Heuristic (AI) calibration for delta printers on Smoothie

User avatar
forrie
Printmaster!
Posts: 159
Joined: Wed Apr 16, 2014 7:15 am
Location: Crab Nebula

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by forrie »

I bought about a dozen different switches to test out...they only cost a buck each so I thought...what the heck. I haven't found any real difference between no lever, lever and lever with a little curl yet between a few different brands and odd switches I had laying around. It actually gives about the same result as FSRs. Next is to try the lever with a roller, extended lever and the plunger on a spring type. It seems repeatability goes down the drain once you heat the bed though. Fun times!

I'd be interested to see any mounting solutions on the stock SeeMeCNC hotend platform...I stuffed around a bit but couldn't come up with something rigid enough, so ended up using the Tricklaser pen holder with a simple aluminium sheet metal mount.
I'm not an alcoholic...I'm Australian!
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

gestalt73 wrote:Hey KAS,

What's interesting about that switch is that it is basically the limit switch that we use on the printers, but without the arm.

I've had my suspicions about the flex in the arm and its effect on repeatability. I may remove the arm and see if I get better repeatability with the probe.

Thanks!

I was definitely seeing the flex on the first mount I used. Basically it was just mounted on one of the effector tabs. The one above mounts the same way, but the top of the mount touches the support rim on the bottom of the effector. That worked a little better. I notice that the wires caused a lot of variation. I ended up holding those in the air while completing the calibration.

I'm thinking that a longer arm would reduce the force needed to activate the switch and not wiggle the switch as much. I do have a set of FSR's, maybe I'll try that as-well. Heating the bed caused a lot of variation as-well. Could of been the heated PEI with a stiff switch and a small foot print pressing into it.
Mac The Knife
ULTIMATE 3D JEDI
Posts: 1409
Joined: Sun May 11, 2014 6:18 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by Mac The Knife »

If I didn't have all three layer fans on, I would consider using one of the 3/8" dia. standoffs to clip a switch bracket to the effector.
R-Max V2
Eris
Folger Tech FT-5 R2
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by 626Pilot »

I've been using a cheap microswitch from Amazon and getting about the same repeatability I did with my Hall-O probe.
User avatar
gestalt73
Printmaster!
Posts: 345
Joined: Wed Aug 07, 2013 11:10 am
Location: Anaheim, CA

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by gestalt73 »

Hey 626Pilot,

It looks like we still have some type of memory issue with GLCD enabled.
With the 6/12 build I pulled I was unable to complete the SA step with the GLCD enabled on either of my printers.
I thought that maybe in was something introduced since 6/9, but when I tried that build I saw the same lockup/reset.

Once I disabled the GLCD, I was able to complete all calibration steps the first time no problem.

Oh, here's the calibration results from one of my deltas:

Code: Select all

[PD]                                   -0.010
[PD]
[PD] [  --  ]    -0.010     -0.010     -0.005      0.000      0.005    [  --  ]
[PD]
[PD] [  --  ]     0.000     -0.005      0.005      0.005      0.000    [  --  ]
[PD]
[PD]   0.005      0.000      0.005      0.000     -0.005      0.000      0.020
[PD]
[PD] [  --  ]    -0.015      0.000      0.000      0.005      0.005    [  --  ]
[PD]
[PD] [  --  ]     0.010      0.000      0.015      0.020      0.020    [  --  ]
[PD]
[PD]                                    0.020
[PD]
[PD] Best=0.000, worst=0.020, min=-0.015, max=0.020, mu=0.002, sigma=0.007, energy=0.007
plasma
Printmaster!
Posts: 61
Joined: Tue Dec 30, 2014 2:41 am

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by plasma »

gestalt73 wrote:Hey 626Pilot,

It looks like we still have some type of memory issue with GLCD enabled.
With the 6/12 build I pulled I was unable to complete the SA step with the GLCD enabled on either of my printers.
I thought that maybe in was something introduced since 6/9, but when I tried that build I saw the same lockup/reset.

Once I disabled the GLCD, I was able to complete all calibration steps the first time no problem.

