Welcome to the SeeMeCNC Community

Hall-O mount is up on Thingiverse: http://www.thingiverse.com/thing:161753

I haven't got the deployment finger designed yet, but if anyone has a Hall-O board and wants to try it out, this will work. Includes wiring and calibration instructions.

626Pilot wrote:Hall-O mount is up on Thingiverse: http://www.thingiverse.com/thing:161753

I haven't got the deployment finger designed yet, but if anyone has a Hall-O board and wants to try it out, this will work. Includes wiring and calibration instructions.

I'm buying the parts now. Reading the instructions and I have a question. Did you really mean to say print with a layer height of 0.01mm? Or did you mean 0.10mm?

daftscience wrote: I'm buying the parts now. Reading the instructions and I have a question. Did you really mean to say print with a layer height of 0.01mm? Or did you mean 0.10mm?

Dang place value. Fixed.

I tried to get the Marlin firmware running, since they already have a "working" autolevel for the Rostock. I could not get it to compile. The example Configuration.h they supply for deltas is outdated and doesn't have any of the autolevel stuff in it. I tried copying and pasting but it still didn't like it. I'm giving up on that for now.

I have opened a ticket here: https://github.com/repetier/Repetier-Fi ... issues/167

Anyone who has feedback, please post there so he gets the info he needs. I predict that this will never work until the Repetier dev fixes it.

The auto level (of any stripe in any firmware so far) for deltas requires you have a perfectly calibrated printer. If your printer is perfectly calibrated, there's no need for auto level.

g.

Based on his comments, he levels the endstops only on delta printers and levels the bed only on Cartesian printers. He is thinking the answer is to do both. He solicited advice about how to find the true center. I gave some, but I haven't done anything with trig for ages so I'm not sure my theory is good enough. If someone who's better than me at math and kinematics can contribute some advice, it would probably go far.

PartDaddy wrote:Here's an experimental magnetically detachable probe for the Rostock MAX. DXF files also on our SeeMeCNC GITHUB.

Probe Detatchable.jpg

Probe Detatchable2.jpg

BOM:
1-3/8 dia x 3/16" thick neo magnet
1-6-32 x 3/4"L ferrous steel flat head machine screw
1-6-32 x 1"L flat head machine screw
2-6-32 stainless nylon lock nut
1-1/4-20 x 1/2"L button head
1-10-32 x 3/16"L ferrous steel set screw (grub screw)
1-t-nut for t-slot rail
1-limit switch (micro switch)
1 each of the three laser cut wood parts 1/4" thick


The 6-32 x 3/4 ferrous flat head screw is used to attach the "Probe Guide" to the platform of the R-MAX. It also attracts the magnet on the "EZProbe". Next, the 10-32 x 3/16"L ferrous steel set screw (grub screw) is inserted into the "Probe Mount" and attracts the neo magnet when you return the "EZ Probe". The hooked portion of the "EZProbe" catches the 6-32 x 1" flat head. The bevel on the 6-32 x 1 flat head also helps guide the probe when it is returned.

The "Probe Mount" is attached to the Z axis (rear vertical rail). Adjust the tightness of the 6-32 x 3/4"L ferrous steel flat head to adjust how well the "EZProbe" fits when magnetically held in place. I programmed the probe to pick up, touch the indicator, and return the probe many times. Using an indicator, I measured a repeatability of .002".

70721 EZProbe.DXF

70722 Probe Mount.DXF

70723 Probe Guide.DXF

Are we any closer to implementing this?
I am beginning to see the do it yourself deltas using this or something very similar to it under Marlin.
Why is it taking so long to implement for the Rostock MAX? SeeMeCNC started this thread, any responses from the staff?

geneb wrote:The auto level (of any stripe in any firmware so far) for deltas requires you have a perfectly calibrated printer. If your printer is perfectly calibrated, there's no need for auto level.

g.

