Unsolved Mystery. Weird Z0 behavior around build perimeter.

[KAS]

The tricky part is verifying the left/right lean of the tower is square. Hard to do without a digital angle finder.

[Jimustanguitar]

For the record, I've successfully used RollieRowland's calibration script, and it got me between +.01 and -.01mm with no shims or other correction. I highly recommend using his HTML calibration page (and eventually his .exe or even a pull into the Repetier firmware itself) if you're having trouble but don't want to upgrade to a ZProbe or FSR's. It will require a dial indicator and a way to mount it, so know that. That turned out to be the most accessible to me, so that's what I did.

http://forum.seemecnc.com/viewtopic.php?f=36&t=8698

[d1rron]

I might try that to calibrate with a little more precision. So far I seem to be having better luck. I'm printing a ~3 inch diameter Saturn V model with pretty decent success. I do want to get an angle finder to double check tower lean, if for nothing else than piece of mind. I'll probably also get a dial indicator, since that seems pretty handy from what I've seen. FSRs might be unnecessary for now, but I'm also always interested in further automating calibration. Lol Thanks for everyone's input, between this thread and the other questions I've had help with and threads I've read in this forum you've all really helped me out. Really glad I went with my Max V2, great company and great community!

[jd52wtf]

Has anyone heard anything from the Support staff at SeeMe? To me this looks like an ongoing issue with the design. This appears to be a geometry issue in how the vertical towers are mounted in relation to each other in addition to the lack of adequate adjustments.

Doesn't that make this a faulty product of sorts?

As I am a mechanical designer I "can" design a bolt in solution to this problem which would include the ability to properly align/adjust the uprights for this unit. Would need to modify the top assembly as well. Not sure I should have to redesign a product that isn't living up to it's billing.

Not overly excited about these issues as I am a new customer and have found some serious issues after the fact. If anyone has some feedback from support please share.

Thank you.

[KAS]

This typically is corrected with mechanical calibration. SeeMeCNC is aware, but the Rostock isn't the only delta printer with this issue. The procedure in the manual has you squaring the towers to the glass, although that only accounts for one angle(front/back). Most don't verify the left/right lean is square until they notice it's an issue. This "product" is a kit made out of melamine and aluminium extrusion which also relies somewhat on the user ability to go above and beyond in regards to setup and calibration.

I'm sure you can build some type of adjustments, but how would you measure it?

Couple pointers that have helped me: Use a depth micrometer on your Onyx to verify if the PCB/glass is the same consistent height from the top plate of the base section; at every point possible. Use a digital angle finder to square your towers left/right. I went through three carpenter squares before finding one that was actually square. After using a digital angle finder, found out that my "eye squaring" ability was not good enough.

This is a good read about Delta calibration and the effects of minor differences.
http://boim.com/misc/CalDoc/Calibration.html

And this is probably the Delta Kinematics bible.
http://www.reprap.org/mediawiki/images/ ... tics_3.pdf

[geneb]

I actually recommend that you square against the base if there's any warp to the Onyx bed.

g.

[geneb]

There doesn't need to be much movement for adjustment. If the base is built square and the top set fully on to the tops of the towers, there should be little if any misalignment. Make sure that the towers are seated on the bottom stop screws and the top stop screws are seated fully on the top of the tower.

Don't modify the structure - there's simply no reason to.

You might want to make sure that the towers are straight and true. UPS could dent a depleted uranium box.

g.

[edit: it's annoying when people delete the message I'm replying to. ]

[KAS]

Too funny, I was just replying as well.


gotcha, I delete the quote

[jd52wtf]

Sorry for deleting as you were commenting. I hate it when that happens as well. I deleted it because I felt that I might have been a bit hasty in my comments about the platform.

After this evening I now realize that something is most certainly out of sorts.