Oh, here's the calibration results from one of my deltas:

Code: Select all

[PD]                                   -0.010
[PD]
[PD] [  --  ]    -0.010     -0.010     -0.005      0.000      0.005    [  --  ]
[PD]
[PD] [  --  ]     0.000     -0.005      0.005      0.005      0.000    [  --  ]
[PD]
[PD]   0.005      0.000      0.005      0.000     -0.005      0.000      0.020
[PD]
[PD] [  --  ]    -0.015      0.000      0.000      0.005      0.005    [  --  ]
[PD]
[PD] [  --  ]     0.010      0.000      0.015      0.020      0.020    [  --  ]
[PD]
[PD]                                    0.020
[PD]
[PD] Best=0.000, worst=0.020, min=-0.015, max=0.020, mu=0.002, sigma=0.007, energy=0.007

I also have to turn GLCD off in order to get through the SA. BTW amazing results on your calibration my best energy was 0.023. My glass plate has lots of high spots, so thats the best i can get.
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

I'm down to .005 but that's running cold. Soon as I heat the bed, the values jump all over without decent repeatability. I'm hoping the new switches and 24awg silicon wire help a bit.
User avatar
gestalt73
Printmaster!
Posts: 345
Joined: Wed Aug 07, 2013 11:10 am
Location: Anaheim, CA

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by gestalt73 »

You know what? I did run that complete calibration cold.

Normally my Energy is around 0.010 - 0.030 for both printers, it does seem to vary a bit, but I'm not sure if it's because of the randomness of SA solutions, or accuracy of the limit switch.

Occasionally, I'll get a repeatability score of 4-5 steps, which is awful, then when I retry it I get 1-2 steps.

I think it has something to do with the flex in the limit switch arm. I need to rework that to see if it improves repeatability or not.
User avatar
forrie
Printmaster!
Posts: 159
Joined: Wed Apr 16, 2014 7:15 am
Location: Crab Nebula

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by forrie »

KAS wrote:I'm down to .005 but that's running cold. Soon as I heat the bed, the values jump all over without decent repeatability. I'm hoping the new switches and 24awg silicon wire help a bit.
Yup..same here...I can only get a good calibration using FSRs on a heated bed so far...limit switches don't seem to like the heat. I need to try the plunger type next.
I'm not an alcoholic...I'm Australian!
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by 626Pilot »

You can run the simulated annealing cold. In fact, I recommend it. When you heat the bed, the PCB is going to swell and push against the glass, deforming it and causing it to swell as well. I have noticed differences in the ~40 micron range between 25C and 50C, and that's significant. Better to anneal cold, and then G31 A after the bed is heated up, so it can correct for how it is during an actual print.

The SA algorithm will produce slightly different results each time because it exploits random numbers to help filter some of the noise out of the system. The randomness makes it harder for any potential solution to bubble up to the top, so only those solutions that are strong enough to overcome that noise can compete. Without the random movement, the system actually does far worse!
User avatar
gestalt73
Printmaster!
Posts: 345
Joined: Wed Aug 07, 2013 11:10 am
Location: Anaheim, CA

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by gestalt73 »

Based on 626Pilot's earlier mention, I decided to recompile and give 9x9 a go...

Code: Select all

[PD]                                              -0.013
[PD]
[PD] [  --  ]   [  --  ]    -0.025     -0.038     -0.038     -0.013     -0.013    [  --  ]   [  --  ]
[PD]
[PD] [  --  ]    -0.038     -0.013     -0.025      0.000      0.000      0.000     -0.013    [  --  ]
[PD]
[PD] [  --  ]     0.000     -0.013     -0.013      0.013      0.000      0.000      0.025    [  --  ]
[PD]
[PD]   0.013     -0.013      0.013     -0.013      0.000     -0.013      0.013      0.000      0.013
[PD]
[PD] [  --  ]     0.000      0.000      0.000      0.025      0.050      0.025      0.025    [  --  ]
[PD]
[PD] [  --  ]     0.013      0.013      0.013      0.038      0.050      0.038      0.062    [  --  ]
[PD]
[PD] [  --  ]   [  --  ]     0.025      0.038      0.038      0.050      0.050    [  --  ]   [  --  ]
[PD]
[PD]                                               0.038
[PD]
[PD] Best=0.000, worst=0.062, min=-0.038, max=0.062, mu=0.005, sigma=0.019, energy=0.020
User avatar
forrie
Printmaster!
Posts: 159
Joined: Wed Apr 16, 2014 7:15 am
Location: Crab Nebula

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by forrie »

gestalt73 wrote:Based on 626Pilot's earlier mention, I decided to recompile and give 9x9 a go...
Wow thats cool. Did you have the lcd, ethernet etc enabled?
I'm not an alcoholic...I'm Australian!
User avatar
gestalt73
Printmaster!
Posts: 345
Joined: Wed Aug 07, 2013 11:10 am
Location: Anaheim, CA

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by gestalt73 »

Hey forrie,

looks like we still need to disable glcd, ethernet, webserver etc to complete the calibration steps.
I haven't tried disabling them individually, so I don't know exactly which ones cause the lockups.
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

Running into an issue with the G31 OPQRS command. After about 5 consecutive runs, the lift function of each probe gets lower and lower until it starts dragging the plunger across the bed. If I reset the smoothieboard, it starts out fine again until I notice the button dragging.