I can agree yet I can disagree. It may require a perfectly calibrated printer to begin with but once the level is reached the auto-leveling feature should keep the printer within tolerence
without the user's intervention for quite some period of time. We know that in the future ALL delta printers will incorporate an auto-level feature, in fact, I believe that it is going to be
a major selling point for the kit or assembled manufactured products. It is being done somewhat kicking and screaming but it appears to be mainly Marlin based. SeeMeCNC even has
the BOM posted to produce the interface device. The problem, no software as of yet. To say that the auto-leveling feature is not needed brings us back to the times when they were
arguing if springs and shocks were needed on cars. They run fine without them but would you want to ride in one? To stay ahead, remain competitive and just for ease of use we not
only need the feature if we don't have it we will have the largest delta printer out there without it, not a good place to be.

Eaglezsoar wrote:We know that in the future ALL delta printers will incorporate an auto-level feature, in fact, I believe that it is going to be a major selling point for the kit or assembled manufactured products.

I would think whoever has to answer support requests would be stoking this fire every day. I've done lots of painfully difficult programming just to cut down on the amount of requests I get from users. Every over-taxing hour I spend on that comes back as dozens that I don't have to spend doing something that should really be automated.

When the guy who wrote the autolevel support for Marlin was talking about that, he said that probing can be used to figure out all sorts of things, even arm length. There are many ways for a Rostock to be just ever so slightly out of alignment. Not all of this can necessarily be compensated for in software, but SOME of it can. I say, let's get the software as close to perfect on that score as possible. Every little thing the firmware can do is one less thing that THOUSANDS of people using that firmware have to worry about.

626Pilot wrote:

Eaglezsoar wrote:We know that in the future ALL delta printers will incorporate an auto-level feature, in fact, I believe that it is going to be a major selling point for the kit or assembled manufactured products.

I would think whoever has to answer support requests would be stoking this fire every day. I've done lots of painfully difficult programming just to cut down on the amount of requests I get from users. Every over-taxing hour I spend on that comes back as dozens that I don't have to spend doing something that should really be automated.

When the guy who wrote the autolevel support for Marlin was talking about that, he said that probing can be used to figure out all sorts of things, even arm length. There are many ways for a Rostock to be just ever so slightly out of alignment. Not all of this can necessarily be compensated for in software, but SOME of it can. I say, let's get the software as close to perfect on that score as possible. Every little thing the firmware can do is one less thing that THOUSANDS of people using that firmware have to worry about.

Thankyou, ^^^^^^ I couldn't agree more. Seemecnc guys will get this more feature filled and absolutely right than we ask I would imagine.

Eaglezsoar wrote:Are we any closer to implementing this?
I am beginning to see the do it yourself deltas using this or something very similar to it under Marlin.
Why is it taking so long to implement for the Rostock MAX? SeeMeCNC started this thread, any responses from the staff?

I'm aware that I'm treading on flame-war material, but this is a legitimate question/comment...

Why the pervading fear of Marlin around here? I know that there are a few that use it, but otherwise I have only formulated two possible explanations for the prominent feeling I pick up in these forums regarding firmware:
i) My machine can't possibly run on anything but Repetier!?
ii) DIE MARLIN, DIE! LONG LIVE REPETIER!

I'm being facetious about (ii), of course, but my larger point remains.

We all have printers and access to design software. So, if you want a probe, follow these easy steps:
1) design and print a mount
2) assemble the required electronics - it is nothing more than a minimum end-stop for the z-axis (Cartesian)
3) compile firmware that supports desired auto-level feature
4) test and provide feedback to developers

The reason the DIY deltas are using probes is, well, the name: Do It Yourself. You don't have to be a firmware developer to test the feature. If you were capable of assembling the RostockMAX kit and calibrating it for use without a probe, you're more than capable of installing your own.

(tip-toes out of room...)

From what I've seen Marlin has much "jerkier" moves than Repetier does and Repetier allows you to set a LOT of parameters without having to re-flash the board.

You guys need to keep in mind that SeeMeCNC isn't a software shop - they're hardware guys. (How many programmers does it take to screw in a lightbulb? NONE! That's a hardware problem!)