[img]http://protech3d.com/wp-content/uploads/2015/09/Y-Z.jpg[/img]
[img]http://protech3d.com/wp-content/uploads/2015/09/X-Z.jpg[/img]
[img]http://protech3d.com/wp-content/uploads/2015/09/X-Y.jpg[/img]

I know this isn't the most accurate way to measure this however it is just fine for the purpose of this post.

As you can see from the images that the distance between the X and Y upright is about 1mm wider than the other two distances. Interestingly enough the area where I am seeing the most "saddle" issue is between these two posts. There is literally no additional adjustments that can be physically made to this platform to eliminate this error. I also measured these distances at the top of the structure and they stayed the same. This leads me to believe the the towers are are straight, parallel, and true but are simply too far apart.

Further investigation leads me to believe this may actually be a rotational issue with the placement of the uprights. I was able to shim the rod holders on the X and Y carriage to rotate them into the center slightly, to account for the 1mm difference in distance between the two towers. While the issue did not completely disappear it did improve by about 1/2 the total error.

As of now I am investigating ways to account for these issues in the firmware. I believe modifying the DELTA_ALPHA settings will help to eliminate this issue but I need to look into it more.

As always any ideas, comments, or constructive criticisms are welcome.

[jd52wtf]

So this is still a pretty large issue. Would feel a lot better to hear something from support on this. This is a tolerance issue with the laser cut parts. Over that distance 1mm is quite substantial.

[bot]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

[forrie]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

I still haven't gotten around to building my max metal (because I couldn't do a good enough full bed print to make the parts, even with pilots awesome smoothie calibration), but the best thing I have done to date is install the longer 325mm Tricklaser arms on my Max2. They are magic! Its like a whole new printer. I can now print right out to the bulldog clips with ease! I'm printing the max metal smoothie enclosure as I type.

[ccavanaugh]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

Improved rigidity may have helped a bit, but the real gain was being able to start with a very square and equi-distance frame. I did find that some inconsistencies in the molded parts were contributing to some additional error and I had to effectively latterly shift the offsets of one set of arms at the truck to compensate for inaccuracies in the molded parts. Use of the trick laser arms made this easy, but it would be relatively easy to do with the stock arms a well. I mostly likely contributed to those inaccuracies by poor sanding and fitting of the original joints.

Once I shifted the arms, printed part angularity errors went away completely, and I have only a small region on the bed where the initial layer thickness is not consistent by about .002" which I think is more a function of glass and bed flatness. I am using no error mapping what so ever on a smoothie board and the result should be the same with a Rambo as well.

It's a relatively simple process that requires printing a calibration piece and use of a 30/60 drafting triangle. If anyone is interested in the process I used, I can do a write up.

[enggmaug]

As you can see from the images that the distance between the X and Y upright is about 1mm wider than the other two distances.

Hello, the proxy at work won't let me see your pics, but I'll have a look at your post at home later on.
My investigations on my RMAX V1 led me to the exact same conclusion.
Do you have a V1, or V2 ?

My RMax is currently dismantled, and I am waiting to find a proper solution to this before building it again. I want things to be as perfect as they could.

People are often refering to a digital angle finder here... would a cheap "2in1" as we can find on Ebay be sufficient for this task ? What angle finders do you guys own ?

[bot]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

Improved rigidity may have helped a bit, but the real gain was being able to start with a very square and equi-distance frame. I did find that some inconsistencies in the molded parts were contributing to some additional error and I had to effectively latterly shift the offsets of one set of arms at the truck to compensate for inaccuracies in the molded parts. Use of the trick laser arms made this easy, but it would be relatively easy to do with the stock arms a well. I mostly likely contributed to those inaccuracies by poor sanding and fitting of the original joints.

Once I shifted the arms, printed part angularity errors went away completely, and I have only a small region on the bed where the initial layer thickness is not consistent by about .002" which I think is more a function of glass and bed flatness. I am using no error mapping what so ever on a smoothie board and the result should be the same with a Rambo as well.

It's a relatively simple process that requires printing a calibration piece and use of a 30/60 drafting triangle. If anyone is interested in the process I used, I can do a write up.

