EZ Probe Attachment for Rostock MAX

[Eaglezsoar]

Sorry cope, but some of us like to take our tools with us to solve certain problems at workplaces.....and time is money. So speed of calibration is key. As for my manual calibration.....I have had lathes give me more hell calibrating and zeroing. This one was easy. The first print was spot on perfect. The good instructions on zeroing were very good.

Just remember.....some of us have more than one type of machine to tend to, and need to shave our time down so that other machines can be set up for totally different jobs that have nothing to do with 3d printing.
A 3d printer salesman would agree with this totally. Demos are the stuff with which you bargain with in the tool industry. Makefaire is a perfect example, which is where I saw the Rostock Max in action, and bought the kit later on.
Lazy? No. Busy fixing other machines while this one hums away. I never considered the rostock max a toy.

+2

[cope413]

I didn't mean to suggest the probe doesn't have a purpose. I was more curious as to why some people are so anxious to get it. I guess I improperly assumed that the anxious people were those that are needing to constantly recalibrate...

[vile_fly]

Every time you move a large piece of machinery, it will distort ever so slightly, and throw the calibration off a bit when it is set down to be used again, and you have to recalibrate if you desire any degree of accuracy. You may not have to do it everytime, but the consequences of not calibrating loom above you like a guillotine, just waiting to getcha at the the exact wrong time. It would be mortifying to be demonstrating it when it decides to crash into the bed, splitting the glass.
By that reckoning, the people in motion a lot with their printers will want to calibrate more often.
I suppose I am preaching to the choir on this statement, by the way. you all probably know this already from experience.

[dsnettleton]

I haven't recalibrated my printer very many times either. If I had something like this, though, that would change. The easier you can make it for users to perform maintenance tasks, the more apt they will be to perform them.

[geneb]

The firmware is _almost_ there. Both versions of "auto-level" that are in Merlin from Johann and Repetier .90 Beta (from Roland) do NOT offer "auto-radius", which is the feature that's needed here.

ScribbleJ on #seemecnc has been working on getting this going with a branch of Marlin. It's close, but not quite there yet.

g.

[inventabuild]

geneb, thanks for the awesome news. By the way, did you find a way to incorporate the auto-level functionality into the RMAX without having to take the top off?

Also, I thought I might add to the discussion on the benefits of auto-level by pointing out that with the abundance of plastics we can print with now (and I'm sure many more in the pipeline) there is a real possibility we might want to switch out build plates to print on. This will occur because different plastics will stick and release better with different build plates. Auto level will help immensely in this regard as each build plate might have a slightly different thickness.

[geneb]

When ScribbleJ gets the system nailed down, you won't need a probe that can automatically deploy. You'll use it once for calibration purposes and you should be able to stow it until you change something on the machine that could effect the level or the radius adjustment.

g.

[626Pilot]

I didn't mean to suggest the probe doesn't have a purpose. I was more curious as to why some people are so anxious to get it. I guess I improperly assumed that the anxious people were those that are needing to constantly recalibrate...

Calibrating it myself involves mounting a depth gauge and fiddling around with endstop screws for at least 15 minutes, and then I have to figure out the height after I have the hot end reinstalled. If I switch from glass to kapton, clean the hairspray/glue off the surface, run the bed at a different temperature, change the clamps I use to hold down the glass, etc. the height will change and it might not be level with how it was before. If I mount something that will not only take the endstop screws out of the equation, but reset the bed height to where I can print at .01 or .008 without having to go through five test prints first, then that saves me time. Running a small business can be extraordinarily time consuming, so something like this is definitely the way to go: whatever time you spend on it will be paid back a hundredfold over the life of the device.

I'm currently working on a derivative of this Hall-Θ mount that will work on the Rostock MAX. It will extend and retract itself without needing to bump into anything (using two neodymium magnets held just the right distance apart from one another), and it will work with stuff you can get from Amazon and McMaster-Carr. (The actual Hall-Θ board comes from Europe but it's cheap and it ships fast.)

