New drama

[theverant]

So things were progressing nicely and got some okay test prints... Then I noticed... something...

Something is very off with the levelling - each of the posts are dead on with the centre. But between each of the posts I have ~ 0.4mm rise.

Is that bizarre? I think it's bizarre.

[JohnStack]

I think it's bizarre.

Bizarre. You should try mic'ing all of the sides of the t-slot if you want bizarre or taking them off and seeing what kind of twist they have.

(Not that SeeMeCNC has any control over this part of it.)

How are you going to resolve?

[theverant]

I wish I knew what to do! I don't actually have any idea what to look at for this. I tried relevelling, and some tweaks were done, firmware reuploaded, but the problem persists - the further from centre, and the more between the posts the head is, the higher it gets.

[theverant]

Okay, playing more - something is definitely eff'd:

If I go from the y post (G0 Z0 X77.94 Y-45 F3500) to the x post (G0 Z0 X-77.94 Y-45 F3500) the head moves up, then arcs back down and touched the print surface. But if I G28 then go to the x post (G0 Z0 X77.94 Y-45 F3500) I get a perfect 0.1" spacing.

Ideas?

[theverant]

Okay, when the print head moves it arcs up quickly, then comes back down, it doesn't matter if I do it X-Y, Y-X, Z-Y, Y-Z. So, basically the high point of the arc is the same distance from the starting post, not the same point on the bed for any random move. Very mysterious.

[lordbinky]

Since there's alot to releveling, did you explicitly check what JohnStack suggested?

That is, make sure your towers are square to the base their whole length up (or just under the cheapskate at least) on each side of the the towers, as in don't assume they are perfectly straight and square their entire length. It's important to rule that out as a possibility. Also if the top plate wasn't perfectly matched to the base as mentioned in the new guide, you'll a similar effect.

[lordbinky]

Since you're issuing travel commands, make sure there isn't a setting that has performs a Z-Lift before/during traveling.

[geneb]

Sounds like you've got a doming problem. You might want to copy down your EEPROM values, change EEPROM_MODE to zero, update the firmware, power cycle the board and change it back to 1.

g.

[theverant]

Since there's alot to releveling, did you explicitly check what JohnStack suggested?

That is, make sure your towers are square to the base their whole length up (or just under the cheapskate at least) on each side of the the towers, as in don't assume they are perfectly straight and square their entire length. It's important to rule that out as a possibility. Also if the top plate wasn't perfectly matched to the base as mentioned in the new guide, you'll a similar effect.

Thanks for the input.

I thought it was odd that the height is relative to the starting point, not the actual physical position. IE the maximum dome is the same distance from the post it starts from, not a specific geometry of the print surface. It's as if when it starts some kind of resistance is causing the print head to pop up, then settled back down. But, I will double check the square of the towers. I did do it just recently and fixed some issues.

[theverant]

Sounds like you've got a doming problem. You might want to copy down your EEPROM values, change EEPROM_MODE to zero, update the firmware, power cycle the board and change it back to 1.

g.

Thanks for the idea. I did have a problem with the firmware/EEPROM before, I will try this and see if anything changes. If it is a doming problem, it's an irregular one (see my post above)

[kbob]

Is this effect speed-dependent? I've seen some nonlinear motion when I tried to move the carriages faster than they could go.

[lordbinky]

You know what? If your going to ask simple things that were overlooked while we rack our brains on less plausible causes, go away.

Really though, good suggestion, I facepalmed myself for myself when I read that. I was picturing the problem in slow motion thinking what could affect it like that, so it would have been a loooong time before I came anywhere near that thought. I give up thinking for today.

[kbob]

The next thing would be to post/paste the dimensions from your Coufiguration.h. What do you have for DELTA_DIAGONAL_ROD, END_EFFECTOR_HORIZONTAL_OFFSET, CARRIAGE_HORIZONTAL_OFFSET, PRINTER_RADIUS, and DELTA_RADIUS?

It would probably be best to just paste in that whole section of Configuration.h -- C source code is tricky if you don't speak the language.

[theverant]

So, I'm pretty sure it's mechanical.

If I go between the posts @ F3500 I get the rise fall behaviour. If I do F100, it's perfectly level. Too much resistance on the effector?

[geneb]

That sounds pretty plausible. Have you noticed any print quality changes?

g.

[theverant]

That sounds pretty plausible. Have you noticed any print quality changes?

g.

To be honest I've never really gotten the printer to a point where I was getting usable prints. I've been pecking at it for awhile, this is the first time where I've been able to really spend a bit of time trying to sort it all out. I did sand down the arms as in the Delta Arm Blues thread, and I double checked that my posts were square (they weren't). Had that whole eeprom bug. It's been a very trying process. I'm at a loss as to what might be causing this new issue, was hoping the boards here would have ideas. Definitely seems mechanical, though. I'm on the verge of ordering the carbon arms from TL and doing a complete teardown and rebuild with your updated manual (I used a very early version back around xmas)

[geneb]

Do you have one of the Indiegogo kits, or something after that? Are you using the SeeMeCNC machined (15 tooth) gears or the new GT2 20 tooth gears for the drive belts?

g.

[theverant]

Do you have one of the Indiegogo kits, or something after that? Are you using the SeeMeCNC machined (15 tooth) gears or the new GT2 20 tooth gears for the drive belts?

g.

I got a kit just after Indiegogo, Rambo 1.1, 15 tooth gears. Stock arms, peek fan w/ SMCNC's shroud, Onyx bed. I also installed some isolators on the motors to quiet down the machine.

Are the GT2 gears better?

