Accuracy issues

So I've been struggling for a few weeks to make my self-sourced Mendelmax 2-based reprap machine (ramps 1.4, drv8825 motor drivers and nema17 stepper motors) a little more accurate. I have tried calibrating backlash settings in the firmware (repetier FW with repetier host, slicing with Slic3r) but my measurements indicate that I have very little hardware backlash so there is almost nothing to compensate. However, I still have the following problems:

1. Accuracy. Small dimensions turn out too big and and large dimensions - too small. I created a calibration piece and have compared the dimensions to the stl model:


2. Walls not sticking to the infill. You can cleary see it in the upper left quadrant as well as on the right wall of the small square:


3. Holes not really round. I made a small part just to test the holes. Here it is, front and bottom side:




What I have tried so far:
- troubleshooting any mechanical issues. Hardware seems to be completely ok.
- increasing motor current
- experimenting with different flow rates
- double-checking the steps per mm
- software backlash compensation

What can I do apart from that? Pretty sure the reprap machine can do better

It looks like you have major over extrusion issues. Luckily it's not that hard to try and fix!

Step 1: Calibrate extruder with Thomas Sanladerer's guide. Google Tom's extruder calibration. Watch some of his other videos while you're there.

Step 2: Mesaure your extrusion width. Heat your nozzle up, extrude filament through nozzle, measure in multiple places, put value into the slicer. My nozzle from E3D for example came with a .4mm nozzle that extrudes .5mm plastic!

Step 3: If you have a Chinese knockoff hotend, especially one from Folger Tech throw it in the spares bin. Since I got my E3d my part accuracy has gone up exponentially. Nuts fit into their traps, calabration cubes 20+-.01mm. It's worth it. See picture, blue is Folger tech, red with real E3D. Same settings for both prints.

Edit: picture didn't upload but trust me, it's much better. Also forgot to mention long rectangular prints tend to contract a bit so keep that in mind.

Edited 1 time(s). Last edit at 11/07/2015 10:11PM by gmckee.

---

Prusa i3 Rework - Ramps 1.4 - E3d Lite6 - Full Graphic LCD Controller

Thanks

I had a chinese hot end, which was disastrously bad, so now I have an original e3d v6. I have calibrated the extruder exactly according to Thomas Sanladerer's youtube video (extruding 100mm and adjusting the steps per mm until the extruded filament length is exactly 100mm). Then I calibrated the flow rate, so that when printing a cube with 0% infill the wall thickness is exactly 1,05*(nozzle diameter). In my case: 0,4mm nozzle diameter and 0,42mm wall thickness. So this is why I thought the flow rate is well calibrated. I could try to reduce it some more... but then the large dimensions would become even smaller, correct?

I am going to try the method you suggested in Step 2. If all is correct, I should get 0,42mm extrusion width. If not I will change the extrusion width in the slicer.

Any other thoughts?

Use the "E steps Fine Tuning" procedure in this page: [reprap.org]

Trying to measure the plastic width coming out of the extruder is extremely difficult to do accurately and precisely. Even if you have a precise, accurate instrument, if you let the plastic hang from the extruder nozzle, as it gets longer and heavier, the apparent width coming out of the nozzle will decrease because the it's being pulled by gravity, a situation that doesn't resemble the way the printer works at all. Even the Chinese hot end makers have 0.35 or 0.4 mm drill bits. I would trust the nozzle to be pretty close to the specified size, and at the very least, it will be a constant size. If you're trying to measure the width of a printed wall rather than the noodle coming out of the hot-end, you may still have a problem. If your printer's layer registration isn't perfect (z wobble, etc) the wall will measure thicker than the extruded lines that make up that wall. When you get down to measuring a few tens of microns, there are just too many ways for error to creep into the measurement.