There's a few folks out there working on auto-level code for deltas. It's no where near as simple as it is for cartesian machines and that's why it's not done yet.

Yes, there IS an auto-level routine in Marlin that works (for some value of "works") for delta printers. However, unless your printer is very well calibrated, it's basically useless or makes things worse.

Cartesian printers benefit a LOT from auto-level code because of how difficult it is to get everything geometrically perfect when you're building a cartesian. With most (if not all) Rostock/Kossel delta printers, you get a perfectly level and square bed by default (unless you've REALLY screwed something up, or the parts are defective). This means that the "traditional" auto-level code isn't going to bring much to the table.

What delta users are really after is auto-radius. When you've got the delta radius perfectly calibrated, the nozzle is going to be at the same height above the bed for a given setting, throughout the build volume.

The problem is that auto-radius seems to be a very, very difficult problem to solve. For example, given two basically identical Rostock MAX machines, auto-radius on one machine produces a nearly perfect result with a +/- .2 "tilt" across the bed. On my machine that tilt explodes to +/- 2mm from x-110 to x110. We have no idea why and there just hasn't been the available time to get it figured out.

SeeMeCNC did their part - they provided a reference design for a nice little bed probe. (WAY overly complex, but that's just how Steve rolls. *laughs*) It's up to the community to put forth the effort to make it actually work. Auto-radius is a "nice to have" feature, not a "must have". I say this because with a small amount of effort, you can get the same result as auto-radius. Nobody is being prevented from getting the most out of their machine by not having auto-radius available.

g.

Some of my personal experiences with both are varied and dependant on version and the amount of fiddling I do with each indavidual specific version at the time.

In general, I have found Marlin more likely to simply just work right out of the box so to speak if you plug in the correct values for your diagonal rod length etc then it requires less detailed and in depth fiddling to get it just right.

However, I agree repetier is better in the movement and planner departments and seems to be on track to have a more feature rich offering.

Both appear to have very responive developers although I have heard complaints about each as well, never had any bad experiences myself.

Currently running v0.90 alpha repetier, gonna give the delta specific version of Marlin a shot, perhaps early this week. I find it easy to swap back and forth without too much trouble if I am careful with keeping my saved changes of course with each and the settings that simply work with each as well.

Thanks for your thoughts, guys.

I will concur that I think Repetier has smoother movement, but probably not due to planning, and IMO not enough so for it to be a definitive "win" over Marlin. I suspect, IIRC, it to be due to using long ints instead of floats to track position, as the motion planner is quite similar. There is much more overhead with floating-point math than integer math, especially on an 8-bit MCU.

geneb wrote:You guys need to keep in mind that SeeMeCNC isn't a software shop - they're hardware guys.

This is basically the reason I asked my question. There seems to be a great deal of desire here for SeeMeCNC to provide the full solution to bed leveling, software and all. While that isn't necessarily out of the realm of possibility, in an open-source environment, they don't make that much more money putting extensive effort into software development when their niche is selling machines that can operate with a variety of firmwares and/or electronics packages.

geneb wrote:Nobody is being prevented from getting the most out of their machine by not having auto-radius available.

Can't say it any better. What I learned first-hand, and am starting to see become pervasive, is the thought that an auto-level probe can fix everything. Nope. I made this mistake because I've used probes on CNC routers in the past to map surfaces. The Delta kinematics are not necessarily that much more complex than standard Cartesian, but enough so (and non-linear) that errors in machine geometry manifest themselves in very non-intuitive manners. As I suspected, my described tower-adjustment method of leveling my machine has only worked for me as far as I know, not because I know magic, but likely because I've spent waaay too much time trying to understand the Delta platform. I do have a background in automatic controls and kinematics, so maybe that has helped...

