Bed Calibration Issue

I am having some trouble calibrating my machine. I have a Kossel Mini with a Duet board and IR auto level probe running DC-42 firmware. Heated bed and test done both at heat and not.

I can calibrate it till I am blue in the face using 13 points and my results are reproducible. But when it does a auto level something just it not working. It will work fine, then I remove bed an clean it and it won't calibrate again. I spent the next few hours fussing with it and then I get maybe a print out of it. It is worse with the glue on the glass than without but nothing sticks without it. I have a hard enough time getting it to stick with it.

I get a thin section and a semi thick section and the rest is around what it should be. The thin section is around P1 (Outer edge between Z & Y Tower). The thicker is opposite. My probe offset based on the data cluster below is 0.67.

My results from point testing the last time are:
Point Nozzel Probe Difference
P0 1.00 1.65 0.65 Z Outside
P1 0.55 1.20 0.65 Z-Y Outside
P2 0.95 1.60 0.65 Y Outside
P3 1.10 1.75 0.65 Y-X Outside
P4 1.10 1.75 0.65 X Outside
P5 0.60 1.25 0.65 X-Z Outside
P6 0.65 1.40 0.70 Z Inside
P7 0.80 1.50 0.70 Z-Y Inside
P8 1.10 1.95 0.65 Y Inside
P9 1.30 1.95 0.65 Y-X Inside
P10 1.20 1.90 0.70 X Inside
P11 0.90 1.60 0.70 X-Z Inside
P12 1.05 1.70 0.65 Center

When I run auto calibrate after these points my adjustments to the config.g was : R - 104.65, H - 248.51 Endstops: X -0.21 Y 0.31 Z -0.10 (Without Glue)
Last run it auto calibrated before printing: R - 104.64, H - 248.41 (With Glue).

I am totally at a loss as to what to do next. Nothing I do seems to be working. Any help would welcomed.

Two things:

1. Please explain what you mean by "it won't calibrate again" after removing the bed for cleaning.

2. If you have one high spot and one low slot opposite that, and you have calibrated 6 factors, that suggests to me that the effector is tilting as it moves between those points, and this tilt is affecting the relative heights of the sensor and the nozzle. Check this by measuring the sensor trigger height at both points.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

1. Please explain what you mean by "it won't calibrate again" after removing the bed for cleaning.

I mean I can get it working, run a print with it level and smooth on the bottom. Then I remove the bed to clean the glass, put it back and next auto calibrate create the thin / thick areas again. After spending a few hours testing and adjusting the config.g defaults I can get it going again till I clean it again.

2. If you have one high spot and one low slot opposite that, and you have calibrated 6 factors, that suggests to me that the effector is tilting as it moves between those points, and this tilt is affecting the relative heights of the sensor and the nozzle. Check this by measuring the sensor trigger height at both points.

Those points are included on the list given. Outside & Inside Z-Y - thin; Outside and inside X-Z - thick; outside and inside X-Y, X, Y - normal. All other points are a difference in between. Sorry the list is hard to read. It removed my formatting.

When the table is removed and placed back on the measurements are all different, but still relatively close to those.

I ended up changing points in my bed.g file to include the 'H' parameter. My bed.g no looks like this:

;13 Point Calibration
G30 P0 X0.00 Y85.00 Z-99999 H-0.05
G30 P1 X73.60 Y42.50 Z-99999 H-0.50
G30 P2 X73.60 Y-42.50 Z-99999 H-0.10
G30 P3 X-1.00 Y-84.00 Z-99999 H0.05
G30 P4 X-73.60 Y-42.5 Z-99999 H0
G30 P5 X-73.60 Y42.50 Z-99999 H0.05
G30 P6 X0.00 Y42.50 Z-99999 H-0.40
G30 P7 X36.80 Y21.25 Z-99999 H-0.25
G30 P8 X36.80 Y-21.25 Z-99999 H0
G30 P9 X0.00 Y-42.50 Z-99999 H0.20
G30 P10 X-36.80 Y-21.25 Z-99999 H0.10
G30 P11 X-36.80 Y21.25 Z-99999 H0.50
G30 P12 X0 Y0 Z-99999 S6

Is there some way to physically change the machine to eliminate the H values? I cannot seem to figure out a way. My bed seems flat (using a straight edge and checking for a gap) but I also do not have any metric feelers that are narrow enough (0.05 from the table above) to test against.

