High spots between towers.

[Toddwagner]

UPDATE TIME:
I got in the effector plate and carbon fiber arms from trick laser today and installed them on my printer. Once it was all assembled I ran the delta calibration wizard twice through, set up a full bed print and pressed start. One hour of impatient waiting later, a full bed completely flat print was waiting for my on my printer! Wow!

[Gdot]

Good News Todd!

I also received my carbon fiber delta rods and aluminum effector plate. I installed last night but did not have much time to run calibrations, but my initial pass through delta calibration still showed high spots between the towers as reported back by the z probe. I ended up only using the bottom half of the effector plate since I do not yet have a E3D extruder. It all fits together so much tighter now and the slop is completely eliminated, plus it just looks much better. I didn't do a full clearing of the EEPROM, but I will tonight and start from scratch with all the new gear on it. Todd, do you mind sharing a photo of your new gear so I can compare along with what settings you arrived at for Horizontal Radius? Did you run G29 and the delta wizard with default probe points from the firmware? What deviation did you end up at after the second iteration of the delta calibration?

[Toddwagner]

Gdot,
I'll attach photos of my complete EEPROM, so you can check everything out. You'll notice that I have the rod length about 7mm longer than what my rods from TrickLaser are measured at. That is because after running the delta wizard, it suggested I change the number to the 307.01 in the EEPROM, so I did it...

As far as my process with calibrating went, I ran the G29 once with Hackneys 135mm radius probe points (I'll Also include a photo of that at the end). Then I ran the wizard and wrote down the ten z-offsets and plugged them into the site. I took the complete list of new EEPROM settings from the site and manually entered them into my system. Then repeated this again on a warm bed (G29 and then probing) and arrived where I am now at about .03 deviation. The only thing I had to manually adjust was the extrusion settings to ensure proper welding of the lines on the bed.

I just printed my 50x50x50mm test cube and it measured about .6mm less in the x and y and about .15 less in the z, so I'm sure I will have to adjust the rod length again. I did the math quickly and I'm going to change my rod length to ~303.76 and recal and try it again, so I'll be updating this thread alongside you this weekend.

Now, attached are also some photos of my setup (he280 probe board and hot end, trick laser arms, and trick laser effector plate). I am not using any layer fans as of now because I haven't designed a new mounting system for them to use on this effector platform. I took the photos I think you would need to see, so just let me know if you need anything different!

[Gdot]

Toddwagner

Thank you very much for taking the time to respond with pictures and settings. I will compare tonight after work (I should be working anyway... ). As far as fan shoruds, I really like these http://repables.com/r/832/ . They fit right on the little tabs at the 3 locations on the bottom effector plate. Tricklaser even included the allen screws with the plate as well to attach something there. I'll attach a pic tonight.

[Toddwagner]

Toddwagner

Thank you very much for taking the time to respond with pictures and settings. I will compare tonight after work (I should be working anyway... ). As far as fan shoruds, I really like these http://repables.com/r/832/ . They fit right on the little tabs at the 3 locations on the bottom effector plate. Tricklaser even included the allen screws with the plate as well to attach something there. I'll attach a pic tonight.

I did print that duct out, but ever since I installed these new parts my plate is much closer to the bed, so I can not fit those ducts underneath it. As you can tell by the pictures it is MUCH lower, so I will redesign my own.

[Gdot]

Well I'm afraid there has been no change for me. After installing the new arms and effector plate, I cleared the EEPROM and and reflashed. I then ran G29 multiple times followed up by the Octoprint delata calibration plug in. It can only get me to .17 deviation. When running the gcode for the same probe points, my sine wave pattern is still there with the towers at (negative)-.20 +/- to the mid tower points being (positive).20 +/- . I even made sure all my screws were adjusted to give me endstop offsets of 2,3,0. I even bought a digital angle gauge to check for tower lean and everything was with in .1 to .3 degrees. What am I missing??????

[Toddwagner]

Well I'm afraid there has been no change for me. After installing the new arms and effector plate, I cleared the EEPROM and and reflashed. I then ran G29 multiple times followed up by the Octoprint delata calibration plug in. It can only get me to .17 deviation. When running the gcode for the same probe points, my sine wave pattern is still there with the towers at (negative)-.20 +/- to the mid tower points being (positive).20 +/- . I even made sure all my screws were adjusted to give me endstop offsets of 2,3,0. I even bought a digital angle gauge to check for tower lean and everything was with in .1 to .3 degrees. What am I missing??????

Try doing exactly what I did to see if that works for you?

[Gdot]

If I use the least squares calibration and choose 7 factor calibration, I get a .03 deviation with a new DR length of 307.81mm, which is up from the 300.15mm of the original length. From what I have read though, this is not the solution since it will affect the scaling of the print object. Have you tried to print a test cube and take measurements yet? I will do that this morning and let you know.

[Gdot]

Printed the 50 mm test piece and came up with 49.5 to 49.55 mm dimensions. So I dont consider 7 factor calibration to be the solution.