[inventabuild]

geneb, thanks for the information. Can you tell me what it is going to mount to so I can make sure I have the right component installed on my RMAX. The reason being I have done some modifications as I'm sure many have and I just want to make sure I'm ready to go w/ the auto-level when it's ready for prime time. I definitely need it.

[626Pilot]

BTW, what would this plug into, Z-min?

[geneb]

The probe I'm using is a modified version of PartDaddy's original design. I lengthened the part that attaches to the effector so that you can mount it to the top surface. I use LED light rings so mounting it under the effector isn't practical. The probe itself is held in place with a .25x25"x.125" neodymium magnet.

It does attach to the Z min input.

g.

[GarageBay9]

Is there any estimated eta on this, or at least the script to run it? I'm drooling just watching that demo video...

[geneb]

Unfortunately, no.

I'll let you guys know just as soon as I can get a good calibration process out of this. Right now the probing works, but the issue is that after the probe run, my bed profile is "tilted" up toward the X axis and down toward the Y. The software shouldn't be doing this and the author hasn't figured out why - it doesn't happen on his machine.

g.

[lordbinky]

Must be a gravity well. Just move the printer over 3 feet.

Really though, I hope you figure it out painlessly as possible. Curious, how significant is that tilt? And is it the bed profile it generates/stores have this tilt, or is it how the profile is processsed that you get this tilt? Sorry if I'm asking questions you haven't gotten to yet I know that can be frustrating.

[geneb]

Its 2mm high @ X=-130 and 2mm low (I'm assuming, I'm not going to screw up my build glass) at x=130.

g.

[Eaglezsoar]

Imagine the poor author's elation when at last he said Eureka, I have found it and then Gene tells him
it doesn't work on the RMAX, what a let down.

[geneb]

What's worse is that he's developing it on a Rostock MAX...

g.

[Eaglezsoar]

What's worse is that he's developing it on a Rostock MAX...

g.

OUCH!

[626Pilot]

I put together a Hall-Θ holder that fits the MAX and wired everything up. The "stop" wire is plugged into the third pin of Z-min (no other connections necessary there), +5 and ground to the power supply. This is with Repetier 0.90 alpha.

Following these instructions I got it to do a four-point calibration. I ran G32 S2 P1 several times until the x and y deltas had gone down to 0.0 - 0.01 (it gets more and more accurate every time, just like when you use a dial gauge.) Initial results are pretty good and it seems to glide over the surface just right. I'm waiting for some fire cement to cure on the nozzle so I can't do a test print, but I expect good results. It may not be as absolutely perfect as the Marlin version that tests many more points, but it's as good as doing a manual calibration with a dial gauge.

[Eaglezsoar]

I put together a Hall-Θ holder that fits the MAX and wired everything up. The "stop" wire is plugged into the third pin of Z-min (no other connections necessary there), +5 and ground to the power supply. This is with Repetier 0.90 alpha.

Following these instructions I got it to do a four-point calibration. I ran G32 S2 P1 several times until the x and y deltas had gone down to 0.0 - 0.01 (it gets more and more accurate every time, just like when you use a dial gauge.) Initial results are pretty good and it seems to glide over the surface just right. I'm waiting for some fire cement to cure on the nozzle so I can't do a test print, but I expect good results. It may not be as absolutely perfect as the Marlin version that tests many more points, but it's as good as doing a manual calibration with a dial gauge.

Man, that is great news!!!!
When you get all the bugs out, would you please do a write up or better yet, perhaps you can come out with a kit and make a few bucks.
GREAT NEWS!!!

[626Pilot]

I'll just give it away on Thingiverse. The only thing you need is a 3D printed part, two neodymium magnets from Amazon, and one smooth rod and a wood screw from McMaster-Carr.

Oh yeah, and the Hall-Θ board. That's the only thing you can't get in the States, but they ship DHL so it's not a long wait.

[Glacian22]

626Pilot, you are a gentleman and a scholar.

[626Pilot]

Its 2mm high @ X=-130 and 2mm low (I'm assuming, I'm not going to screw up my build glass) at x=130.