Next Issue: Getting Prints to stick well.

I can get basic items to stick that do not have a lot of fill, but big flat circles etc I have trouble getting to stick well to the platform. The outside and inside rings stick, but when it starts the infill it gets small bubbles when it comes off the bed randomly, or when making a curved infill it starts to curl inwards, I would guess from shrinking while it cools. Depending on how bad, this causes a failed print since the second layer cannot be made over it.

Any suggestions on how to resolve that curling/sticking problem?


Thanks

1. Varying trigger height on a delta is usually caused by effector tilt varying with XY position. Various things can cause this:

- Diagonal rods in a pair not being quite the same length (measured between bearing centres)

- Diagonal rod spacing (measured between bearing centres) not being quite the same at both ends of a parallel pair of rods. Fix this by adding shims under the bearings.

- Carriages not quite aligned with the effector (ideally, both the carriage bearing pairss and the effector bearing pairs should be at intervals of 120 degrees). If using wheeled carriages, fix this by adding washers or shims between the carriages and the wheels.

- Play in the joints, coupled with the sideways force exerted by the Bowden tube changing direction depending on its XY position. You can use springs to preload the joints, and elastic to support the weight of the Bowden tube form the top of the frame.

2. There are several things you can do to improve first layer adhesion:

- Correct heated bed temperature. If you have a glass bed then 60C is about right for PLA with most bed surfaces (you get about 5C temperature drop across the glass)

- Bed surface. Some PLA filaments can be printed directly on bare glass primed with vinegar. Other popular surfaces include Kapton tape, hairspray, masking tape, BuildTak, PrintBite, and PVA glue. I now use PEI which is one of the best, but this is still difficult and expensive to obtain in Europe.

- A low first layer speed

If you are getting actual bubbles in the extruded filament, then perhaps you have bad filament.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
dc42
1. Varying trigger height on a delta is usually caused by effector tilt varying with XY position. Various things can cause this:

- Diagonal rods in a pair not being quite the same length (measured between bearing centres)

- Diagonal rod spacing (measured between bearing centres) not being quite the same at both ends of a parallel pair of rods. Fix this by adding shims under the bearings.

- Carriages not quite aligned with the effector (ideally, both the carriage bearing pairss and the effector bearing pairs should be at intervals of 120 degrees). If using wheeled carriages, fix this by adding washers or shims between the carriages and the wheels.

- Play in the joints, coupled with the sideways force exerted by the Bowden tube changing direction depending on its XY position. You can use springs to preload the joints, and elastic to support the weight of the Bowden tube form the top of the frame.

I have added bands to my rods to achieve those settings, was worse before. I will try to check out the rods and work forward. Is there any problem for now with leaving the H values in there to correct it?

Quote
dc42

2. There are several things you can do to improve first layer adhesion:

- Correct heated bed temperature. If you have a glass bed then 60C is about right for PLA with most bed surfaces (you get about 5C temperature drop across the glass)

- Bed surface. Some PLA filaments can be printed directly on bare glass primed with vinegar. Other popular surfaces include Kapton tape, hairspray, masking tape, BuildTak, PrintBite, and PVA glue. I now use PEI which is one of the best, but this is still difficult and expensive to obtain in Europe.

- A low first layer speed

If you are getting actual bubbles in the extruded filament, then perhaps you have bad filament.

I heat my bed to 65C for first layer, 60C for each after.

