Unsolved Mystery. Weird Z0 behavior around build perimeter.

[lordbinky]

Thanks, I converted from what I had for my firmware to what was packaged with yours since I had extra settings I made up.

I'm using .9° steppers, the probe is just a regular microswitch from what I had laying around so it does have extra travel after the trigger point (I tested a few and oddly the roller plunger was the most consistent). When it crashes I don't get squat on the terminal or gcode log or a dump on the board and I don't have a jtag to get information that way. Is there something else to try I overlooked?

Oh boy I can't wait to try it again

[626Pilot]

When it crashes I don't get squat on the terminal or gcode log or a dump on the board and I don't have a jtag to get information that way. Is there something else to try I overlooked?

Actually, yeah! Smoothie has a couple of ways for debugging stuff like this. They thought of everything.

First, in config, change second_usb_serial_enable to true, save, unmount the SD card, reset the printer. When it's finished coming back, connect to it with a terminal application like minicom or cutecom (or whatever comes with your OS). You will need to leave the terminal open while using the printer, because if you don't, it will hang. (It has a serial buffer that won't flush without a connected console, so if you ask for that second serial console, you HAVE to use it.)

On the serial console, type help for a list of commands. One of them should be "mem" or "memory". You can run that command periodically to see how much RAM is available. See if there are any unexpected drops in RAM, and where it is when it crashes. Let me know what you observe.

Second, Smoothie contains something called Monitor for Remote Inspection, or MRI. If the board crashes, MRI should take over. See this page for info on what to do. All the files you need are already included with the Smoothie source, so you don't have to download anything else. I'd need you to create a full crash dump, not mini, then upload it to Gist or some other file sharing place. Instead of sending it to the Smoothie community for analysis (they won't even know about my code), post the link here so I can look at it.

I haven't tried MRI with second serial enabled. That setting could conflict with MRI. I don't know. If you're not able to connect to MRI, try disabling second_usb_serial_enable to false, saving, ejecting, rebooting, and running the commands again.

[lordbinky]

It seems my config file was offending it in some way. I've been happily playing with it and so far have gotten it to be at worst .019 off after depth correction. Having fun and it's interesting to see where it goes with the adjustments it makes.

[626Pilot]

It seems my config file was offending it in some way. I've been happily playing with it and so far have gotten it to be at worst .019 off after depth correction. Having fun and it's interesting to see where it goes with the adjustments it makes.

I'm really glad to hear it was just the config file.

Can you post the console logs of the calibration? I'd like to see what it was like before and after. Did you run the heuristic calibration, depth map correction, or both?

[lordbinky]

I'll post some logs for you, I've been playing with it how the firmware converges. I'll post a log of a series of 31 OPQRS and Z followed by a G31 A, one from a clean G32, one from my settings I got to when I did my manual method, and one from my initial settings with a .1 multiplier on OPQRS (ie O.1P.1Q.1R.1S.1). I am also considering doing a run with low multipliers and high annealing iterations, and halving the binary search width and overrun divisor.

[lordbinky]

Ok, so I tried to get a dump with the offending config file. The com ports are unresponsive to even GDB, you hear the disconnection noise in windows a few moment after the board crashes. So no crash dump there.

But here are the console logs of different calibrations. One is the standard G32, G31 OPQRS , G32 A. The others have it initialized at my best measured settings and then a similar run through with different option values. I also added the config file so you can see things like that I have the

[0110-m-p]

Think I may have cracked this one (at least in my case)...entire bed within 0.001" (25 micron) at 110mm radius.

http://youtu.be/pnY6eCkasQw" onclick="window.open(this.href);return false;

I'm planning on writing out my entire calibration procedure for others to try out. One requirement to perform it though...you really need a dial indicator and tool holder.

[BenTheRighteous]

That is incredible!

[jckrieger]

Think I may have cracked this one (at least in my case)...entire bed within 0.001" (25 micron) at 110mm radius.

http://youtu.be/pnY6eCkasQw" onclick="window.open(this.href);return false;

I'm planning on writing out my entire calibration procedure for others to try out. One requirement to perform it though...you really need a dial indicator and tool holder.

I'm curious, was your tweaking just based on tower adjustment, delta radius adjustments, and tower angle adjustments? I found that I had very poor Z0 performance without changing the tower angles. Let me know if this is included in your calibration. I may print a dial indicator holder and evaluate my calibration. Of course, I need to understand a little more about how Gcode for this printer works first.

[bot]

Nice work, now let's see these printers print. Having a gauge tell you numbers is one thing, but you know where they say the proof is... print some pudding.

[0110-m-p]

Nice work guys, now let's see these printers print. Having a gauge tell you numbers is one thing, but you know where they say the proof is... print some pudding.

How's this? (accidentally printed it 10mm off center...oops)

[img]https://farm8.staticflickr.com/7558/15558781573_f5d88a6673_h.jpg[/img]

[0110-m-p]

I'm curious, was your tweaking just based on tower adjustment, delta radius adjustments, and tower angle adjustments? I found that I had very poor Z0 performance without changing the tower angles. Let me know if this is included in your calibration. I may print a dial indicator holder and evaluate my calibration. Of course, I need to understand a little more about how Gcode for this printer works first.