Does this happen to anyone else?



Also, is this normal?
RECEIVED: [HC] Reticulating splines...
RECEIVED: [HC] Try 0 of 50, energy=0.044 (want <= 0.010)
RECEIVED: [HC] Try 5 of 50, energy=0.070 (want <= 0.010)
RECEIVED: [HC] Try 10 of 50, energy=0.051 (want <= 0.010)
RECEIVED: [HC] Try 15 of 50, energy=0.046 (want <= 0.010)
RECEIVED: [HC] Try 20 of 50, energy=0.056 (want <= 0.010)
RECEIVED: [HC] Try 25 of 50, energy=0.048 (want <= 0.010)
RECEIVED: [HC] Annealing has stalled - aborting.
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by 626Pilot »

KAS wrote:Running into an issue with the G31 OPQRS command. After about 5 consecutive runs, the lift function of each probe gets lower and lower until it starts dragging the plunger across the bed. If I reset the smoothieboard, it starts out fine again until I notice the button dragging.

Does this happen to anyone else?
Would need to see console output to be certain, but I'm guessing it probably gets the best solution it can on pass 1 or 2 and then gets progressively worse. It's best to watch the energy level, type M500 to save the config every time it gets better... and don't save it if it gets worse. Sometimes you have to let it get worse before it gets better, but if it's crashing into the glass, it's not getting better.

The thing to remember is that the code doesn't know how to account for tower lean. I tried, but the math is more of a pain in the ass than I have patience for at this time. If someone who's good at math wants to take a look at it, I have a "trig help" thread somewhere in the Lounge. Because it can't account for tower lean, it is always going to try to adjust other things in order to get the energy down, but it's like if you have someone working for you and they don't know their entire job. They will do their best, but they can only compensate for so many un-accounted-for variables.

I would like very much to get the tower lean thing nailed down. As it is, the test is biased by the fact that it always happens at (hopefully close to) the same Z height. Correcting for tower lean would improve not just the first layer, but all of them. As you increase the elevation you print at, the tower lean is going to throw things off differently than it does down at the glass. However, I think it can be corrected for just with Z=0 probing because the carriages have to climb up and down a great deal on that first layer, especially as you get out further from the center. Should be enough to at least roughly infer the tilt vector for each tower.
Also, is this normal?
RECEIVED: [HC] Reticulating splines...
RECEIVED: [HC] Try 0 of 50, energy=0.044 (want <= 0.010)
RECEIVED: [HC] Try 5 of 50, energy=0.070 (want <= 0.010)
RECEIVED: [HC] Try 10 of 50, energy=0.051 (want <= 0.010)
RECEIVED: [HC] Try 15 of 50, energy=0.046 (want <= 0.010)
RECEIVED: [HC] Try 20 of 50, energy=0.056 (want <= 0.010)
RECEIVED: [HC] Try 25 of 50, energy=0.048 (want <= 0.010)
RECEIVED: [HC] Annealing has stalled - aborting.
Yes, it will "stall" and stop working if the energy level doesn't move enough for a long enough time. It figures it's already got as good as it will without probing the surface again.
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

I'm not having any luck getting a decent calibration. With the Rambo I never had issues with the effector lifting or dipping between towers. I calibrated with a dial indicator at three points and the center and everything was good to go. I remeasured the towers with digital angle finder and everything was the same. Nothing physically moved, just replaced the controller with the smoothie.

In my many many attempts, I thought it might be the G32 causing the initial variation as the end stops never reported back repeatable values. So, I decided to manually adjust the end-stops until they were perfect within the G32. Below are the different attempts ending with a perfect 0,0,0.

When I put on the dial indicator with the G32 showing 0's, I had a variation of .009 thousands between towers. Something just isn't working and not sure what it is. G29 runs perfect and appears repeatable with the same number of steps.