Arm lengths, like forrie mentioned, are definitely a good start at improving the work envelope. You make some great points, too. I think a write-up would definitely be welcome about your process.

.002" is quite a small deviation. How are you measuring that? I have been able to get my bed to within .009" measured by doing a constant outward spiral from the center to the outer edges. This doesn't just measure a small set of points, but a constant sweep where the dial indicator can be monitored the whole time. .009" of full indicator movement (about +.005/-004). I would love to get this to be within .002". The thing is, when I measure a set of discrete probe points, they are also all within .002", it's just not revealing the whole story.

[ccavanaugh]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

Improved rigidity may have helped a bit, but the real gain was being able to start with a very square and equi-distance frame. I did find that some inconsistencies in the molded parts were contributing to some additional error and I had to effectively latterly shift the offsets of one set of arms at the truck to compensate for inaccuracies in the molded parts. Use of the trick laser arms made this easy, but it would be relatively easy to do with the stock arms a well. I mostly likely contributed to those inaccuracies by poor sanding and fitting of the original joints.

Once I shifted the arms, printed part angularity errors went away completely, and I have only a small region on the bed where the initial layer thickness is not consistent by about .002" which I think is more a function of glass and bed flatness. I am using no error mapping what so ever on a smoothie board and the result should be the same with a Rambo as well.

It's a relatively simple process that requires printing a calibration piece and use of a 30/60 drafting triangle. If anyone is interested in the process I used, I can do a write up.

Arm lengths, like forrie mentioned, are definitely a good start at improving the work envelope. You make some great points, too. I think a write-up would definitely be welcome about your process.

.002" is quite a small deviation. How are you measuring that? I have been able to get my bed to within .009" measured by doing a constant outward spiral from the center to the outer edges. This doesn't just measure a small set of points, but a constant sweep where the dial indicator can be monitored the whole time. .009" of full indicator movement (about +.005/-004). I would love to get this to be within .002". The thing is, when I measure a set of discrete probe points, they are also all within .002", it's just not revealing the whole story.

I was printing a bed calibration part, letting it cool, and then carefully peeling it off the bed and measuring the variation in height with calipers. Not super scientific, but easily done without use of a probe and it creates a continuous layer that can be measured in many places.

This is the write up

DeltaCalibrationGuide.pdf

Delta Calibration Guide

(1005 KiB) Downloaded 937 times

and the calibration object can be found here https://github.com/ccavanaugh/Max-Metal ... alibration. Let me know if there needs to be any clarification to be done. It was fun doing the write up and I hope others can benefit from it.

[626Pilot]

I think the issues are much much deeper than anything to do with the lasercut parts. Ask the folks who went with the max metal frame if the rigid frame has helped alleviate this problem.

YES. I'm buying two more of them. Way easier/quicker to put together. Part count is nothing compared to a Rostock MAX 1 or 2, and it's designed to be self-aligning. So much less fiddly stuff to mess with. My calibration system finds delta radius offsets near zero, which suggests to me that the frame is close to perfect. (Way closer than my old Max v1.)

[d1rron]

Well I don't know what happened.

I checked my tower alignment as bet I could and made sure everything was tight, upgraded a few parts, and everything seemed to be working well for the past 4-5 months. I didn't seem to have the bed level issue between towers anymore. I went to print a large part a few days ago and once again I'm having issues between the towers. I sent the nozzle to the bed while trying to recalibrate, and decided to send some travel commands while it was just barely dragging the paper. The nozzle got closer to the bed at first and then moved away until it was about .3mm above the glass at the edge.

One thing I did notice is that I never updated my Printer Radius in the firmware after installing Trick Trucks -- I'm also using TL 325mm CF delta arms, but I adjusted that in the FW the day that I installed them. I made the printer radius change and recalibrated the 3 towers and center, but the spaces between are still way off, and especially near the edge of the plate. I've been reading through all of the various answers on this and my head is spinning. I'm considering disassembling the whole thing and reassembling, but I'm not sure if that'd be the best idea.