[geneb]

Oh yes! They won't devour your belts if the nozzle smacks the bed!

They're well worth the upgrade hassle. I'm pretty sure that SeeMeCNC is selling the all-metal ones now as well. (The original GT2 gears were metal cores on injection molded plastic.)

Just don't forget to tweak the steps_per_mm in your firmware when you go to the new drive gears.

I'd order a new set of belts as well - if you've been running on those sharp gears, you've probably worn the belt teeth quit a bit by now. This may be a contributing factor to the "arch" you're seeing when moving from tower to tower.

g.

[edward]

Since this thread seems to describe the same issue that I am experiencing (XYZ posts and Center are all level, effector raises at FLR - see attachment image or link), I'm going to jump in to the discussion. I've had this issue since my machine was first operational (purchased beginning of June), and have been getting amazing prints as long as I stay within a ~100mm diameter circle right around the center.

Originally I had a borosilicate glass build surface on top of the Onyx. It was domed even when cold by 0.5mm high in the center. I have recently made a huge overhaul and upgrade (which I had naively hoped would fix the problems) by going to xnaron's mag-arms and a 5/16" cast aluminum build plate that I verified to be flat to better than 0.001." My towers were recently re-squared with a 10"x18" machinist's blade square and the inter-tower distances set with a large caliper to within 0.002." The six carbon fiber arms are measured to have <0.2mm variance with the chrome balls in place.

Right now I've got a dial indicator in my effector. As I said, just like the OP, I can get XYZ and C exactly level. Move to F, L, or R and I'm 0.01-0.018" high (points F & R are 0.018" high and L is 0.01" high...I haven't been able to explain the difference or clearly identify the source).

The only irregularity that I have found is between the Y and Z towers. Starting around half-way up the distance begins to increase to the point that it is 0.012" wider at the top than at the base. To me this suggests a slight twist in the Y tower, but the XY distance matches top and bottom exactly (and is identical to the XZ distances). All towers are square to the aluminum plate, up the 18" side of the square, to better than 0.001" (that is the thinnest feeler gauge that I have and I can't insert it at any place along the square).

At this point I am completely at a loss for ideas. I've tried EEPROM on. EEPROM is currently disabled. I've tried Marlin-Delta. I'm back on Repetier. Since it's now so easy I've swapped around the positions of the mag-arms to no effect. I've measured and mic'd every thing I can and just don't get it.

Do I understand that most of you get XYZ & C level and it's all gumdrops and rainbows at points FLR?

Sorry if this sounded like a mini-rant. I'm frustrated and really interested in getting this fixed.

Any thoughts?

Image obtained here: http://forums.reprap.org/read.php?178,2 ... msg-201301

[edward]

Well...maybe I scared everyone off

After going through the Delta forums at RepRap.org I found some insightful comments for calibrating the delta platform with regards to the auto-leveling routine.

After a short bit of adjusting the tower positions (I had to move the X and Z towers toward the center by something like 1-1.5mm) I managed to get the FLR error to be consistent. Still high, but equal at all three points. More than a couple hours later of slowly adjusting the DELTA_DIAGONAL_ROD and PRINTER_RADIUS parameters I now can sweep the bed to better than 0.008" (~0.2mm). A huge improvement, but it still seems like quite a bit of variation compared to what I am reading around here.

I've yet to print with these settings, but I'm worried that since both of the aforementioned parameters are so much higher than the actual values (several millimeters over measured) my scaling will be wrong.

We'll see...

[theverant]

My problem seems to be speed related, though. Have you tried moving the effector very slowly to see if you get the same doming? If I run between the posts at different speeds I get different results. F100 = no doming vs F3500 I get massive doming. So, it seems my problem is resistance/inertia, rather than the mechanics causing the effector to follow a certain path consistently.

Clear. As. Mud.

On the plus side it sounds like your machine is mechanically corrected to an insane tolerance. Wanna come to my house and do mine?

[edward]

I haven't noticed speed related issues. To that, I have mag arms which should have all but eliminated joint resistance/friction. If I lightly touch my dial indicator to the surface and issue a G2 or G3 command at 100mm radius I can watch the change. F100 or F2000+ provides the same results. I should note that using a dial indicator on a moving machine is not a good practice as it can easily damage them. Mine is <$20 so I'm not worried.

I can't recall seeing any explicit mentions of total bed variance here on the forums. I'm hoping that someone else might be able to comment for comparison if they've mic'd their entire bed.

On the plus side it sounds like your machine is mechanically corrected to an insane tolerance. Wanna come to my house and do mine?

I just noticed your from Nova Scotia. Never been there...so...you paying travel expenses?

[edward]

Well my suspicions have been mostly confirmed. I have been scouring Delta related forums this weekend and found more than one post indicating that setting the DELTA_DIAGONAL_ROD length to anything other than the actual, measured, dimension will cause scaling issues in the X/Y cartesian axes. So, I've reverted those changes and I'm back to square one.

I also noticed that when touching off at the perimeters of the build surface (I'm using a 100mm radius circle for the calibration points) the effector was tilted toward the center. An additional find was a comment in the Google Groups Delta list that comments on this exact issue - it relates to the pairs of arms being exactly parallel. What do you know? The ball joints on the effector that I printed is 2mm farther apart than the ball joints on the carriage.

I'll continue to update as things progress...

[theverant]

Sounds like you are up to your eyeballs - good R&D though - will you reprint the carriage arm mount? Was that the way it was by design, or did you have scaling issues? Have others using the design not seen problems?

I'd be happy to pay your travel expenses - you fit into a FexEx medium box, right?