I have glass, was using Pritt Stick, tried elmers and a few others. Nothing quite perfect yet. Small prints do fine, just larger prints. I will have to look into PEI. Do you just you really thin tape to adhere it to your bed?

When I get it to work I run it at 50% normal. When it fails is at 100%.

Bubbles are not bubbles in the plastic but spaces between two lines where one line pealed up some and the next line came around and pushed it down somewhat. Makes for a ugly bottom if it does not fail.

The problem with relying on large H corrections is that the underlying effector tilt will probably cause other issues such as scaling errors.

I tried Pritt Stick briefly but had limited success. I found that some PLA filaments would adhere to plain glass primed with vinagar. For the ones that didn't, painting solvent pipe cement (neat or diluted with acetone) on the glass worked. But since I discovered PEI, that's what I use. I sprayed the underside (the less shiny side) with black heat resistant paint for compatibility with the IR sensor, cured the paint in an oven at 170C, then used a sheet of 3M 8153LE adhesive to attach it to the glass. You could attach it to the aluminium directly, but I prefer removable bed plates.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Quote
c3dprinting

Quote
dc42
1. Varying trigger height on a delta is usually caused by effector tilt varying with XY position. Various things can cause this:

- Diagonal rods in a pair not being quite the same length (measured between bearing centres)

- Diagonal rod spacing (measured between bearing centres) not being quite the same at both ends of a parallel pair of rods. Fix this by adding shims under the bearings.

- Carriages not quite aligned with the effector (ideally, both the carriage bearing pairss and the effector bearing pairs should be at intervals of 120 degrees). If using wheeled carriages, fix this by adding washers or shims between the carriages and the wheels.

- Play in the joints, coupled with the sideways force exerted by the Bowden tube changing direction depending on its XY position. You can use springs to preload the joints, and elastic to support the weight of the Bowden tube form the top of the frame.

I have added bands to my rods to achieve those settings, was worse before. I will try to check out the rods and work forward. Is there any problem for now with leaving the H values in there to correct it?

Ok, I checked the rods. Length is near identical on all rods (0.01). Each pair of rods - 2 of the sets where around .2 off from top and bottom measured across. I have corrected. I do not have a good way to measure the last item. I am going to run tests again with the H factors removed and see what happens.

I was using the H factors and it worked fine, until I added glue on the glass to hopefully help with adhesion. Now its way off again

Quote
c3dprinting

Quote
dc42

2. There are several things you can do to improve first layer adhesion:

- Correct heated bed temperature. If you have a glass bed then 60C is about right for PLA with most bed surfaces (you get about 5C temperature drop across the glass)

- Bed surface. Some PLA filaments can be printed directly on bare glass primed with vinegar. Other popular surfaces include Kapton tape, hairspray, masking tape, BuildTak, PrintBite, and PVA glue. I now use PEI which is one of the best, but this is still difficult and expensive to obtain in Europe.

- A low first layer speed

If you are getting actual bubbles in the extruded filament, then perhaps you have bad filament.

I heat my bed to 65C for first layer, 60C for each after.

I have glass, was using Pritt Stick, tried elmers and a few others. Nothing quite perfect yet. Small prints do fine, just larger prints. I will have to look into PEI. Do you just you really thin tape to adhere it to your bed?

When I get it to work I run it at 50% normal. When it fails is at 100%.

Bubbles are not bubbles in the plastic but spaces between two lines where one line pealed up some and the next line came around and pushed it down somewhat. Makes for a ugly bottom if it does not fail.

So far I am not having much luck with anything. I will keep at it. Maybe if I can fix the bed level without the H this problem will just go away.....

Well after testing that did not work. Without the H involved the result were the same as before. Is there anything else that could cause low and high spots?

Just for fun (and extreme frustration) I took a piece of wood painted black and removed the glass and black paper. It calibrated first time out with a .05 margin of error. Powered off, moved the board, powered up and calibrated again. Once again same very small margin of error. No H figures no nothing.

