PrintBed leveling - prusa

Hi Everyone,

First time leveling my print bed (w/ Heat) and I'm having some issues. I'm following these instructions:

[reprap.org]

I just can't seem to get this thing level. Usually when I move to the secon diagonal I begin to have problems. The top of my bed is glass, so should be pretty darn flat. My springs came with a kit, they looked about the size of those that come inside pens, didn't look to incredibly flimsy. I have them compressed within 4mm of full compression - brings it about 3 mm above the pully.

Does anyone use an alternate technique that works better - or is faster? Here is my pattern:

FL->BR->FL, BL->FR->BL, FL->BR->FL then repeat if necessary. I leave FR loose (F=Front, B=Back, L=Left, R=Right)

Any suggestions? I'm so close to a testing point - but so frustrated I'm stuck on such a mundane (but important) item.

Also - anyone recommend against that type of spring and recommend somthing different?

Thanks in advance!

Ian

You could try this method: [hydraraptor.blogspot.co.uk]

---

[www.hydraraptor.blogspot.com]

If the adjusting screws are not at the same position as the corners of the glass. Do the checking inline with the adjusting screws, that way the bed position won't change when the other pair is adjusted. If you move the rear up, the front edge will move down, so measure inline with the adjusting screws.

You have to get the nozzle level with the bed, doens't matter if it's perfectly level with he table or whatever.

Try bringing the nozzle close to the bed on one side like 0,5cm or something.
(Try to make sure it's more or less level pretty level on both x ends)
Move the nozzle back and forth on x and adjust your bed screws up or down until it gets closer and closer to level with your nozzle.
Do it for the y back and forth kind of interchangeably until you can go all 4 corners and the nozzle looks perfectly the same height.
(Using a piece of paper or something can help to get it the same)


When you do a print, watch carefully as the plastic goes onto your surface, and slightly adjust your z end-stop lowering the nozzle until it's how you want sticking to the bed.

Thanks for all the feedback.

nophead - thanks, I might try that once all else fails, trying not to spend more money.

stephenrc - not really following your recommendations. My adjusting screws are certainly not in the same position as the corners of the glass, perhaps that is the case for some Prusa variation, but not with mine. I have a much smaller under-base that is attached to Y axis that the adjusting screws clasp to. Perhaps if you could add some detail I might better understand if your suggestion applies to me.

dustynus - I'm not focusing on printer to table leveling, just print head as you say. You say use a piece of paper which I have seen recommended many places, but I'm a bit confused on your dimensions 0.5 cm. That is pretty big, did you mean .5mm? Even so, that is pretty thick for paper. I'm using two sheets of standard US paper, which is about 0.2mm. Also - in your suggestion when you are going back and forth on X-axis - what is your Y position during this process? In the center? Or font/back? Same question for when you check Y-Axis, are you at center position with Y?

General question for the rest of you. I'm beginning to think that I will be unable to make my print bed level with the head. Is there some well known construction issue that could cause that to happen? Anything specific I should re-check on the structure itself?

At my wits end with this. I Always end up with one corner "floating" because the other three positions set correctly results in the final corner being forced into a height that is too far away from the print head (so, I get 2 opposite corners at 0.2mm, a third at 0.2mm, and the last corner is like 0.5mm).

Any ideas/recommendations?

Quote

nophead - thanks, I might try that once all else fails, trying not to spend more money.

You don't need to buy a dial gauge. You can use the same method with some form of feeler gauge, such as paper, or as I prefer: a round rod.

If the Y bars are not parallel you wont be able to level it, so that is probably your problem. This is one of the many issues I addressed with Mendel90.

If you replace the Prusa bar clamps with the Sell's Mendel version the bars are clamped against the studding, so are aligned more accurately. You then rely on the feet all being exactly the same length and it standing on a flat surface.

---

[www.hydraraptor.blogspot.com]

The clamps at the bottom for the Z axis were not aligned exactly with the Y-axis threads - I adjusted them, so X-axis bars should now be parallel. However, still have the same issue - didn't seem to help.

