surfaces so it works fairly well. But if you can figure out how to do it with a phone, that would be awsome.
Either way, that sort of point cloud collection would be my first go to for lasering or milling on a non flat surface. Probing is just sooooo slow to do with good results unless the surface is supposed to be relatively flat to begin with, and it provides far more accuracy then is needed for the process.
If you are milling the surface then you shouldn’t need to capture it at all. Just use the mesh from the model you milled to get your point cloud. Should be close enough.
If you get somthing working let me know. Sounds like it would be cool!
Yes get the original one for the Xbox 360… they say it works better with the software anyway.
Edit: here is the all3dp article about it:
If you’re going to try this, remember to increase the laser power the larger the slope of the patch the beam wait is on. The correction is 1/cosθ, where θ is the angle the surface normal makes with vertical. To be more accurate, that’s the first-order correction; you probably won’t need higher ones.
The other caveat about doing this on a Snapmaker is that you’ll need to accurately register the laser head to the CNC surface as cut. Even if you leave the work piece clamped in the same place on the table, it only solves part of the problem (rotation). There’s very poor repeatability when you mount a work head on the gantry, so you’ve got the problem of finding work origin. This is less of a problem in the Y-axis and rather more along the X-axis. My suggestion would be to engrave some registration targets outside the final piece and cut them off when you’re done.
Those are great pieces of info, thank you.
I think that will push me towards writing a post processor that will handle both Z and also laser corrections. It won’t be trivial, but I think the math is straightforward.
Initial findings on the IR height sensor.
3.3-5V tolerant, not compatible with the 24V source, requires a voltage regulator.
Expected digital output expected to be inverted 3.3V signaling. IR sensor output is non-inverting. This actually works fine since a level shifting circuit would’ve been required anyways.
I tried to do a quick demo to prove it works in the SM but I don’t have the required JST2.0 connector to do it properly. Works on the bench though.
Very interested in your findings. I have glass with zyflex on it and want a sensor for the laser and CNC for bed leveling. Been pushing support for the BlTouch or similar so one sensor would work for all surfaces and modules. Make bracket that stays on x-axis and a quick connect plug. Keep up the work !!!
no habeis probado placa de fibra de vidrio negra,a mi me funciona y es mas resistente,y lo de calibrar es usar la magnetica que trae,y calibrar y cuando lo haga y haya que subir o bajar el hosten,sacar la placa ,y poer la otra,entonces subir y bajar como la placa original,ydarle a save y ya esta,hola buenas noches
se pude cortar a la medida la cama,osea funciona tanto con abs o pla