I'm running into a similar issue. The deflection is about the same. With software leveling turned on, it will start to dig into the Kapton at around x=50, not by much but it will leave a tiny line behind itself.

This is a weird problem. It happens to you and I, but not to the guy who is developing the software. That tells me there is some physical variation between machines that throws it off, or it's a software issue, or both.

Hardware -

• Tower alignment: My towers were aligned with these clamps a month ago and the printer hasn't been moved. I don't think they are out of whack. I don't know about you and the other guy, but his are probably no better aligned than yours.
• Differing endstop screw lengths: I tried setting them all to the same height with some calipers. Made no difference.
• Differing arm lengths: I have Trick Laser CF arms, assembled by Brian himself. I don't know if they are all the same length. With depth gauge calibration to +- 1 thou at all three towers, and PRINTER_RADIUS or whatever it's called calibrated so the center is within +- 2 thou of one tower (it WILL NOT let me calibrate that closely to all three), I still see the nozzle either lifting or dipping as it gets close to the edge in certain places. Not enough to dig into the surface, but enough to stall the extruder.
• Carriages and belts: They are all tightened down. There is no wobble in any direction, nothing is loose.
• Small errors in how the T-slot notches are cut into the top/bottom plates: I can believe this. Even with the clamps tightened down, pulling all three towers toward the center with some force, I still saw places where the towers would not come in contact with the plates. Also, my towers are not equidistant. There is a slight variation, maybe 1/16" to 1/32". Others have noticed the same thing.
• Different RAMBo boards: I have 1.0. I don't know what you have but I wouldn't be surprised if it's the same. Developer might have 1.1 or above, in which case he gets 16x microstepping compared to my 8.

Software -
Why does it work for him, but not us? I'm using Repetier, dev branch (0.9), pulled three nights ago.
Possibilities: Either it works for him and the problem is triggered by hardware differences, or it doesn't, but he thinks it does because he's not trying to move the print head far enough off center.

Next piece of information I'd want is whether he's tried pushing the print nozzle all the way towards all three towers and seeing whether it will make it without deviating. This is my #1 (unpaid) project right now, so if anyone needs help with anything in particular I'm available. I sent a message to ScribbleJ on IRC.

BTW, if anyone's curious, I've attached an image of the mount. It's the same idea as the mount on Thingiverse, except that one won't fit a Rostock MAX and requires a metal rod in a diameter I didn't find, and neodymium magnets in a size I couldn't find on Amazon. (Rule of thumb: if it isn't on Amazon or McMaster, I'm not using it unless I have to.) This one will require two Nd magnets and a 3"x1/8" metal rod. The first magnet goes into the square opening near the top (it's "fluted" so it's easy to push in, and then it will force-fit.) The second magnet goes into the long slot below, and the metal rod sticks to it via magentic force. The Hall-Θ board is pressed into place and held by friction. As the probe descends, the metal rod runs into the build surface and starts pushing the magnet upward. Once it's close enough to the Hall sensor, Z-min will go HIGH and it will back up to measure the next position. If the metal rod is pushed in a little bit further, it will immediately snap up to the top of its slot due to the other magnet sitting under the sensor board. At that point it's stowed and will not project below the print nozzle.

Deployment will be via a "finger" that will be mounted to the bottom plate (the big round thing under the Onyx), replacing one of the white spacers under the Onyx. G-code to make it all work can be embedded directly into the controller via Configuration.h. I had initially thought to make deployment happen by rapidly moving the effector down and then up, but it's too tricky to do that without risking an accidental deployment.

[geneb]

Blue Max has been my test machine for this and like yours, uses Brian's hand-built Trick Laser arms.

Note that the deviation I see _only_ happens after I do the calibration and while under ScribbleJ's branch of Marlin. Under Repetier .80, it works without issue.

g.

[626Pilot]

I'm getting the same issue but it isn't with his branch. I pulled Repetier with --branch=development.

There's a post in another thread about altering the angle of the three towers to fix the problem I have. I'll take a look at it.