Code: Select all

RECEIVED: [PK]      Endstop offsets: {-0.930, 0.000, -0.585}
RECEIVED: [PK]      Endstop offsets: {-0.930, 0.000, -0.497}
RECEIVED: [PK]      Endstop offsets: {-1.053, 0.000, -0.263}
RECEIVED: [PK]      Endstop offsets: {-2.284, -0.100, 0.000}
RECEIVED: [PK]      Endstop offsets: {0.000, -0.567, -0.682}
RECEIVED: [PK]      Endstop offsets: {0.000, -0.213, -0.320}
RECEIVED: [PK]      Endstop offsets: {-0.243, 0.000, -0.098}
RECEIVED: [PK]      Endstop offsets: {-0.213, 0.000, -0.102}
RECEIVED: [PK]      Endstop offsets: {-0.111, 0.000, -0.030}
RECEIVED: [PK]      Endstop offsets: {-0.081, 0.000, -0.026}
RECEIVED: [PK]      Endstop offsets: {-0.051, 0.000, 0.000}
RECEIVED: [PK]      Endstop offsets: {-0.047, 0.000, -0.009}
RECEIVED: [PK]      Endstop offsets: {-0.034, 0.000, -0.004}
RECEIVED: [PK]      Endstop offsets: {-0.045, 0.000, -0.026}
RECEIVED: [PK]      Endstop offsets: {0.000, -0.058, -0.053}
RECEIVED: [PK]      Endstop offsets: {0.000, -0.037, -0.026}
RECEIVED: [PK]      Endstop offsets: {0.000, -0.015, -0.007}

RECEIVED: [ES] Endstops: Difference => 0.005 (want 0.030) 
RECEIVED: [ES] Endstops are within tolerance.
RECEIVED: [DR] Delta Radius - Depths: Center=4.705, Tower average=4.703 => Difference: 0.002 (want 0.030), within tolerance.
RECEIVED:  
RECEIVED: [PK] Current kinematic settings:
RECEIVED: [PK]           Arm length: 269.000
RECEIVED: [PK]         Delta radius: 134.007
RECEIVED: [PK]      Endstop offsets: {0.000, 0.000, 0.000}
RECEIVED: [PK] Radius offsets (ABC): {0.000, 0.000, 0.000}
RECEIVED: [PK]  Angle offsets (DEF): {0.000, 0.000, 0.000}
RECEIVED: [PK]     Virtual shimming: {0.000, 0.000, 0.000}, vector={0.000, 0.000, 1.000}, d=0.000, Disabled
RECEIVED: [PK] Depth (Z) correction: Disabled
RECEIVED:  
RECEIVED: [IC] All done! Save settings with M500.

jlmccuan
Printmaster!
Posts: 75
Joined: Mon Feb 09, 2015 11:10 am

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by jlmccuan »

I'm having two issues. I believe the first caused the carnage. Even after disabling automount, my Win8 machine still opens the sdcard every time I plug in the USB cable. I've set dsd_disable to both true and false, no difference. So I went ahead and did my G29 and my probe was pretty consistent with the high to low spread .005mm and that is my stepper resolution. Only one reading was .005 off, the rest being exactly the same. I moved on the the G32 and ran into trouble. The printer became unresponsive and had to be rebooted. This is happening way too often, sometimes within seconds of connecting. At any rate now I have one of the variables set "Determining the probe-from height [new line] Probe-from height = 50520723.000" How do I wipe it all out and start over? M502 does not do it.
jlmccuan
Printmaster!
Posts: 75
Joined: Mon Feb 09, 2015 11:10 am

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by jlmccuan »

Wow, what a long 17 hour day on this. For any of you in the same boat some day.

issue an M502 from your terminal program
reboot Smoothie
erase firmware.cur or whatever other extension it might have
copy a new copy of your firmware.bin to the sdcard
power down the Smoothie and disconnect USB cable
place the sdcard into the Smoothie if you did the copying on your computer rather than USB
plug the USB cable back in
wait for the flashing lights to stop
check the directory of the sdcard to verify that firmware.bin was changed to firmware.cur or in my case Win8 shows no extension but it is a cursor file
power back on start over on probe calibration etc.
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by 626Pilot »

KAS wrote:I'm not having any luck getting a decent calibration. With the Rambo I never had issues with the effector lifting or dipping between towers. I calibrated with a dial indicator at three points and the center and everything was good to go. I remeasured the towers with digital angle finder and everything was the same. Nothing physically moved, just replaced the controller with the smoothie.