If you can do a successful 4-point calibration (7-point is definitive), but you can't produce good prints over the full bed area without a probe, it's a slim chance that you're one of the edge cases where auto-leveling of any kind will cure your ills.

geneb brings up another, related, topic:

For clarity, the Marlin G29 auto-level is more appropriately called "bed mapping," as its task is exactly that. "Auto-level" as a description is at least more accurate with its implementation than with the probe-based auto-whatever methods in other firmwares. With a properly calibrated machine, Marlin's G29 should allow you to print on any surface contour you can put under your nozzle, within sensible limitations, but it expects that the user has done the diligence to ensure a calibrated delta-radius.

As geneb states, the current implementation of "auto-level" in both Repetier 0.90alpha and a custom version of Marlin, for which he has been aiding the development, should be called "auto-radius," or "auto-calibration," as it is not actually leveling anything. I think it has become a common mistake to refer to the calibration process as "leveling," and is likely feeding the confusion surrounding Z-probes.

edward wrote: 3) compile firmware that supports desired auto-level feature
4) test and provide feedback to developers

I'd love to compile it but the example Configuration.h for delta printers doesn't have auto level support and I couldn't figure out how to make that work, even with copying and pasting what were supposed to be the settings. I also didn't see support for our LCD panels in there. My RAMBo came loaded with Marlin way back in the day because SeeMe has their own branch that's patched for our printers. But, this branch is old and has no auto-level support. Also, Marlin is weird. I remember when I was using it, I wanted to do a four-point calibration, so I was jogging the platform down to the surface in small increments. It would skip some of the 0.1mm steps. Don't get me wrong, authoring something like Marlin and giving it away for free and supporting it is a huge effort and whoever is behind it is a saint. However, that behavior makes me wonder how accurate it is, if it doesn't think a tenth of a millimeter is a large enough distance that it should bother moving the motors.

For the auto-level feature in Marlin, you need Johann's deltabot branch.

Where to connect in the Rambo Z probe?

http://www.thingiverse.com/thing:169980

Since i'm not using stand offs I modified 626Pilot's probe to lower it a bit. Should work as-is if you don't have magnet arms. If you do, then you need another stationary magnet to retract the probe.

mrblade wrote:Where to connect in the Rambo Z probe?

Here ya go:
This is from http://www.thingiverse.com/thing:161753/#instructions

626Pilot wrote: You will need three wires going to the board: +5V, GND, and STOP. (The board has a fourth pin, SIGNAL, but we don't use it.) +5 and GND can come from your power supply. STOP plugs into the third pin of the Z-Min plug on your RAMBo. (The other two pins on that header are unused.)

daftscience wrote:http://www.thingiverse.com/thing:169980

Since i'm not using stand offs I modified 626Pilot's probe to lower it a bit. Should work as-is if you don't have magnet arms. If you do, then you need another stationary magnet to retract the probe.

Cool. Now I don't have to do another version for the stock hot end

I am trying to get my head around why the need for a probe? If you put a few drops of water on the glass and squeegee a piece of foil down it will stay down flat.
Put an alligator clip on this and the glass. The other end of the circuit to the nozzle. What am I missing? I may need to make up a buzz box with an LED and use this to fine tune leveling after the paper round of leveling.

That will certainly work and I use a similar touch probe on my CNC mills. But, putting a piece of foil down takes effort and the 3D printing community is a lazy lot! (j/k). The idea is to be able to recalibrate frequently, like maybe everyday or more often. Putting foil down would be time consuming. But, touch probes like you describe are very precise and reproducible.

One of the differences here is that once a mill is dialed in, you only ned to touch off the stock to establish your Z datum. The goal of these printer probes is to calibrate planarity of the head movement to the print surface and to establish Z (but mostly the former).

Make sense?

There's literally no point in using a buzz box on a 3D printer. As long as the travel is perfectly parallel across the bed, you're fine. You're trying to apply mill techniques to a decidedly non-mill thing.

g.

See, I knew you guys would have the answers. Let me ask this then; is it best to have a strong stable base (table), set it up and leave it alone or does this not make much of a difference? This is assuming all the screws got tightened up nicely and things were straight the first time.

I do think that having these Rostocks mounted to a strong and heavy work bench/table is a good idea. They can really get flying!