P003A HE280 [Solved?] Auto Leveling Issue/Bed Leveling Issue/Calibration

[Bones056]

When that big print finishes tomorrow at around 7PM. I will separate the End Stop Wiring form the Stepper Motor Wiring. I have them tucked together in a bundle to clean the top up. I will also run 3 cal's like mhackney did with the Ferrite on, Ferrite off, and then I will do one after I separate out the wiring on top. Should happen sometime on Thursday.

[mhackney]

Also do the calibrations (3 sets) before and after you separate the endstop and stepper wiring. If there is an issue, you will see it in greater variation in the offsets.

[morgandc]

I thought it was interesting that the "deep" way of mounting the bed was recommended on the other thread. That is the opposite of what I have done, I followed the directions

[Bones056]

Will do....

[geneb]

I suspect it boils down to a math issue in the controller. If someone were to port dc42's calibration page to Python as an Octoprint plugin, the issue would vanish with that calibration. It would require Kraegar's tweak to the firmware in order to be able to probe "normally" though.

g.

[miaviator]

Check my mechanical notes on the v3 build. This can also be the cause of many first layer leveling issues:
http://forum.seemecnc.com/viewtopic.php?f=112&t=11104

Here at the shop, not one of us have put ferrite on our machines. I agree that ferrite can't hurt. If miaviator was in house at SeeMeCNC, my team would say whoa, you changed too many things at once to make certain ferrite is absolutely necessary. We would try for the problem machine and change that one thing after observing the error to see the result. It's sometime painstaking to troubleshoot, but necessary. And it's very important to have good connections in the wiring of course. We've built a lot of machines and don't see a noise issue. So a wiring noise issue is likely corrected by the ferrite. It really can't hurt anything, but we're not convinced it's necessary.

Thanks for posting Steve it helps for everyone to know (vs just being told) that your team is actively working on this.

What's quoted above is the important point about our "Fix". We developed a working fix for one machine in a few hours because someone brought a V3 in for repairs. While we did use -- single change, observe, next change -- we only made those changes to a single machine until it started working. Later we made the same bulk changes to our in house V3 and other systems.

I do hope that I've stressed this is a temporary fix until the longer term testing --- Finding machines with the issue or breaking them, isolating each possible cause, testing solutions on multiple machines, repeat for different build qualities, repeat, repeat, etc -- can be done at which point an actual solution can be found.

[dc42]

Does it make any difference if instead of probing with everything cold, you heat the bed and hot end to operating temperatures, then turn the heaters off and probe?

[mhackney]

No it does not David. I've done it both ways. The issue is that the bed level is way off - and consistently. Simply adjusting the endstop offsets manually will result in a very good calibration. Of the 18 or so users I've collected info from, the out of level is very consistent - we are all seeing a high area near the X tower (thin first layer) and low are opposite that between Y and Z (thick first layer). This happens cold or hot (heated and then bed and hotend turned off).

[Bones056]

Here are the results from my testing that I performed today. I've included the hires picture in a zip file and a pdf of my results and how I did it. If you have any questions.... let me know. I will get back as soon as I can.

Jeff

[mhackney]

@Bones056: make no bones about it, very consistent results indicating that all of these combinations are just fine.

[ComJak]

So I swapped nozzles to an .35 nozzle and the problem is back.

[duvdev]

What id the status of the new firmware fix?
I have the same problem in 3 printers

[mhackney]

I have run the firmware since Friday and have calibrated 25 separate times ALL with excellent results. I've got about 20 hours of print time on it too with no problems related to the firmware. I'm sending my feedback to SeeMeCNC and I suspect they will make the release official today.

[morgandc]

I am running the dev version, didn't run enough tests to be scientific. I think it is a bit better, but still not level. I was more focused on the new PEI bed and abs. Too much changed to know if it improved or not. The 75mm test disks look better, but not perfect, so I would not expect the large ring to look much different.

[iKeyborg]

Does it require a hardware modification?
I will try "dev" firmware tonight.