In my many many attempts, I thought it might be the G32 causing the initial variation as the end stops never reported back repeatable values. So, I decided to manually adjust the end-stops until they were perfect within the G32. Below are the different attempts ending with a perfect 0,0,0.

When I put on the dial indicator with the G32 showing 0's, I had a variation of .009 thousands between towers. Something just isn't working and not sure what it is. G29 runs perfect and appears repeatable with the same number of steps.
Can you do these three commands, and then post the full output? I need to see what's going on.

G29
G32
G31 OPQRS (don't omit any of these letters)
User avatar
626Pilot
ULTIMATE 3D JEDI
Posts: 1720
Joined: Tue May 14, 2013 12:52 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by 626Pilot »

jlmccuan wrote:I'm having two issues. I believe the first caused the carnage. Even after disabling automount, my Win8 machine still opens the sdcard every time I plug in the USB cable. I've set dsd_disable to both true and false, no difference. So I went ahead and did my G29 and my probe was pretty consistent with the high to low spread .005mm and that is my stepper resolution. Only one reading was .005 off, the rest being exactly the same. I moved on the the G32 and ran into trouble. The printer became unresponsive and had to be rebooted. This is happening way too often, sometimes within seconds of connecting. At any rate now I have one of the variables set "Determining the probe-from height [new line] Probe-from height = 50520723.000" How do I wipe it all out and start over? M502 does not do it.
G32 wipes all the calibration variables before it runs. You can also do M503 to see what values are stored under M665, and then run M665 with all the variables set to zero (except L and R, which are arm length and delta radius).
jlmccuan
Printmaster!
Posts: 75
Joined: Mon Feb 09, 2015 11:10 am

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by jlmccuan »

When I ran G32 I got the following lines and a huge crash into the bed.
Determining the probe-from height
Probe-from height = 50520723.000

I got that fixed as in my above post. The big issue now is getting the USB to stop stalling. I can be jogging around, trying a G29 z0, G32, dry run a program, or the printer just sitting idle and it stops and is unresponsive to any commands. If I unplug the USB and plug it back in, it will continue any uncompleted commands but the host software loses its connection and the COM port is no longer available. Network cable does not connect either. I've tried auto IP and manual. No dice either way.
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

626Pilot wrote:
KAS wrote:I'm not having any luck getting a decent calibration. With the Rambo I never had issues with the effector lifting or dipping between towers. I calibrated with a dial indicator at three points and the center and everything was good to go. I remeasured the towers with digital angle finder and everything was the same. Nothing physically moved, just replaced the controller with the smoothie.

In my many many attempts, I thought it might be the G32 causing the initial variation as the end stops never reported back repeatable values. So, I decided to manually adjust the end-stops until they were perfect within the G32. Below are the different attempts ending with a perfect 0,0,0.

When I put on the dial indicator with the G32 showing 0's, I had a variation of .009 thousands between towers. Something just isn't working and not sure what it is. G29 runs perfect and appears repeatable with the same number of steps.
Can you do these three commands, and then post the full output? I need to see what's going on.

G29
G32
G31 OPQRS (don't omit any of these letters)
Right now it's lifting between the Y and Z towers, and dipping right at the X tower. I was incorrect when I said nothing changed before switching to the smoothie board. I changed the mount to hang the hotend below the effector. It's very possible that it could be tilted in the mount. I'm just not sure why the calibration values for G32 don't relate to the a dial indicator for each tower as both are mounted on the effector.




G29

Code: Select all

RECEIVED: [BH] Probe-from height = 345.780
RECEIVED: [BH] Bed height set to 350.480
RECEIVED: [PR] Test  1 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  2 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  3 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  4 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  5 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  6 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  7 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  8 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test  9 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Test 10 of 10: Measured 940 steps (4.700 mm)
RECEIVED: [PR] Stats:
RECEIVED: [PR]   range: 0 steps (0.0000 mm)
RECEIVED: [PR]      mu: 940.000 steps (4.700 mm)
RECEIVED: [PR]   sigma: 0.000 steps (0.000 mm)
RECEIVED: [PR] Repeatability: 0.0000 (add a little to be sure)
RECEIVED: [PR] This is your best score so far!
RECEIVED: [PR] This score is very good! 
G32

