Strange Z-height issues

[Vinculum]

Hi all!

Before i begin, i must admit that a do not own an awesome SeeMeCNC machine but i am building the Rostock Mini. Since this is basicly the same machine but smaller i hoped it would be ok to post my question here since i'm stuck at reprap.org.

The basics of my machine:
Rostock Mini build like on reprap.org/wiki/Rostock_mini
Marlin firmware based on Johann's (latest) release of june 7th from the kossel branch. Modified only mechanical settings to adapt to my setup
RAMPS 1.4 with LCD and SD option
Calibrating using Prontrface for Mac
Slicing with KISSlicer (but not relevant just yet)

My problem is this. I can calibrate z-height beneath the towers nicely with the set-screws. I can just move a piece of paper underneath the hot end at these points which should be ok. Software z-height is 200mm at these points.
Going to Z0 @ center slams into the bed. After some carefull testing the bedcenter seems to be 0.9mm higher that the rest. The bed itself is borosilicate glass.
Off course the problem should be in DELTA_RADIUS. But here's the strange thing. It doesn't matter if i lower the DELTA_SMOOTH_ROD_OFFSET or raise it. Not even by steps of 50. Every time i re-upload after a change it still slams in the bed. The EEPROM setting is turned off so my firmware should reflect what's actually in the Arduino.

What am i doing wrong?

I really hope you can help me, i get no reaction at forums.reprap.org in the deltaforum. The dutch part of it is helpfull but no one has a delta. I'm stuck...

[September]

Hoi,

Hoe gaat het met uw?

That's about the limits of my Dutch, anyway I had a similar problem.

Firstly is your bed level? Grab something with a straight edge, also make sure each tower is square.

For myself I changed PRINTER_DELTA_RADIUS and added an offset to DELTA_RADIUS

If your nozzle is raising you should increase the value by 0.5 at a time.

Also make sure your end stops are hitting at the same location each time. For me I actually shifted the end stops back a little so the screw heads weren't hitting the the tip of the end stop.

I too did try changing by huge amounts just to see if it actually made a difference, which it didn't.

Tot ziens

[Vinculum]

Hoi,

Hoe gaat het met uw?

Heel goed, dank u wel!
"Uw" is actually more like "your". It should be just "u". But that's totally off-topic...

Firstly is your bed level? Grab something with a straight edge, also make sure each tower is square.

Will re-check but i'm fairly sure it's level.

For myself I changed PRINTER_DELTA_RADIUS and added an offset to DELTA_RADIUS

I'll try that.

If your nozzle is raising you should increase the value by 0.5 at a time.

That's how i've started but after that didn't do it i've tried to go extreme with incements of 2 digits.

Also make sure your end stops are hitting at the same location each time. For me I actually shifted the end stops back a little so the screw heads weren't hitting the the tip of the end stop.

Again, pretty sure they are ok. I´ve G28´t and then gone to the tower position a hundred times and it hits the spot (under the towers) again and again. It couldn't do that if the endstops were shifting.

I'd sure wish there was a manual or something to implement Johann's automatic bed probe...

[September]

Ah I must have it mixed up, I was thinking uw was the formal you, perhaps it was jou that I'm thinking of or the reverse and u is the formal.

I've rechecked my changes, I ended up leaving the offset as zero and just altered the PRINTER_RADIUS, granted that in the end they all get calculated and make up the DELTA_RADIUS.

Another to check is to make sure that your pulleys teeth is set correctly, the rod lengths, END_EFFECTOR_HORIZONTAL_OFFSET and CARRIAGE_HORIZONTAL_OFFSET are correct or at least near to it.

Are you testing with the hotend and hotbed on? As apparently and I'm sure it does happen, they will expand and change the distances. So make sure you test with the heat set close to or the same as what you'll be printing at.

There is an Excel spreadsheet floating around here somewhere, I'm not sure how accurate it is, I tried to use it myself but ended up going back to trial and error. Basically it takes the XYZ towers and the centre Z value and calculates the expected offset for the DELTA_RADIUS, it might work for you.

From memory I think it would calculate your offset for DELTA_RADIUS to be +0.90

The Z values you are judging by, is that by the program or from the LCD display? For the Repetier I've read that sometimes the values reported can't be trusted and to execute GCODE M114, this gets the results from the printer itself. I presume the readout on the LCD is as accurate.

I'm assuming M114 would produce the same results and report back the XYZ of the head.

Good luck, it took me many days to get it right.

[kbob]

Instead of going directly to Z=0, what happens if you gradually lower the effector until it touches the paper? Then record your Z value. After you make a change to DELTA_RADIUS or equivalent, does the Z height change?

I ask because I wonder whether somehow you are not actually updating the firmware.

Also, are you using Repetier Host + Repetier Firmware? If you are, there is also a setting R-Host for the printer's Z height. It is under Printer Settings -> Dimensions -> Delta Printer -> Height.

[Vinculum]

Yesterday eveneing i've spent some time doing the calibration dance. This time i removed the hotend and replaced it with a micrometer. To compensate for the lower meter i've increased the max height in the firmware by 20mm. This also proves that my firmware uploads are coming through. Right under the towers is top notch for all 3 within 0.05mm. Every adjustment is doing a G28 and than going to a point under the tower (in my case G1 X60 Y-60 Z0 F1500 for thy Y tower for example). Hits the spot every time. The micrometer is approx at 1.8mm at those points. After checking and rechecking thos points i moved it from G28 to Z0. Micrometer hits 3.75mm.
I've made a video showing what happens.
[youtube]http://www.youtube.com/watch?v=2E7-6AEHdaA[/youtube]
Don't mind the actual numbers on the micrometer. You can clearly see that it's dipping in the center.