You think you are measuring a flaw in the hot-end nozzle but it is more likely that you are measuring an error in the filament diameter (compounded by printer precision). The slicer assumes that filament diameter is constant, and the size you told it. If the filament is larger than the diameter you entered, you'll get over extrusion. If the filament is smaller, you'll get underextrusion. It isn't a problem with the hot-end, it's a problem with the filament.

Filament "diameter" varies from one spool to the next and even on the same spool. When you calibrated the extruder, or printed your test part, did you use a measured value for the filament diameter or just plug in the nominal 1.75mm diameter? Filament normally has an oval crossection, and can easily vary by a 1/10th of a mm depending on the orientation of the caliper when you measure it. For that reason, I always measure 20-30 places spread over 5-10m of filament, varying the orientation of the caliper, and calculate the average value. That average value is what goes into the slicer/printer when I print. In more than 2 years of doing this I have had only one spool of filament where the average measured diameter was 1.75mm. A few 1/100ths of a mm make a big difference in over/under extrusion. Be sure and mark the average measured diameter on the spool so you'll know it the next time you use that same spool.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

Thanks for the advice. Truly there is some Z wobble that adds about 0,1mm.

As for the filament diameter, I didn't just plug in the nominal 1,75mm but measured the diameter at about 10 places. However, I didn't spread the measured places over 5m. Instead I measured over 1m and just calculated the mean value which was almost 1,75mm. It is a good idea to spread the measurement point over a higher length of the spool. It is nevertheless not really easy to measure the diameter because the filament is usually not perfectly round. What I find even more frustrating is that measuring the diameter accurately with a digital caliper is next to impossible because the pressure on the caliper alters the measurement by a few hundredths of a mm.

Anyway, even I have a small error in the filament diameter, calibrating the extrusion steps and the flow rate should compensate for that. I did Triffid Hunter's E-Steps fine tuning and printed a small part (5x5 cm) which turned out pretty precise (+/- 0,1mm which is actually my goal). My printer is currently occupied with parts which I actually need for my projects so I will be able to print the same test parts on Wednesday in order to compare the dimensions. I will definitely post if the accuracy has become any better,

Two more questions:
1. What about my problem #2, walls not touching the infill? Is this also connected to over-extrusion?
2. I have a chinese ramps 1.4 board and chinese arduino and motor drivers. Could it be that Chinese electronics has an effect on accuracy? I wouldn't think so but I'd like to bring up the subject anyway

If the pressure applied by the caliper causes the value to vary, you're applying too much pressure to the caliper- the filament is pretty solid/hard and won't give, so you're causing the caliper to flex. Use a lighter touch.

1) if the infill doesn't touch the perimeter walls, increase the infill overlap setting in slic3r- it defaults to 15%. If you'ree already at 15%, you may be under extruding. You can also check the infill line width- I usually set it equal to the nozzle diameter.

2) I doubt the electronics has any effect unless the motor driver isn't capable of driving sufficient current into the motor. It is a good idea to use higher drive voltage- I run the X and Y axis motors in my printer at 32V using separate power supplies and DSP stepper drivers. Higher drive voltage to the motors may result in more vibration...

When you get done calibrating the extruder, you might consider checking the orthogonality and accuracy of the printer's axes. I designed a test cube to print and wrote a companion spreadsheet that calculates the required corrections based on measuring the printed test cube. If the axes aren't orthogonal, you'll have a tough time printing gears that mesh and screw-on lids for round containers, etc., because the things you designed to be round won't print that way.

Here it is now: [www.youmagine.com]

Edited 1 time(s). Last edit at 11/09/2015 08:15PM by the_digital_dentist.

---

Ultra MegaMax Dominator 3D printer: [drmrehorst.blogspot.com]

So after experimenting further with the E steps per mm, the hot end and bed temperatures and filament diameter I think I have achieved sufficient accuracy. Small dimensions are fairly precise (+/- 0,05 mm) and large dimensions are about (+0; -0,2mm), probably due to material shrinkage during cooling.
I am going to consider this when creating my models.

Thanks for the help