Code: Select all

RECEIVED: [ES] Endstops: Difference => 1.120 (want 0.030), out of tolerance by 1.090.
RECEIVED: [ES] Setting endstops to {-1.300, 0.000, -1.456}.
RECEIVED: [DR] Delta Radius - Depths: Center=5.505, Tower average=5.432 => Difference: 0.073 (want 0.030), out of tolerance by 0.043.
RECEIVED: [DR] Changing delta radius from 134.081 to 134.228
RECEIVED:  
RECEIVED: [IC] Iteration 2 (max 20)
RECEIVED: [ES] Endstops: Difference => 0.135 (want 0.030), out of tolerance by 0.105.
RECEIVED: [ES] Setting endstops to {-1.235, 0.000, -1.280}.
RECEIVED: [DR] Delta Radius - Depths: Center=4.605, Tower average=4.573 => Difference: 0.032 (want 0.030), out of tolerance by 0.002.
RECEIVED: [DR] Changing delta radius from 134.228 to 134.291
RECEIVED:  
RECEIVED: [IC] Iteration 3 (max 20)
RECEIVED: [ES] Endstops: Difference => 0.030 (want 0.030), out of tolerance by 0.000.
RECEIVED: [ES] Setting endstops to {-1.274, 0.000, -1.306}.
RECEIVED: [DR] Delta Radius - Depths: Center=4.685, Tower average=4.677 => Difference: 0.008 (want 0.030), within tolerance.
RECEIVED:  
RECEIVED: [IC] Iteration 4 (max 20)
RECEIVED: [ES] Endstops: Difference => 0.015 (want 0.030) 
RECEIVED: [ES] Endstops are within tolerance.
RECEIVED: [DR] Delta Radius - Depths: Center=4.665, Tower average=4.653 => Difference: 0.012 (want 0.030), within tolerance.
RECEIVED:  
RECEIVED: [PK] Current kinematic settings:
RECEIVED: [PK]           Arm length: 271.106
RECEIVED: [PK]         Delta radius: 134.291
RECEIVED: [PK]      Endstop offsets: {-1.274, 0.000, -1.306}
RECEIVED: [PK] Radius offsets (ABC): {0.000, 0.000, 0.000}
RECEIVED: [PK]  Angle offsets (DEF): {0.000, 0.000, 0.000}
RECEIVED: [PK]     Virtual shimming: {0.000, 0.000, 0.000}, vector={0.000, 0.000, 1.000}, d=0.000, Disabled
RECEIVED: [PK] Depth (Z) correction: Disabled
RECEIVED:  
RECEIVED: [IC] All done! Save settings with M500.
G31 OPQRS