I'm running Marlin firmware right now. The configuration.h is in the attachment.
Maybe i'll try the repetier firmware but i'm not sure it is capable of working with an lcd/sd but i'll look into that.

[Polygonhell]

You probably need to adjust the PRINTER_RADIUS or Marlin equivalent, I think it's covered in the latest manual.

[Vinculum]

You probably need to adjust the PRINTER_RADIUS or Marlin equivalent, I think it's covered in the latest manual.

In know. The problem is that changing it doesn't do anything.

O've tried the Repetier firmware and host but that's not a succes yet. After making the obvious changes and copying most from my Marlin config, i tried it. Homing went in the oppositie direction so i inverted all axis which fixed that issue. But now it keeps slamming into the endstops. M119 does defect them correctly, L by default and H when triggered. I've copied the endstop settings from my Marlin config as posted earlier. Attached you'll find my Repetier config. I must me doing something stupid with the min/max settings.

Repetier host on Mac is strange as well. I can't send gcodes. Only GUI clicks get through. So far, it's giving me more headaches than solutions...

[Polygonhell]

Repetir Host on Mac works fine, you do need to capitalize all letters used and correctly format GCode
i.e. G1 Z50 is not the same as g1 z50

OK I just realized you don't have a Max, so you are having to manually configure the firmware.
The printer dimension, specifically DELTA_RADIUS and DELTA_DIAGONAL must both be at least approximately correct before you will get useful results.

[Vinculum]

Repetir Host on Mac works fine, you do need to capitalize all letters used and correctly format GCode
i.e. G1 Z50 is not the same as g1 z50

I did not now that. I was to lazy to try capitals...

OK I just realized you don't have a Max, so you are having to manually configure the firmware.
The printer dimension, specifically DELTA_RADIUS and DELTA_DIAGONAL must both be at least approximately correct before you will get useful results.

I know off course i have to change settings. Although it says n00b, i'm not (well not totally at least).
The Max and the Mini are only different in size. The rest is the same. So i altered all size related settings and off course steps/mm settings etc. Also added SD and LCD support. Other than that it's standard repetier. And writing this makes me smack my head against the desk. I've based my firmware on the Repetier firmware, not the Max firmware from this site. Damn, a n00b after all...
I'll alter it and try it asap. Oh i'm so embarrassed now...

[JohnStack]

No one mentioned flashing problems - a possible.

Why build a smaller format Rostock? Just curious... Is it for travel in the NS Train? LOL

We're all noobs - always....

[kbob]

Would this have anything to do with your problem?

http://forums.reprap.org/read.php?178,2 ... msg-209694
(Yet another helpful post from the faceted inferno man)

[Jimustanguitar]

No one mentioned flashing problems - a possible.

I second that. I had a heck of a time getting my EEPROM to update "by the book". After beating it to death, it finally took.

[626Pilot]

I'd say to get a framing square (narrow enough to fit in the grooves on your towers) and a couple of Irwin Quick-Grip clamps or whatever the equivalent is in the Netherlands. I realigned my Rostock a couple of days ago by clamping the top and bottom of the framing square to my towers. It works very well. Make sure you get all your towers as close to perfectly square as possible, top and bottom. (On top it's more important to ensure equivalent tower height.) Your glass build plate should be perfectly level and it should NOT bow upwards in the middle, which, based on your dial gauge video, appears to be the case... but I don't think it's really bowed, I think it's a geometry problem. That suggests to me either tower alignment or (as you say) software. Might as well check them both at the same time.

I did everything by hard-coding changes into Configuration.h but after awhile I started using EEPROM. (It was erroneously reading garbage because Marlin was installed before, but I changed the EEPROM version to 10 to force it to initialize from scratch and it worked after that.) Speeds it up quite a lot. I haven't had any issues with changes not taking. As soon as you save from the EEPROM screen, it's live. The only gotcha is that you have to press Enter after you make your last change. If you don't press Enter, it won't save the last edited value.

If you don't trust the EEPROM, change a value, exit Repetier, go back in, and then check the EEPROM to see if your change took place. Once you pass that test your testing cycle will be 5-10x quicker. You don't have to reset the board or anything like that for the change to take place, it's immediate.

[Vinculum]

Thanx for all the great help guys!
I don't know what happened but yesterday i went back to Marlin (Max Repetier firmware didn't compile?!?) and something happened when i changed the radius!
So after tinkering with it for some time (was very late last night) this is my result.
http://youtu.be/r5JQ47dgs50
I know, focus wasn't great but it was late. It's flat within 0,05mm!
There is some play on the traxxas joints but some rubberband fixed that.

[Vinculum]

My first "print"!
[youtube]http://www.youtube.com/watch?v=enRH4h760Po[/youtube]

It's in reverse though. That's kinda strange. But it's nice and level as far as i can judge. Now for calibrating the extruder and making a real print.

[Vinculum]

Why build a smaller format Rostock? Just curious...

Haha, at the risk of sounding like Bill Gates who asked once who would ever need more than 256k memory but i don't need the big space. I like it because it is small. Plus, i don't own a huge american house but a small dutch one...

[lordbinky]

Ugh, quit making sense. What you should do though is make it multi-function, such as adding another Tier and make an end-table out of it, or add a lamp to it. Oh Oh! or a clocks, clocks always improve anything, even other clocks.