I've been strongly considering a Max Metal frame, but it's a bit more than I can afford to spend on this right now. I was looking at OpenBuilds extrusions and trying to figure out if I could hobble something together with their V-Slot stuff and maybe some printed corner pieces or something. I'm just not sure what to do. I had planned to do a couple of rather large prints for family members, but now I'm not sure how soon I'll be able to, if ever.

I'm sure I screwed up somewhere, but I just don't know what to do at this point.

[nebbian]

You might have introduced frame twist, that causes weird Z behavior between towers. Check with a large square, ensure that all towers are parallel and not twisted.

[bob64]

Not sure if this applies to you... but check if your snowflake/top part of the melamine base is flat... Mine is domed and it was throwing off my calibrations for the longest time.

[d1rron]

Thanks for the replies. I'm going to check towers for square and then perhaps that bed piece. Would having all of the Melamine piece cut from acrylic make it more rigid and hold it's square better? I really wish I could afford to order a Max Metal frame. Lol

[Jimustanguitar]

Would having all of the Melamine piece cut from acrylic make it more rigid and hold it's square better?

I believe that the original IndieGoGo kits that were acrylic were prone to frustration. Since the acrylic doesn't have any give or squish, it was hard to keep the fasteners from loosening. It's also nasty to cut on a laser table, so the people that do, do it often wish that they didn't.

[d1rron]

Thanks for the info. I'd just seen a couple in Google Images that were Acrylic and thought it might be a good way to go, but I'll stick with the Melamine until I can get a metal frame kit. Is there any practical way to make the melamine frame more rigid and hold square (assuming that's what my issue is)?

I'm heavily considering the Max Metal frame, though I know it's not a drop-in kit. I was really looking at OpenBuilds V-rail to just put something together myself, but it seems to be 4/5 the size of the 1" Rostock rails so I assume my Trick Trucks wouldn't work and I'd still have to figure out some way to connect the corners and vertical rails and then figure out all of my new dimensions for the software which sounds kind of intimidating. Lol

I wonder if carefully disassembling and reassembling might resolve it. I'm still a little unsure whether I properly aligned my towers. I assumed, since they were connected by those 4 screws at the bottom, that loosening and tightening different screws slightly would lean the tower in the intended direction until it's square and tight. Hopefully I was right in that assumption.

I think I need to pick up a digital angle finder. I have framing squares and speed squares, but it seems like it'd just be easier with an angle finder.

Sorry for sort of rambling, I'm just overwhelmed with this thing right now. Lol

[Xenocrates]

I personally like the idea of doing something using the 80/20 Inc. 30 degree profiles (Only available in 40MM sadly, but you can get/make/print alignment brackets to connect the 40MM frame to the 1IN towers). Use pairs of them at the corners that are seperated to make a superstructure that is in the right shape, which can also hold an enclosure, and then go ahead and use another pair to hold the runner between the sides of the triangle that the tower connects to. Then for the bed, just use runners out from the sides that connect to either 2 more, or a sixty degree profile. Very high rigidity, built in enclosure, and commodity parts all around.

That being said, warming the Melamine in a controlled fashion, say with a space heater in a box with it, while it is between two flat objects and under a decently large amount of force. Playing with tower alignment has helped others as well. Acrylic might be stiffer when cool, but once warmed sufficiently (Such as with a heated chamber, or very high bed temps), it can bend pretty well. I have taken the bend out of the corner pieces used on the max with nothing but a heat gun and gravity (They bent under their own weight). I would not be a big fan of it in high temp environments, especially not as a structural element.

[d1rron]

All good points, Thanks! I still haven't resolved it, but I'm leaning towards selling my GoPro and ordering a Max Metal frame (as I mentioned in that thread about making a black one) just to eliminate the tower lean variable as much as possible. If I still have issues, at least it will be very unlikely to be related to tower lean. Lol