Code: Select all

RECEIVED: [BH] Probe-from height = 346.055
RECEIVED: [BH] Bed height set to 350.410
RECEIVED: [DM] Depth to bed surface at center: 870 steps (4.350 mm)
RECEIVED:  
RECEIVED: [PD]                                    0.185                                  
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.040     -0.130     -0.075     -0.050      0.025    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.010     -0.035     -0.050     -0.060     -0.110    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]   0.130     -0.090     -0.070      0.000     -0.065     -0.085     -0.125 
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.020     -0.085     -0.030     -0.030     -0.070    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.010     -0.115     -0.050      0.015     -0.005    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.045                                  
RECEIVED: [PD]
RECEIVED: [PD] Best=0.000, worst=0.185, min=-0.130, max=0.185, mu=-0.020, sigma=0.056, energy=0.065
RECEIVED:  
RECEIVED: [HC] Generating carriage positions for a printer with this configuration.
RECEIVED:  
RECEIVED: [HC] Starting test configuration: Arm Length=271.214, Delta Radius=134.250
RECEIVED:  
RECEIVED: [HC] ***** Simulated annealing pass 1 of 1 in progress *****
RECEIVED: [HC] Existing calibration has energy 0.065
RECEIVED:  
RECEIVED: [HC] Reticulating splines...
RECEIVED: [HC] Try 0 of 50, energy=0.059 (want <= 0.010)
RECEIVED: [HC] Try 5 of 50, energy=0.082 (want <= 0.010)
RECEIVED: [HC] Try 10 of 50, energy=0.054 (want <= 0.010)
RECEIVED: [HC] Try 15 of 50, energy=0.066 (want <= 0.010)
RECEIVED: [HC] Try 20 of 50, energy=0.069 (want <= 0.010)
RECEIVED: [HC] Try 25 of 50, energy=0.050 (want <= 0.010)
RECEIVED: [HC] Try 30 of 50, energy=0.047 (want <= 0.010)
RECEIVED: [HC] Try 35 of 50, energy=0.052 (want <= 0.010)
RECEIVED: [HC] Annealing has stalled - aborting.
RECEIVED:  
RECEIVED: [HC] End of annealing pass (energy=0.052)
RECEIVED: [HC]  
RECEIVED: [HC] Heuristic calibration complete (energy=0.052)
RECEIVED:  
RECEIVED: [PK] Current kinematic settings:
RECEIVED: [PK]           Arm length: 271.332
RECEIVED: [PK]         Delta radius: 134.045
RECEIVED: [PK]      Endstop offsets: {-0.739, 0.000, -0.839}
RECEIVED: [PK] Radius offsets (ABC): {0.000, 0.246, 0.246}
RECEIVED: [PK]  Angle offsets (DEF): {-0.865, -0.876, -0.834}
RECEIVED: [PK]     Virtual shimming: {-0.625, -0.358, -0.867}, vector={-0.002, 0.003, 1.000}, d=0.585, Enabled
RECEIVED: [PK] Depth (Z) correction: Disabled
RECEIVED:  
RECEIVED: [HC] Checking calibration. If it's worse than it was before, you may have to run this several times!
RECEIVED: [DM] Depth to bed surface at center: 832 steps (4.160 mm)
RECEIVED:  
RECEIVED: [PD]                                    0.145                                  
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.020     -0.110     -0.065     -0.025      0.090    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.040     -0.015     -0.040     -0.030     -0.050    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]   0.240     -0.030     -0.055      0.000     -0.040     -0.020     -0.015 
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.075     -0.070     -0.025     -0.015     -0.035    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.035     -0.110     -0.060      0.010      0.005    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.015                                  
RECEIVED: [PD]
RECEIVED: [PD] Best=0.000, worst=0.240, min=-0.110, max=0.240, mu=-0.003, sigma=0.055, energy=0.053
G31 AF

Code: Select all

RECEIVED: [!!] Forcing re-probe.
RECEIVED: [DC] Depth-mapping calibration in progress. Press Reset to abort.
RECEIVED: [DC] /!\ PROBE CRASH DANGER /!\ Don't press buttons, send commands, or access the SD card.
RECEIVED: [DC] Probing bed for depth correction...
RECEIVED: [BH] Determining the probe-from height.
RECEIVED: [BH] Probe-from height = 346.125
RECEIVED: [BH] Bed height set to 350.220
RECEIVED: [DM] Depth to bed surface at center: 818 steps (4.090 mm)
RECEIVED:  
RECEIVED: [PD]                                    0.140                                  
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.020     -0.110     -0.065     -0.030      0.085    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.040     -0.015     -0.045     -0.040     -0.065    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]   0.240     -0.030     -0.060      0.000     -0.055     -0.040     -0.035 
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.070     -0.080     -0.040     -0.030     -0.055    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.030     -0.120     -0.065      0.000     -0.010    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.015                                  
RECEIVED: [PD]
RECEIVED: [DM] Probing at circle's edge for extrapolation...
RECEIVED:  
RECEIVED: [DM] Complete output, with extrapolated points (numbers in brackets):
RECEIVED:  
RECEIVED: [PD]                                    0.140                                  
RECEIVED: [PD]
RECEIVED: [PD] [ 0.295]     0.020     -0.110     -0.065     -0.030      0.085    [ 0.276]
RECEIVED: [PD]
RECEIVED: [PD] [ 0.167]     0.040     -0.015     -0.045     -0.040     -0.065    [-0.005]
RECEIVED: [PD]
RECEIVED: [PD]   0.240     -0.030     -0.060      0.000     -0.055     -0.040     -0.035 
RECEIVED: [PD]
RECEIVED: [PD] [ 0.191]     0.070     -0.080     -0.040     -0.030     -0.055    [-0.055]
RECEIVED: [PD]
RECEIVED: [PD] [ 0.136]     0.030     -0.120     -0.065      0.000     -0.010    [ 0.138]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.015                                  
RECEIVED: [PD]
RECEIVED: [PD] Best=0.000, worst=0.295, min=-0.120, max=0.295, mu=0.016, sigma=0.092, energy=0.058
RECEIVED:  
RECEIVED: [DC] Surface transform saved to SD card.
RECEIVED: [DC] Checking calibration...
RECEIVED: [DM] Depth to bed surface at center: 818 steps (4.090 mm)
RECEIVED:  
RECEIVED: [PD]                                   -0.005                                  
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.050      0.010      0.005      0.005     -0.005    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.020      0.010      0.010      0.000      0.005    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]   0.015      0.015      0.015      0.000      0.010      0.010      0.015 
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.030      0.015      0.015      0.015      0.020    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.030      0.020      0.015      0.020      0.020    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.025                                  
RECEIVED: [PD]
RECEIVED: [PD] Best=0.000, worst=0.050, min=-0.050, max=0.030, mu=0.006, sigma=0.013, energy=0.015
RECEIVED:  
RECEIVED: [DC] /!\ IMPORTANT /!\ Type M500 to save!
G31 Z