So.... My guess is either the IR probe is having trouble with the paper, glass or both. What other kinds of things can I use / replace the glass / paper with? I have an aluminum heated bed below.

What do you use on the back of you PEI? Paint or?

Quote
c3dprinting
Just for fun (and extreme frustration) I took a piece of wood painted black and removed the glass and black paper. It calibrated first time out with a .05 margin of error. Powered off, moved the board, powered up and calibrated again. Once again same very small margin of error. No H figures no nothing.

So.... My guess is either the IR probe is having trouble with the paper, glass or both. What other kinds of things can I use / replace the glass / paper with? I have an aluminum heated bed below.

What do you use on the back of you PEI? Paint or?

I guess it helps if I go back and read things again. I will work on purchasing some PEI, tape and paint.

Glue on top of the glass will affect the trigger height a little, so you may need to change the G31 Z parameter. However, if you have glue at all the probe points and not just some of them, the H corrections should remain the same.

Have you checked that the glass is flat and of uniform thickness? I bought 3 float glass plates, and one of them turned out to be slightly bent. So if I didn't put it back on the bed in the same orientation as when I took it off, it threw the calibration out. I threw that piece away when I discovered this.

The recipe I now use is PEI on a glass sheet (checked for flatness this time) as I describe towards the end of [miscsolutions.wordpress.com].

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

What paint did you use? I have found 2 choices: A high temp flat black (Rust Oleum High Heat Enamel Flat Black Spray Paint) and a high temp satin black ( Rust Oleum Bar-B-Que Black Satin High Heat Spray Paint). Will the satin black work OK, or should I go for the Flat Black?

I went with Opaque Natural for the PEI. Will that work OK or do I need to find a clear (have not found one yet)?

Quote
c3dprinting
What paint did you use? I have found 2 choices: A high temp flat black (Rust Oleum High Heat Enamel Flat Black Spray Paint) and a high temp satin black ( Rust Oleum Bar-B-Que Black Satin High Heat Spray Paint). Will the satin black work OK, or should I go for the Flat Black?

I went with Opaque Natural for the PEI. Will that work OK or do I need to find a clear (have not found one yet)?

I used this [www.diy.com] from my local store. A matt finish would be better than satin.

You don't need clear PEI. The ideal would be a PEI that is completely opaque to near IR light (then you wouldn't need to paint the back black). The two samples of PEI I have are both translucent yellow, and both are highly transparent to IR.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I agree the PEI is the best option however i have this very problem trying to compensate for my bed with the H parameter so i cant stick anything really because i cant auto calibrate correctly yet :/

Well the end result:

PEI on bed, print sticks great but....

Still need to use the H to correct a couple points. I think there is a few spots I got a air bubble between the glass and the PEI. Around the Z tower is .15 to low and around the X tower is .18 to high. But all is working well so I will leave it. I really don't want to tear the machine apart and I am not printing anything that needs precision lower than .1 or .2.

I followed you post (DC42) about the PEI when prepping except I did not bake it. I just let it sit overnight to really dry well. Cutting was a pain in the butt with a blade but it was OK once I made it through. I decided to cut and shape after I stuck it on the glass to give it the best possible shape.

The PEI does seem to scuff up really easy. Will this affect my printing? If so how do I fix it in time?

I'm glad you got it working well. One of the reasons for painting the underside of the PEI is to avoid bubbles in the adhesive affecting the sensor. I don't know how hard the paint gets or sticks if you don't bake it, I only tested the adhesion after baking.

It's hard to get a perfect mechanical build that completely avoids effector tilt, so having to use small H corrections is nothing to worry about.

My PEI is a little scuffed but this doesn't seem to affect the prints. Some of the scuffing may be because of filament residues, and this can be removed by heating the bed to 100C and then rubbing it with a tissue moistened with acetone.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].