[cloneit3d]

I have run the firmware since Friday and have calibrated 25 separate times ALL with excellent results. I've got about 20 hours of print time on it too with no problems related to the firmware. I'm sending my feedback to SeeMeCNC and I suspect they will make the release official today.

I too have been running the new dev firmware and I have not had one bad calibration. I also removed the ferrite cores I had put on before I loaded the new firmware so the printer is back to stock.

[Qdeathstar]

Interesting, for me the dev firmware resulted in a poorer calibration. y is too low, x and z are too high. I'm going to try and manually calibrate again... Wifi duet is calling

[briankb]

Interesting, for me the dev firmware resulted in a poorer calibration. y is too low, x and z are too high. I'm going to try and manually calibrate again... Wifi duet is calling

I tried the dev branch last night on my RMax v3 and the built in calibration from LCD didn't work, tried it 5 times each time after a on/off of the machine. Three times it was stuck on the Calculating Endstop phase and the last two it moved the arms down but then stopped. I just went back to the master/release branch and hope to see the released version soon. I just don't know enough about Delta's to help trouble shoot beyond testing the built in firmware features like calibration.

[geneb]

Try it using the g-code script and not through the front panel.

g.

[Qdeathstar]

When I tried it a few times it would keep going down on one axis and tilt the effector a bit... after a few goes it would finally do its ditty but the results where meh. I was aiming for better than what I could achieve manually.... not there yet.


I couldn't wait any longer, I broke down and thesuet is comming with an IR sensor....

I hate the idea of throwing money at the problem, but the duet has a lot of other advantages too.

[briankb]

Try it using the g-code script and not through the front panel.

g.

I do have the g-code script and even the modified version mhackney posted to turn off the HE and that works great. But the new firmware (currently in dev) is supposed, AFAIK, to have better calibration. I've also seen PDaddy saying to use the LCD Calibration instead of g-code.

and I did clear the EEPROM each time between firmware uploads.

[geneb]

The calibration in-firmware is effectively the same as running the g-code. The code that interprets the g-code for calibration has been changed, not the g-code itself.

g.

[mhackney]

Here's the low down...

As geneb said, the firmware interprets the gcode (calibration script in this case).

In theory, it should not matter if you run the gcode from your PC host or from the SD card but in practice, because USB basically sucks as a realtime network for 3D printing, running from a PC host over USB can introduce problems due to odd delays, stutters, etc. Running from a RaspberryPi is equivalent to running from a PC host but since the Pi is dedicated to the printer, there is much less likelyhood of a USB issue so you should be fine. This is how I do it.

Now the confusion with the LCD... some of the code that implements the LCD menu options has hard coded gcode. There is a "calibration" option in the LCD but it DOES NOT call the new gcodes that SeeMeCNC added to repetier for the new auto calibration. So when you run "calibrate' from the LCD in the current firmware it IS NOT the same as running the gcode from the calibration script.

What PartDaddy MEANT to say was, run the calibration script from the SD card using the LCD "print from SD" command. Not to use the calibrate command that is there. This, of course, means you have to load the calibration script on to the SD card first but once it is there, you can just use it.

Here's a detailed overview of the script I use to calibrate:

Code: Select all

M104 S0 ;turn off hot end so as not to interfere with the probing - this is a convenience in case I forget to do it!
G69 S2 ; this is the command that does the endstop Calibration
M117 ENDSTOPS CALIBRATED
G68 ; this is the command that does horizontal Radius Calibration
M117 HORIZONTAL RADIUS CALIBRATED
G30 S2 ; this is the command that does Z height calibration
M117 Z Height Calibrated
M500 ; and this saves the new values to EEPROM
G4 S2 ; this just waits for 2 seconds
M117 SAVING CALIBRATION

All of the M117 commands simply print a message to the console or LCD. Note that you can run each of these commands on their own if you do not want or need to do a complete calibration or for test purposes. For instance, I added a button to OctoPrint to set Z height. All it does is call G30 S2. You could also enter this from the command line in your host to calibrate just the Z height.

[cloneit3d]

Here's the low down...

As geneb said, the firmware interprets the gcode (calibration script in this case).

