Welcome to the SeeMeCNC Community

Hi everybody. I hope Gene will read/critique this as he may want to publish it or something like it to add to the already OUTSTANDING user guide.

I first loosened all my tower screws and carefully with a good 24" framing square with short side on the towers, long side on the glass bed - tweeked tower position such that all were 90 degrees. I can't emphasize how much easier this last tower angle check makes calibration (reduces the number of iterations as I dramatically found out trying to calibrate before doing this). When tightening the tower screws I tightened all number 1 (see build instructions) base screws,checked angles, tightened all number 2 base screws, checked angles, tightened all number 3 base screws, checked angles, tightened all number 4 base screws, checked angles, tightened all number 1 top tower screws, etc you get the idea. Rechecking the angles all those times turned out to be totally unnecessary, but having found the angles not right after the below described embarrassing incident and probably the build - and a bunch of less than satisfying calibration iterations, I was paranoid.

But here I have a very embarrassing confession to make and warning. If you have left your hot end .2mm above the bed and you decide to run the z zero set routine, DO NOT FORGET TO HOME THE Z BEFORE YOU GO TWIRLING THAT LCD KNOB!!! I FORGOT AND . . .CRASH. But it was a good learning experience. The force of the crash, which happened at the z tower calibration location, did not harm the glass or hot end (so far as I can tell by close inspection), but was enough force to knock the tower angles off 90 degrees ( but not otherwise harm the towers, Delta arms, or anything else I can see). So don't forget to home before TWIRLING that knob or, I suggest as best practice, entering any other move command.

Next step: set z routine and Z=0.Don't forget to home before TWIRLING that knob! Then run the macro that goes to/pauses in center and all three towers enough iterations with stop screw adjusts and z resets to get the tower position gaps visually equal. I suggest adjusting the high and the low towers each time , not all three and not just one - it won't be a juggling act. Then, if your horizontal radius is visibly off to the observing eye, then follow Gene's correction in .2 iterations. My default number was 142.68. I ended up at 143.68. Set z once more ( with the sheet of copy paper, of course) to set yourself up for the gcode proceedure I describe below.

This is where I start to depart from the published procedure.

First let me make a case for studying the G code in the macro set up instructions just enough to know how it instructs your printer to move. It pays off with HUGE dividends and opens up the use of the terminal window. Yes the canned macros are quicker, but if you learn a modicum of code you can tailor your own macros.If a 5 second pause is not long enough for you to get your act together with your eyes or with feeler gauges, type in S10 or S20 instead of S5. You know much more about how cnc actually makes your machine move. And it's REALLY cool to tell your machine to go somewhere, then watch it do it. It's a confidence builder extraordiniare.

That said: here we go.

Next: don't just get and clean one feeler gauge (.009" suggested in the alternate calibration method discussed in the appendix). Get .005" through .015" , and every thousands of an inch in between.

Next, go to the terminal window in options and type in G28 (Home All). Press Send. Type in G1 Z0.2 F1500. Send. The machine goes slowly to the center of build plate .02 mm above the bed. ( the F value is your move speed; I saw no reason to rush the machine when its headed for a close encounter with the glass after the above described incident. Now get out your feeler gauges and start with the .005" one and swipe them under the hot end until you get to a thickness that has ANY friction whatsoever. The last one that swiped without friction is the gap that you record (keep a chart). My gap measured .010". Next type in and send G1 X-77.94 Y-45 Z0.2 F1500. Measure and record measurement you get. Next type in and send G1 X77.94 Y-45 Z0.2 F1500 and record measurement you get. Finally type in and send G1 X0 Y90 Z0.2 F1500 and record the feeler guage measurement you get. Make your end stop screw and/ or horizontal radius adjustments for this center,x,y,z iteration then reset the z with the piece of copy paper.

Using the manual gcode commands and feeler gauges, I got calibration to within .001" in just two iterations after running the macro just once ( after meticulously squaring up my towers). I think I could get .0001" accuracy if I wanted to bother; the printer is incredibly consistent.

It might be argued that hand typing those gcodes is beyond the average builder's attention span/patience - but I think it's only a small amount more time, teaches ALOT, and ( with the set of feeler gauges) is much more accurate than my 63 year old eyes comparing the gaps in the 5 second macro routine.

Note: if you screw up typing in a coordinate and type in a double negative, for example, the hot end and bed heater shut down ( safety feature I presume;again, a learning experience).

I hope my solutions will be of help to someone.

Thanks you for posting this Eric LB, you have included details that could of help for those squaring up their towers.

Squaring to the glass seems like it would definitely be the way to go. I also found it helpful to use a digital angle gauge, like you'd use to square a table saw.

Hi all,

Now that I had my fun with gcode, and using this rudimentary knowledge, I did create that tower cal macro with Z= 0.2 and 20 second pause times at each location and ending in the build plate center; plenty of time at each location to get three feeler gauge thicknesses between hot end nozzle and glass. And then have end stop screws accessible for adjustments. Recalibration and new z=0 set now takes 5 minutes. I highly recommend this after installing the PEEK fan/ shroud. My z height had increased @.15 mm, enough to make ABS first layer not stick to the purple glue. Also z tower needed minor adjustment, for reasons unknown other than having disassembled the Delta arms for PEEK fan installation.

Having just finished my second marathon calibration I also have a small question and maybe suggestion to make regarding horizontal radius changes in the manual.

Currently the advice is: "If from the center position the nozzle goes up from the glass at all three towers, you'll want to lower the Horizontal Radius by 0.2. Run the Tower Cal macro after each change to check the effectiveness of the change"

I would recommend changing it to something like:"If from the center position the nozzle goes up from the glass at all three towers, you'll want to lower the Horizontal Radius by slightly less (to be safe) than the approximate height gap of the smallest axis gap (in mm). Run the Tower Cal macro after each change to check the effectiveness of the change. Repeat in 0.1mm increments if minor adjustments are still needed."

vice versa for lower at all outer corners

Reasoning
My gap at all 3 corners was 8-12mm! I dutifully adjusted in 0.2 increments for a few painful eternities yesterday and ended around 132mm. Yes i double checked diagonal rod length, temperatures. When a repair forced my to remove the bed and recalibrate today I reset at 144m and ended up at 132mm in <1 minute by measuring the gap.

Now I don't know the mathematics of the delta modeling - maybe this is just lucky circumstance or maybe I was causing my own drift by the following:

Side Note
I got in the habit of resetting z-height (z0) after making horizontal radius changes. Instructions just say to rerun the calibration macro. Seemed like large steps changed that measurement in ways that would've had me grinding glass otherwise. Then again maybe this was me undoing the horizontal radius correction and causing the large drift from expected? Or maybe this needs to be added as a step when significant correction is required. Were the z0 adjustments between radius adjustments required or were they causing my drift? Does this question make any sense?