Looks like you are going by the Mendel90 instructions. Although I use the same bed on my Prusa, I'm not sure I understand what supports the bed in the Mendel90 instructions. In my design I have the M3 bolts going through the particle board (PCB is attached to board, glass is clipped onto PC. On the other side of the board, I have lock nuts down to the base of the board to ensure the shaft remains perpendicular to board. Then I have springs on each followed by a washer in front of the Y-track base. On the other side of the base I have a washer and another lock-nut: so - Bolt->board->Lock Nut->Spring->washer->Y-track base->washer->Lock nut

So in your instructions - I don't see springs involved - the bed construction is a bit confusing - what allows compression of the bed in case of a head crash? The instructions seem to imply that two of the bolts are "fixed". Anyway, I'm just confused by the directions since it seems clearly different than what I have in place now.

No I don't use springs because they let the bed wobble. If the nozzle hits the bed it just pushes the lead nuts out of the X ends, so the maximum force is the weight of the X axis. That will rip tape but not damage the nozzle.

The adjustment procedure can be used however the bed is mounted. On Mendel90 (and also on my hacked Prusa) the PCB is supported above the carriage with mail female M3 hex pillars. Those are tapped into the holes in the carriage so rotating the pillar winds it up or down and the screw through the PCB locks it in place when tightened.

I don't understand your first sentence. It is the Y bars that prevent the bed being leveled when they are not exactly parallel and co-planar. The Prusa bar clamps don't press the bars together so there is an offset that can vary from clamp to clamp depending on how much it is tightened. The Mendel clamps do press the bars to the studding, so they are more accurately placed. But there is nothing to ensure the frame bars are parallel apart from the feet being the same size and being placed on a flat surface. Considering we want the bed level to about 0.05mm it is easy to see why this is hard to achieve with bars suspended in mid air on other bars suspended by feet. The holes in the feet and the bar clamps are all clearance holes, so there is no accurate alignment. If the bars are not parallel the bed changes its inclination as the Y axis moves, so you can then only level three corners.

Edited 1 time(s). Last edit at 07/02/2012 12:47PM by nophead.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
If the Y bars are not parallel you wont be able to
level it, so that is probably your problem. This
is one of the many issues I addressed with
Mendel90.

I should have read your reply more carefully. I now realize you said Y bars and for some reason I was thinking X.
When you say parallel, do you mean parallel to each-other (top down view)? Or do you mean parallel to the resting surface? The first is hard to screw up because the Y-axis would not slide freely (which it does). The latter I don't see how you can screw up because it certainly seems like those clamps keep the bars locked into a certain (parallel to the ground) orientation - and the threaded rods they clamp to are fixed in place by the corner pieces - and all corner pieces have been measured multiple times now and the overall structure is 100% aligned. I'll read your posts again more carefully and see if I better understand now, will follow up with questions if there is something I'm not following. Thanks for your help.

Edited 1 time(s). Last edit at 07/02/2012 02:32PM by ifrisbie.

Yes they must be the same distance apart at each end or the carriage would not slide. Whether they are parallel to ground doesn't matter but they must lie in the same plane and that plane should be orthogonal to X and Z. This is really difficult to get right with the Prusa because it all relies on measurements and the placement of studding in holes that have clearance far bigger than the 0.05mm accuracy required to get the first layer right with say 0.3mm layers.

---

[www.hydraraptor.blogspot.com]

I'd be happy if I could get the corner differences down to 0.2mm. I've checked the Y-bar height above the round (placed it on glass to ensure it was flat. the front two feet are the same, back to feet are the same - front and back differ by about 0.08mm. Seems to me that this relatively small difference couldn't cause a 0.4mm difference between the corners of the bed (3 corners being even - 1 corner being off by 0.4mm). I do now understand all of what you have outlined. The Sells mendal bar clamps do seem like a much better approach - but I'm determined to make this prusa work, for all else so I can print out better bar clamps!

BTW - when you talk about 0.05mm accuracy, are you suggesting that the corners of the bed need to be within 0.05mm (for me 1/2 a paper width) of each other? Yikes, that seems like it might be hard to achieve.

Is there any chance that my sheet of glass has a bend to it? Can glass be bent by my clips by about 0.4mm (they are the medium size metallic clips) when attached to the PCB (which is surely not as flat) which is attached to the board that came with my vitamins kit (which certainly is not flat either). Just wondering how much pressure you can place on it before glass will break - any possibility this is happenning in my case?

0.05mm is enough to notice with 0.3mm layers, which aren't particularly thin these days. 0.05mm too high will make the first layer gappy and not stick well. 0.05mm too low will make a ridge around the bottom and cause the plastic to ball up around the nozzle.