For instance, I added a button to OctoPrint to set Z height. All it does is call G30 S2. You could also enter this from the command line in your host to calibrate just the Z height.

Not to be a complete idiot here but I have asked many times about setting Z-height so that I can get a good first layer height without having to use MaterSlice in Matter Control. Some of the answers have lead me down paths but I am still a bit lost. I thought I could set it in the EEPROM level after you and someone else told me I could but but I am just not getting things right. How are you setting your Z-height to get a good first layer print? Do you put some painters tape or some other offset in the center of the bed and run the G30 S2 button you set? I must be doing something wrong because when I changed what I thought was the offset in the EEPROM my first layer print did not change at all. Maybe I got something wrong in my sequence like maybe I changed it in the EEPROM and then ran the calibration again wiping out my change? I am just stumped and a bit frustrated and embarrassed I can't get this right. I had an M-code I entered on my other printers to adjust the Z-height however that is not supported in this firmware. Thank you, in advance for your assistance.

[mhackney]

Don't feel bad, setting Z=0 is a mystery to most people. Hopefully this will help clarify.

When you HOME your printer, it uses the Z max height value in EEPROM to determine how far the nozzle needs to travel from the home location to touching the bed (Z=0). So, every time you home, you need to make sure that the EEPROM has the correct Z max height value, otherwise the nozzle will be above the bed (if Z height is smaller than it should be) or crash into the bed (if it is greater than it should be). Simple huh! Read on.

There are multiple ways to determine Z=0 and once you have that, it needs to be saved to EEPROM. The three primary ways are:

1) trial and error
2) setting manually
3) setting with autocalibration (G30 S2 with the V3 firmware from SeeMeCNC)

All of these are valid methods and I use all of them depending on circumstances.

1) trial and error - very easy. If you attempt a print and see that the first layer is TOO THIN, take an educated guess at how much too thin it is. Then edit the Z Max in EEPROM. Let's say that the layer is really squished, a good assumption might be .2mm - the actual layer height. So find Z Max and SUBTRACT .2 mm from it. Test print (my test calibration disk is perfect for these tests). Still too thin? Guess at how much and subtract that from Z max. As you gain experience your guesses will be pretty good. On the other side, too thick, you can also guess at how much and ADD that to Z max.

I use this technique throughout the day. If I print and see my first layer is getting a little thin, I do a quick manual tweak and carry on.

2) setting manually - this is using the paper test. Home first, then lower the nozzle (and you can use jog controls or enter gcode in the command line of your controller or even use the knob on the LCD. I prefer to use Gcodes myself as it is more precise) until it "snugs" a THIN sheet of paper. I like to use onion paper (or rolling paper) as it is very thin. Once it snugs, you get your current Z position (from the controller display) and use this value to change the EEPROM Z Height. I am away from home so I can't give details without sitting in front of my printer. But, lets say you are just touching the paper and the MatterControl or Repetier Host show "0.1 mm" for Z. Then you simply SUBTRACT 0.1mm from the current Z Height in EEPROM. If the host shows something like "-.05m" then you SUBTRACT that negative number (minus a minus is add).

3) autocalibration - this one confuses people as they are expecting a silver bullet and it isn't quite that simple. Let me explain the probing process and you will understand why...

I'll keep it to the HE280 probe, all are different. When you probe, the nozzle strikes the bed. In a perfect world there is no flex in the system and that strike will actually be read as Z=0. But in the real world, the bed flexes, the belt stretches, etc. So, the firmware has a fudge factor called Z Probe Height. It is also in EEPROM and by default, SeeMeCNC set it to -0.2mm. This should tell you that the system is flexing .2mm on a probe. But guess what, every machine is different. Your belts might be looser than mine, I might have PEI on my bed that cushions (flexes) a little more, etc. When you probe, the G30 S2 command homes and then measures the distance to the probe "strike". But then it ADDS that Z Probe Height value stored in EEPROM to get the final Z Max Height. If your Z Probe Height is off, you can autoprobe until the cows come home and never get a good Z=0. My post this weekend describes how to determine your Z Probe Height and set it in EEPROM.