Expect a full calibration guide this weekend, but here's the short version that I had to do after initial z-height/printer radius/tower calibration...

To get all the gaps to be equal I had to play with the tower rotations first. Once all the towers were level to each other and all the gaps are level to each other, I changed the tower motor steps/mm until everything is equal.

The much abridged version of my iteration results look like this...

- Towers at 90, 210, 330; 80 steps/mm (Stock settings).
- - Gaps of -0.012", +0.001", -0.007".

- Towers at 89.92, 210.15, 329.85; 80 steps/mm
- - Gaps of -0.006", -0.006", -0.0055"

- Towers at 89.92, 210.15, and 329.85; 81.125 steps/mm
- - Gaps of -0.001", 0.000", 0.000"

One key though...steps/mm need to be mulitples of the configured microstepping (usually 1/16 for a Rostock Max with RAMBo).

One big thing I haven't verified yet (am printing calibration pieces as we speak). How does this steps/mm affect printed part height. I will have those results tonight.

[bot]

Nice work guys, now let's see these printers print. Having a gauge tell you numbers is one thing, but you know where they say the proof is... print some pudding.

How's this? (accidentally printed it 10mm off center...oops)

[img]https://farm8.staticflickr.com/7558/15558781573_f5d88a6673_h.jpg[/img]

That's... a step up! I wanna see a full print. Like, have you printed anything with these settings? Are dimensions accurate, circles round, etc?.. edit: oops, see now that you are printing something now. I'm very excited to hear about your results. This does look like a promising solution for easy calibration.

[0110-m-p]

Currently printing this...

[img]https://farm8.staticflickr.com/7565/15994991209_d465545b6a_h.jpg[/img]

[nitewatchman]

How did you measure the tower rotations? Science, trial and error, or magic?

Maybe a bit of all three.

[0110-m-p]

How did you measure the tower rotations? Science, trial and error, or magic?

Maybe a bit of all three.

Mostly trial and error, but after a while you get a feel for what changes do what to the nozzle/bed gap. Basically increasing the gap between two towers will decrease the gap between the nozzle and the bed.

[bot]

Currently printing this...

[img]https://farm8.staticflickr.com/7565/15994991209_d465545b6a_h.jpg[/img]

I've gotta say, that looks pretty impressive. Are those trick laser arms or custom jobs? The length seems to be greater than stock.

[0110-m-p]

I've gotta say, that looks pretty impressive. Are those trick laser arms or custom jobs? The length seems to be greater than stock.

Thanks.

They are trick laser arms.

[lordbinky]

While I have gotten my X/Y to within .05 mm of intentional size, I glossed over the Z/Height. It stopped being a non-issue when I tried to make a print and circular holes vertical holes were very squished. I'll be running through 626Pilot's firmware and seeing if I can identify any particular setting as the culprit.

[lordbinky]

Well, turned out it was ABS shrinkage hiding well because of support that led up to it.

[mlapaglia]

Hey guys,

Could something like each tower's depth into the base of the printer affect calibration?

The towers I recieved looked like they had been filed down on the sides to ease their insertion into the body of the printer. The screw head that stops the towers is hitting the tower on this rounded edge, possibly causing all the heights to not be the same? This is also mirrored on the top.

[img]http://i.imgur.com/ft1jnvq.jpg[/img]
[img]http://i.imgur.com/KjJiSih.jpg[/img]

[0110-m-p]

Hey guys,

Could something like each tower's depth into the base of the printer affect calibration?

The towers I recieved looked like they had been filed down on the sides to ease their insertion into the body of the printer. The screw head that stops the towers is hitting the tower on this rounded edge, possibly causing all the heights to not be the same? This is also mirrored on the top.

As long as the distance between base and top plate is equal for all 3 towers then it should be fine. The installed depth doesn't matter as long as these are equal. The perpendicularity between the towers and the build surface is important though.

[mlapaglia]

Hey guys,

Could something like each tower's depth into the base of the printer affect calibration?

The towers I received looked like they had been filed down on the sides to ease their insertion into the body of the printer. The screw head that stops the towers is hitting the tower on this rounded edge, possibly causing all the heights to not be the same? This is also mirrored on the top.

As long as the distance between base and top plate is equal for all 3 towers then it should be fine. The installed depth doesn't matter as long as these are equal. The perpendicularity between the towers and the build surface is important though.

I am betting these aren't the same height I will check tonight.. I've also found one of my towers isn't completely perpendicular, but the difference is so small I'm not sure how to fix it.

[0110-m-p]

I've also found one of my towers isn't completely perpendicular, but the difference is so small I'm not sure how to fix it.

I don't really think perfect is required...but the closer the better. Using a digital angle gauge my towers are something like 89.95, 89.85, and 89.8 degrees.

[bot]

0110-m-p, have you printed anything tall? I'd like to see if there are any weird artifacts from the steps/mm adjustment.