You can check the glass is flat by placing a straight edge across the diagonals. In my experience it is much flatter than measurements at the corner relative to the nozzle would make you believe.

---

[www.hydraraptor.blogspot.com]

Well, I've changed a number of things but they haven't panned out for me:

1. I replaced the particle board platform with what came with my kit to a hard-plastic version - which I verified to be flat (the board certainly was not flat). I show how I have the glass and heated PCB mounted to it in pic 1327. Verified that with flat edge, glass is flat on diagonals.
2. I went from a 4 point spring-based mounting to a 3-point spring-based mounting (pic 1329 shows how it is connected - 6-32 Bolt->top->6-32 washer->6-32 nylon lock nut->spring->smaller washer->bottom->6-32 washer->6-32 nylon lock nut)
3. I verified the height of the Y-rails from the ground - which was verified to be flat at all 4 corners (see pic 1326, basically made a block and ensured that it barely fit at all 4 ends)

Pic 1322 shows measurement at the 3rd post with a business card that is 0.33mm thick, barely fits. Pic 1325 shows the HOME position of Y where you can fit 3 business cards under the extruder head (> 1.0mm). On the other end of Y (still X-HOME) at about 150mm up you can no longer fit the 0.33mm car under it. This is the situation when all 3 points (in isosceles triangle pattern) are exactly 0.33mm from the surface (X-MAX/Y-HOME, X-MAX/Y-MAX, X-HOME, Y-MAX/2)

Really at a loss of what to check. It did seem logical that this is the result of non-parallel Y-bars - but they are parallel (both to each-other and all four ends lay in the same plane which is parallel to the surface). There must be another possible cause of this misalignment, or I'm simply not understanding something very basic.

Perhaps the bars are bent.

---

[www.hydraraptor.blogspot.com]

Unfortunately thats a bingo as Hans Landa would say. One is fairly obviously warped in the middle. The other has a very slight bend, but probably a forgiving one. Very frustrating.

Any chance this could have been caused by the weight of the bed over time? I probably did have it in a centered position for up to 4 months.

Anyone know a reputable place to get smooth rods that are actually straight?

I did a bit of an experiment and rotated the rods thinking that since the bend is only on one of the rod's axis then if I rotate it by 90 deg I will eliminate the vertical anomaly. Well, I did reduce it a little, but there is still a general downslope on one side and I only moved the differential from 1mm down to about 0.6mm front right side to back. So back to the drawing board on isolating the problem - guess I have two issues compounding each other - the other one must be in the frame somewhere - or simply a variability in the clamps.

Back to the drawing board - thanks for your help.

I'm a little confused what your trying to fix with the bars, if the bed can slide smoothly without getting stuck or something then the bars should be ok.

Even if the bars are not perfectly level it's fine since you can compensate for that error by lifting and leveling your actual print surface independent from what your actual bearings are attached to.
If your piece of glass is definitely flat then you need to adjust the corners raising or lowering until it's level consistently with the nozzle tip.
The 0,5cm I mentioned before was just arbitrary I just mean bring it close enough to the bed, so you can slide the carriage + bed back and forth over the whole XY area, and just using your eyes you can get it ''in the ballpark'' over the surface then slowly lower and check it more carefully sliding the paper

Hope this helps in some way

Quote

Even if the bars are not perfectly level it's fine since you can compensate for that error by lifting and leveling your actual print surface independent from what your actual bearings are attached to.

No, you can't compensate for the bars not lying in a plane by adjusting the bed. The bed rotates around a line through the Y axis as it moves backwards and forwards if one bar is inclined relative to a another. That means that even if the bed is a rigid flat sheet it will appear to be curved relative to the nozzle.

There is nothing in the Prusa design that ensure the bars are in the same plane to the required accuracy that the bed needs to be level to.

Also the sag in the bars due to the weight of the bed and the extruder make the bed appear to have a saddle shape.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Also the sag in the bars due to the weight of the bed and the extruder make the bed appear to have a saddle shape.

Is this sag guaranteed to happen over time? Does it mean I'm using poor quality rods? Or do they get replaced often?

