Validating Linear (/non) Extrusion Correction Settings

DocTrucker

Hi All,

I go through a process of taking three samples over a range of filament speeds ranging up to the point where my extruder stalls in order to plot the required extrusion correction graph, and calculate the A and B parameters for the extrusion corrections on that spread sheet.

Previously I carried out these tests, calculated the required correction, then tested that by extruding while moving the head. with command sequences such as:

G1 X0 Y-50 F3000
G1 Y50 E100 F40

These no longer seem to have non linear extrusion corrections applied to them. I realise the static extrudes didn't have non-linear correction due to the mid print filament retracts but why would the above not work? Am I going to need to run my command sequence in a build file, or have I missed something obvious?!

DocTrucker

Without reading all the way through the "What's New" file on the github page I can't see anything obvious that mentions this change in behaviour which makes me think I have missed something. I may not be able to do further tests today but I will run a few at some point.

Running on a v0.6 duet and a Shield. The the extruder is running off E0, is a 0.9 degree motor, running through a bondtech BMG clone. The clone will get replaced as and when funds allow!

fcwilt

Hi,

Just out of curiosity what are you talking about?

I have not heard about "extrusion correction" before.

Thanks.

Frederick

deckingman

@fcwilt i think he is talking about this https://duet3d.dozuki.com/Wiki/Gcode#Section_M592_Configure_nonlinear_extrusion

DocTrucker

@fcwilt Yup, M592. If you calibrate e-steps at low speed with a V6 like hotend you will typically find it extrudes less when asked to extrude a fixed amount significantly faster.

fcwilt

Thanks for the info.

Interesting topic.

Frederick

samlogan87

I had a play with it a few nights back and I was having some inconsistency problems as well. I used someone’s macro that they put up on here, did the tests, added the values into my config file, and ran the test again. I was still out by 2-3 mm across the board. Not 100% what is going on tbh.

Sam

DocTrucker

@samlogan87 that's roughly what I saw. Ran a load of tests and determined I needed 0.007 on the linear correction. Checked it, and it was out. Doubled the correction and it was still out. Need to run more tests to check things out better but something weird going on.

samlogan87

Let us know how you get on. I will be intrigued to see what you find out.

Regards
Sam

DocTrucker

Currently running test 3 of 18 mapping the required correction. Will follow it with 6 tests with the extrusion commands within build files, and a further 6 outside build files. My test files are:

G1 X0 Y-50 Z50 F6000
G1 F27
G1 Y50 E50 F27
G1 Y-50 E50 F27

My test F rates are 27, 82, 137, 192, 247, 302. Filament diameter 1.75. Stock E3D V6 with 0.4 nozzle.

DocTrucker

Ran back to back tests before M592 correction, after correction with G1 commands in a build and after. I can confirm they both give the same results and a correction is applied. Not sure what went wrong on my first trial!

Thanks again.

fcwilt

Hi,

So what is the result of this correction? Does it make a significant difference in the printed items?

Thanks.

Frederick

DocTrucker

My working print speeds are 10mm/sec through to 90mm/sec. In this range I was getting perfect 100mm lengths of extruded filament at low speed through to a 5% under extrusion at the higher extrusion rates. With the correction the requested length is all within 1%. It'll have a similar effect to being able to increase the extrusion multiplier by 0.05 for the large open areas, but leaving it at a lower value for slower sections. Makes it easier to tune.

Note: Below 40 is only used if the Slic3r thinks the layer time is too short - in other works I have prepared a bad build, on purpose or otherwise!

fcwilt

@doctrucker

Thanks for the info.

But does that improvement from 5% to 1% result in a significant difference in the printed items?

Frederick

DocTrucker

In the past it has for me yes, so I just continue with the calibration process so that things are as close as possible to requested. The issue I was getting was if settings were good for large areas of solid fill small areas were over extruded, or visa versa. I don't think a benchy will show the difference but I will run two back to back with and without the M592 correction to see if you can see the difference on that. I suspect not.

To be honest if not apparent on the benchys the build times of the sort of files I had difficulties with (Panel Due mounts, Duet mounts, etc) are not something I'm likely to trial for the sake of it in the near future.