On the past fill weeks (months?

(André Frazatto) #1

On the past fill weeks (months? couple years?) I have been on a rant driven effort to properly calibrate my home made printer and I must say I did great progress so far.

But…there is always a but isn’t it?
This week I noticed a strange pattern on my calibration pieces, if I use a 10x10mm cube to calibrate, when I try to print bigger projects everything gets short at the end.
So I steadily increased the size of the test pieces and apparently the bigger the print the bigger the error, it’s some non linear shit and I have no idea where it is coming from.

Sure, I could just calibrate for something bigger, but than the small prints come too big.

The error consequently also occurs on the size of walls and are getting up to 40% thinner than it should.

Any clues???

(Stephanie A) #2

Size error? Like a 40mm cube comes out 38mm?

(André Frazatto) #3

Yep, after calibrating for a small size piece, like a 10mm cube.

(Jeff Parish) #4

You might be adjusting your steps to compensate for thermal shrinkage properties of your filament at 10x10x10. Have you printed any thermal calibration clips?

Also, have you printed an accuracy block? 1" in Y, 2" in X, and 1/2" in Z. This can tell you if the error is step tolerance stack up. If your steps are the right ratios to each other for each axis but there is an error in the actual measurement the error will be multiplied evenly in all directions.

For example if you have an error of 0.1mm in Y you should see an error of 0.2mm in X. For the accuracy block mentioned above.

(André Frazatto) #5

@Jeff_Parish wait…what?
I never heard anything like this before and I’m not sure I understood correctly.

Indeed my X/Y motors are not simetrical. I always thought it was because all my hardware is cheap Chinese stuff.
What you are saying is that if I don’t get the steps spot on, this is what happens???

The termal calibration is spot-on now!
This is what started this whole craziness!!!

(Jeff Parish) #6

@Andre_Frazatto What magnitude of error are you seeing? An average printer will have a tolerance of about +/- 0.1mm. If you print a series of lines say 10 mm apart do they get closer or further apart or stay the same in relation to each other as you move further away from the home location? If you measure from the home position to each incremental line do you get an error stacking up? This will tell you if there is a mechanical imprecision in your machine through a stack up of step setting inaccuracies from a small scale measurement.

At 10mm with an error of 0.01mm you might call that good enough. However at 200 mm with the same amount of error you are now off 0.5mm. The ratio stays the same.

You need to be certain first that your printer moves exactly the distance you tell it to. This is not determined by printing calibration cubes at this point.

For example if you print a small cube that is supposed to be 10x10x10 and lets say the X and Y are both off by the some error factor so you change your steps to compensate you are likely adjusting your mechanical steps based on the thermal expansion or shrinkage properties of your filament for a small mass object. When you scale it up you will see an error related to that shrinkage rate multiplied and the error stack up in the steps is more apparent.

If the error factors are the same ratio wise in all directions then your steps are also the correct ratios to each other. If you know that your travel distance is correct then any refinement of actual measurement left is thermal calibration.

In the end it might be from cheap Chinese stuff. Just trying to help rule out other possible causes.

I set the steps for all the motors on my machines with a measurement and a mathematical calculation. My X and Y steps/mm always come out the same for 1/16 stepping on my drivers. I never adjust them from there. Movement in any direction is pretty close to exact what I tell it to move. Mechanically the printer is precise to within less than .075mm (about 3 thousands of an inch) at any distance.

Filament is the big variable. The first thing I do with a new roll of filament is print a set of thermal calibration “C” clips. I determine the proper melt temperature with a coiled rope (cobra) extrusion test about 35mm above the bed. Then I determine the true thermal diameter by printing the clips and testing the fit. Once calibrated to an easy but snug fit I write the melt temp and the thermal diameter on the spool and those are the settings I use in my slicer.

If I want to test the thermal stability of the filament I will print an accuracy block and measure the error ratio based on increasing mass. Typically not necessary but it does give me additional insight into the behavior of the filament I’m using if i want to go that far.

(André Frazatto) #7

@Jeff_Parish your answer is alsome!
I will have to revisit a lot of details to have a more clear picture of the problems.

Thanks a lot!

(Stephanie A) #8

Belts are typically calculated, not calibrated. The pulley should machined to enough accuracy.
It’s (steps-per-rev * microstepping)/(tooth-pitch * pulley-teeth)
So typically (20016)/(216)=100 steps per millimeter.

Jeff is right in that at small scales, the errors are from slicing errors, filament inconsistencies, math rounding, and overall machine issues like ringing. At larhe scales it relies on the belts and pulleys being calculated correctly and machined accurately.

You can calculate these errors as something like size=filament-error + slicing-error + movement-error*distance. At small movement the first 2 errors are prominent. At large distances the distance error takes precedence.

(André Frazatto) #9

@Jeff_Parish I had some time today and made some changes and tests.

First I tested the X/Y movement and they really were short, now when I say 100mm they move 100mm :stuck_out_tongue:

I printed the C clips with 0.1 clearance and they fit tight but fit, but I had better prints before.
Also printed a 20mm cube and it came out 20.5mm.
And a 100mm internal diameter circumference, which come out 99.4mm

I’m not sure what to make of those results…
It seams to me I should lower a tine bit the extrusion multiplier??

(Jeff Parish) #10

@Andre_Frazatto Sounds like you are over extruding a bit. You can either increase the diameter of your filament (+0.02 to 0.03mm) in your slicer or drop your extrusion multiplier a couple percent. Personally I like to leave my multiplier at 100% and use that to adjust for anomalies on the fly.

If you are getting small gaps between your outer shells in one direction or another then bump the temp by 5C and they will disappear.

Keep in mind that you need to allow for tolerance error in the objects you are using as gauges. Inner circles will typically have an error of 0.2 - 0.3mm radial clearance reducing your inner diameter. Your inner diameter measurement being 0.6 smaller is a little on the tight side indicating slight over extrusion again but this could also be related to the STL resolution. With STL circles and curves the profile is made up of straight lines. Unless you have a really super fine surface resolution you will see a smaller diameter than the outer points indicate because of the flat surfaces between them. This is expected and we need to design for it.

These calibration clips below include interlocking curves to test that tolerance. The example shows how they were used for thermal calibration of two filaments before using them for the checker board alignment calibration.

(André Frazatto) #11

@Jeff_Parish I’m aware of the “making circles” issues :stuck_out_tongue:

I will carry on with the adjustment, but I’m pleased for now.

Thanks for the detailed explanations, it really made a difference.