Esteps and Ultimaker O+ and the magic missing 10%

[neotko 1,417]

Hi...

I been trying to set my esteps of a new mk7 that has more grip and after 4 days of prints, measuring, calibrating, going bananas I found a few things that I would like to know more, if someone has the knowledge to share.

I have 2 different mk7s, one from reprapeurope other form other place. The thing that drives me nuts it's that no matter how good my maths are, measuring the diameter of the mk7s * pi, using reprap calculators, every single time I get that the number it's a constant 10% incorrect to the reality.

So I started to dig on, how Ultimaker comes with their extruder feed number and it's kinda weird.

The Knurled bolt, diamond shaped, any name you want, from Ultimaker has a 8mm diameter. That passing it to the reprap steps calculator I get that, in theory, the ultimaker 2 feeder using 0.9 feeder nema, should use 256 esteps. But UM2 uses 282 as Estep number. That almost 10% (and a tinny little bit more)...

Ok

Now with my Mk7 from reprapeurope, it's quite well documented, it has 11mm radio that * pi gives a pitch of 34.55 and that on the reprap calculator tells me that my extruder using UM2 motor feeder (bough to ultimaker) should use 185.24. But when printing, the print sucks badly (with very good filaments that are 0.03+ margin of error). Ok, then I get that number and do a 10% increase. And bang, magic happens, I print nicely, with top layers without any error...

Also... To increase the mystery of that 10%. I did some digging into how Ultimaker Original feeder (big wooden gears) get's his esteps. Ok. If you get the gears info, the 8mm radio of the bolt and enter the data on the reprapcalculator it should be something like 760-762. And to my surprise, on the Marlin code, the Esteps of Ultimaker original are (760*1.1) on the configuration.h , so it uses the 'math' number, and then goes an extra 10%...?

So... How did ultimaker come with that 10% magical increase on the feeder? Because it's driving my nuts.

And it's driving me nuts because doing the classical extruder calibration I get that, without bowden, the numbers from the reprap calculator are accurate, and when doing the calibration trough the bowden and feeder filament and measuring the filament output, the numbers just go bananas and I get fluctuations of 4% from one test to another. So crazy...

So... If there's a 10% margin of error on the feeder and that it's fixed on umos by using a big gear ratio, to make that % margin of error the less possible or it's just that the bowden pressure adds a flat 10% error ?...

Edited January 29, 2016 by Guest

[neotko 1,417]

Btw for future mk7 users. Mine was 11mm inner dia (34.556 leadscrew pitch (11 * 3.1415) and reprap calculator gived me 92.60 but the perfect esteps (after 25 print testing this and 4 days of adjusting) was 97.6 (for 1.8 nema) and 195.2 (for 0.9 nema um2 style). Now I have 2 printers with different nema printing with the same perfect top layer quality. So the margin or error on 1.75mm filament it's less than 10% probably because it generates less preassure on the bowden.

Anyhow hopefully this will help someone since after countless google pages I found zero about a direct % that affects the esteps.

[peetersm 124]

I though once you got close with your math you are supposed to extrude some filament , like 100mm or so, and measure how much filament actually moved, and fine tune esteps from there.

I know this does not address the mathematical quandary , I agree there should be some theory to explain the 10%.

[neotko 1,417]

Agree. But.

After getting the math ok, triple check with calipers, the think failed to spot the perfect mm/s when:

- Extruding x filament without bowden

- Extruding x filament with bowden on

- Extruding filament to the head and measuring the filament moved.

Also there's a big bug on umo firmware when you change esteps as I wrote here:

https://ultimaker.com/en/community/19301-calibration-error-marlin-ultimaker-original-plus

Other user also got that error and some others:

https://ultimaker.com/en/community/19048-marlin-retraction-off-pause-bug

But... going back to the topic.

The thing it's that there's room for the filament to move inside the bowden, so also I tested the filament going out the bowden, simulating the curve it does. I even printed x3 capsule boxes to let pass 1.9mm filament and be able to see how much filament did move (since it's quite hard to actually measure filament straight unless you pass it through a tube.

After 4 days of madness I even started to redesign the umo+ feeder to allow me to use 1.75mm filament correctly (not posible by default since there's a gap of 1.65 air between the knurted bolt and the plastic bearing (trust me I did that for 2 months long ago). And then I started to think that maybe ultimaker already did this, specially when I saw that the firmware esteps for umo/umo+/um2 are not just math, but they add a flat 10%.

As for a method to properly calibrate the esteps, there's hundreds of posts of people, but not actual math involving bowden curvature. There's even a calculator that let you calculate the mm3 used if you actually can weight the extruded filament, but that also adds the error of the amount of filament that falls because normal drip of the head. So, when I saw that ultimaker machines add a 10%, then I started to use the math number, add 10% and then 2%, then 4%, 5%, 6% etc. Until I got perfect flat top layers. And now when I do a preview on S3D of what I'm going to print, it actually prints like the preview, so I'm happy but this it's been a hell of a crazy week.

So, this also reminds a think to say to @nicolinux about he getting first layers overextruded when using 1.75mm. I bet that if he uses the pure um2 number and * by 1.055 then the extrusion should be exactly like the one with 3mm (or at least with the same underextusion he was getting, that estep number should be 282 - 10% * 1.055, or 254.6 * 1.055, so 270,4 or 268.6 esteps). That means, that to have a really well calibrated extruded, when going from 3mm to 1.75mm the esteps must be readjusted.

I have 3 umo+ (1 it's disassembled partially to change some stuff) but on 2 umo+, one with nema 1.8 other with the nema 0.9, both with the same mk7 model, both now print the same. So I think I might be half right about this.

[gr5 2,265]

Some error might be because the base of the pyramids in the gnurled sleeve are the radius you should be measuring. Or somewhere between the base and the peaks of the pyramids.

More error (if measured while printing) is because the filament slips a bit. The holes that those pyramids make on the gnurled sleeve are not square but elongated. Illuminarti was testing this and noticed that the faster you print, the higher the pressure, the longer those holes get until they get so long that each hole reaches the next hole and at that point the filament gets ground up and the print fails.

[gr5 2,265]

More details here:

http://www.extrudable.me/2013/04/18/exploring-extrusion-variability-and-limits/

In particular look at the graph part way down the page. You can easily get 30% underextrusion when printing near the limit of what the feeder can do.

[neotko 1,417]

Yeah that's old. And not much related to the original question that was already answered by UM and they confirmed that they know that filament works like a spring inside the bowden.

Basically you flex the filament inside and push and hope for the best. Also that shows how much could help a gilament measuring system near the hotend to really know if the filament is moving or not. Anyway this topic continues a bit on http://www.3dprintingforum.us/viewtopic.php?t=4&start=30#p681 where Anders and other talk about that missing 10%.