Code: Select all

RECEIVED: [PK] Current kinematic settings:
RECEIVED: [PK]           Arm length: 271.332
RECEIVED: [PK]         Delta radius: 134.045
RECEIVED: [PK]      Endstop offsets: {-0.739, 0.000, -0.839}
RECEIVED: [PK] Radius offsets (ABC): {0.000, 0.246, 0.246}
RECEIVED: [PK]  Angle offsets (DEF): {-0.865, -0.876, -0.834}
RECEIVED: [PK]     Virtual shimming: {-0.625, -0.358, -0.867}, vector={-0.002, 0.003, 1.000}, d=0.585, Enabled
RECEIVED: [PK] Depth (Z) correction: Enabled
RECEIVED:  
SENT: M105
RECEIVED: [BH] Determining the probe-from height.
RECEIVED: [BH] Probe-from height = 346.125
RECEIVED: [BH] Bed height set to 350.225
RECEIVED: [DM] Depth to bed surface at center: 819 steps (4.095 mm)
RECEIVED:  
RECEIVED: [PD]                                    0.005                                  
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]    -0.005      0.000      0.000      0.005      0.010    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.010      0.010      0.010      0.005      0.015    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]   0.020      0.010      0.010      0.000      0.015      0.020      0.020 
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.020      0.020      0.020      0.020      0.025    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD] [  --  ]     0.025      0.015      0.015      0.020      0.020    [  --  ]
RECEIVED: [PD]
RECEIVED: [PD]                                    0.020                                  
RECEIVED: [PD]
RECEIVED: [PD] Best=0.000, worst=0.025, min=-0.005, max=0.025, mu=0.008, sigma=0.009, energy=0.014
User avatar
mlapaglia
Printmaster!
Posts: 121
Joined: Sat May 24, 2014 10:41 am
Location: Indianapolis, IN

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by mlapaglia »

gestalt73 wrote:Oh crap, what a bummer, I thought I could get away with posting google drive images if the folder was public. Thanks for letting me know.

Here's the results of my G32, G31 A calibration with the new probe.

Code: Select all

[PD]                         0.010
[PD]
[PD] [ 0.000]     0.000      0.020      0.020    [ 0.000]
[PD]
[PD]   0.005      0.005      0.000      0.015      0.035
[PD]
[PD] [ 0.000]     0.000      0.010      0.000    [ 0.000]
[PD]
[PD]                         0.010
[PD] Best=0.000, worst=0.035, min=0.000, max=0.035, mu=0.005, sigma=0.009, energy=0.011
[img]http://i.imgur.com/FGc5DXi.jpg[/img]
Do you have the STL for that mount anywhere?
User avatar
ZakRabbit
Printmaster!
Posts: 186
Joined: Sat Jun 20, 2015 5:07 pm
Location: Orange, CA

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by ZakRabbit »

I've read completely through this thread in preparation for building my unit. If I'm using FSR's, does that mean the top limit switches are redundant, or does this augment them? I'll be using the Smoothie 5X, with the 626pilot branch. Thank you all for putting so much work into this, I hope that at some point I will be able to advance this technology as much as you have.
User avatar
KAS
ULTIMATE 3D JEDI
Posts: 1157
Joined: Thu Dec 04, 2014 6:06 pm

Re: Heuristic (AI) calibration for delta printers on Smoothi

Post by KAS »

The FSR would attach like a probe in one of the "Min" Stops while the "Max" stops would be for the limit switches at the top. You would still need the top switches to home & calibrate the Delta.


[img]http://i.imgur.com/keOPkoC.png[/img]
Post Reply

Return to “Smoothieboard and variants”