[Mac The Knife]

I am curious. What is your goal? If you're just playing around to see who can get closest to 50 mm,,, well have fun. But if you actually need it to be 50mm ±0.001, I think you need to find another process.

[Toddwagner]

Printed the 50 mm test piece and came up with 49.5 to 49.55 mm dimensions. So I dont consider 7 factor calibration to be the solution.

Gdot,
You're aiming for the lowest deviation, but I don't think you should worry about the numbers as much as what the print looks like. The big issue here is getting a flat first layer, not getting the deviation perfect. I just ran the probing wizard again and my deviation was ~.13, yet I'm still getting perfect first layers.
Now I'm not saying a pro at this (which I think is obvious by how bad I am at this stuff), but I think you should go back to the stock eeprom and follow what I did exactly. Don't worry about the numbers, just go through the process and give a single layer print a shot!!! Do not use the deviation numbers as an indicator of print quality, use the print quality itself!

[Toddwagner]

Also yes I have printed cubes, and I had to increase my steps/mm to get the right z-height and I'm adjusting the DR length as we speak. I am ~.1mm larger on the y axis than on the x, so I'm going to also try to fix that.

I am curious. What is your goal? If you're just playing around to see who can get closest to 50 mm,,, well have fun. But if you actually need it to be 50mm ±0.001, I think you need to find another process.

I am just searching for consistent prints as close to the 50mm as I can get, not searching for that percise of a piece out of this machine

[Mac The Knife]

As long as no one pokes an eye out,,,,,

[Gdot]

So yes Toddwagner, I agree, I stopped getting hung up on deviation results and started to just concentrate on test print results. Except, my test prints were failing miserably. I had bad test prints and no deviation lower than .17. Then I noticed something peculiar, any temp much above 50 deg C and my bed was bowing very badly in the center. I was running my bed at 60 deg C, I lowered it and ran the Octoprint delta wizard and still got a .17 deviation, but this time my first test print (111 mm x 111 mm) came out perfect. I will try bigger test prints later this afternoon. Not sure why the bed is popping up so badly in the center at temp, but it was bad enough, my glass would wobble on it when not clamped down.

[Toddwagner]

If you're just printing PLA you should try printing without the heated bed to see if it's any better

[Gdot]

That is one of my troubleshooting steps, eventually, but in relation to deviation numbers, my cold deviation numbers always matched the warm deviation numbers. I need to correct myself in regards to my previous post. I prematurely stated my test print was perfect, that was wrong, after I pulled the test print up, it had 3/4 of the print area with unfused print lines. So I began to hand tweak the endstop corrections and X tower angles. After several iterations of adjusting xyz endstop corrections and x tower angles(each time followed by a z height adjustment), I got to a 95% fused test print. My test print was a very smallish 111 mm x 111mm, so I need to try something bigger to see if my manual calibration still holds up further out to the perimeter. As far as the adjustments, they only amounted to +/ - 10 steps or so for endstops and .1 degrees for tower angles. So that was a huge learning curve for me that very small adjustments can lead to huge gains in calibration results. The other learning curve was that automatic calibration (for me) only gets me part of the way. I will test further tonight. I tried my hand at manual calibration by only using the screws and following the steps outlined in several posts on this site, but I got to a point where I could no longer screw in the adjustment screws so I gave up. It later dawned on me that I could have started out with some increased endstop correction in firmware for each tower, so that I would have had room to go up or down with the settings. I still want to try that again. So I tested my 149 mm test print, didnt look as good as the 111mm. Probably because the calibration issues get worse the farther I go out to the perimeter. But....they look much better than what my first test print looked like right after an automatic calibration before I manually tweaked the settings.Something odd though, I always see these circular patters in the unfused areas,
I think you can see them really well in the first pic.

[Toddwagner]

Gdot,
First of all congrats on getting a better print than when you started, that's good news.
I'm sure you've done both of these things, but just to be sure, have you rotated your bed, including the insulator and heating element to see if that changes anything? How about the bed mounts themselves? It's possible that if you change around the locations of the bed mounts it might help you find the cause of this issue.
The second thing I am curious about is your hot end mounting on the new effector plate. You're absolutely sure it is vertical (or at least parallel with your three X,Y,and Z towers)?

Let me know if you've tried these or not!
Todd

[Gdot]

I'll check all those things, but tomorrow I plan on re-calibrating cold bed and printing to a cold bed to see what differences that makes.

[Gdot]

Ok, I checked all my bed mounts and they all measured the same. I put them back on in random positions. I tried different rotations for the bed as well, same result. I now calibrate with no bed heat as well as print without bed heat and that has improved my first layer by quite a bit as you can see from the picture. It still has odd unfused areas at x to z tower and y to z tower on the outer edges. It looks like a simple horizontal radius adjustment would fix it, but it does not make any improvement.

[Toddwagner]

At this point It might not be unreasonable to think your glass is the defective part, so it is never truly flat especially when heated...

[Jimustanguitar]

Mark the glass, and index it around. Does it follow the glass or the machine?

Rotate your XYZ motors and endstops all over 1 position (YZX instead of XYZ if that makes sense) and see if it follows the mechanics or the electronics.