I'm waiting for parts but my plan is to solve this issue by doing the following:

1. Get replacement rods - one rod has a differential greater than 1mm, badly warped
2. Use Sell's mendel Y-AXIS clamps instead of the standard clamps - to ensure that the rods hang equa-distant from each end.
3. Go to a 3-point system to simplify leveling (already did this).

Once I have all the parts I'll re-post to document how it went. Thanks for all the feedback!

They don't sag over time. They just sag slightly with the weight of the extruder/bed simply because they are not infinitely stiff. They would spring back as soon as the weight was removed.

---

[www.hydraraptor.blogspot.com]

Oh sorry I wasn't being clear, I was trying to say the bed/bars don't necessarily have to be level with earth, only bed and nozzle have to be parallel. I wasn't trying to say it can't be on a plane..

If the whole bed is able to move smoothly back and forth then the bars must be parallel. But I guess you actually have warped bars ?

Well, I have an update. I have my new bars - they are not warped like the last two were. I don't have the Sell's Mendel bar clamps, still using my original ones.

By enlarge I still have the same issue.

Quote
dustynus
If the whole bed is able to move smoothly back and forth then the bars must be parallel.

Certainly not the case for me, I have a very freely moving bed yet the left side (which has two of the spring mounts) are even with the nozzle at the home position and MAX position - but on the right side the HOME end is still about 0.6-0.7mm off from the MAX end.

Ok, so what I'm assuming now is the bars must not be parallel. However, with the small amount of difference I'm looking at here I'm not quite sure how to detect and then address it. Every mechanism I can think of to measure the height above ground of each bar is not precise enough (or repeatable). What I really need is something that is an exact fit under the bar, but that is hard was well - then, how do I adjust the ends? When I have the sells mendel clamps that might help, but I'm concerned that if my problem is that one of my threaded cross bars is warped, it would cause the same issue - but I couldn't address it since its fixed.

Anyone have any suggestions on measuring and adjusting the clamps?

I would measure the height to the base with digital callipers by standing one jaw on the ground and using the other jaw for measuring inside to measure the height under the bar. It will read too low by the width of the bottom jaw but you are only interested in relative distances.

If the studding is bent perhaps rotating it 90 degrees will fix it.

Is the surface you stand it on definitely flat? I recently stood a machine on a cheap laminate kitchen work surface and found it wobbled. I had to pack one corner with two thicknesses of card. This was a machine with a flat MDF base, a Prusa would probably twist.

Perhaps you can fix it by putting shims under the feet to twist one triangle relative to the other.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
I would measure the height to the base with digital callipers by standing one jaw on the ground and using the other jaw for measuring inside to measure the height under the bar.

This is what I have been doing but you must ensure that the caliper is perpendicular with 2-axis. To get it just right for this to measure accurately under 1.0mm:
* rock the caliper on its head length until it feels flush - hold it steady
* Use level to approximate the other axis (caliper head width)

If you don't do this, your measurements will be off because the caliper is angled and you'll be measuring the wrong distance. I find the whole process very error prone and if I measure the same spot 3 time, I'll get three different answers +- 0.4mm - within the error that I'm trying to correct. So unfortunately thats not working for me

Quote
nophead
Is the surface you stand it on definitely flat?

It sits on a piece of hard wood (solid) shelving that I pulled from an old book shelf. I'm sure its not perfect, but my plate of glass sits pretty flat against it so I never questioned it. I think its good.

Your suggestions are good, just need to figure out how to accurately measure the corner rod heights so I can make the adjustments in the correct direction.

My callipers have quite a large flat area on the head so it is fairly easy to stand it orthogonal to the base.

Another way to do it would be with a dial indicator attached to a heavy base. There are digital height gauges but they seem quite expensive. Cheaper to screw a calliper to a block of metal.

You can see now why I designed the Mendel90. The whole purpose of the frame is to hold the bars precisely in orthogonal planes and resist the dynamic force of the carriages moving. The triangular prism frame is not good at either function.

The Mendel90 bars are stood off flat sheets by printed parts, so they are coplanar to the accuracy of the repeatability of the parts. Since they are printed on the same machine together they should be pretty much the same.

---

[www.hydraraptor.blogspot.com]

another possibility might be a very large bolt with two nuts (bolt taller than the bar height). Could lock the lower nut to the bottom of the screw with some glue and use the other nut to adjust to the bottom of one of the 4 ends - then lock it in place. This would give me a flat and relatively orthogonal base to measure the other ends.