M592 activation issue

RomeoFox

Right now I am trying to calibrate the dynamic nonlinear extrusion (M592).

-> https://forum.duet3d.com/topic/6015/guide-for-finding-the-values-for-m592-non-linear-extrusion/5

The result of the table from the thread is the following command: M592 D0 A-0.11 B0.04 L2

=> Should result in an effective correction of 4% at 11.6 mm/s.
Whereby with L2 any limitation should be switched off.

For test purposes, I have tried values up to 0.2 for M592 Bxxx and have the impression that nothing is happening at all.

For testing I use the following command: G91 G1 X-100 E100 F700

However, I cannot see any difference between
G91 G1 X-100 E100 F700 (due to movement with correction) and G91 G1 E100 F700

Does G592 need to be activated elsewhere?

M115
FIRMWARE_NAME: RepRapFirmware for Duet 3 MB6HC FIRMWARE_VERSION: 3.4.0rc1 ELECTRONICS: Duet 3 MB6HC v1.01 or later FIRMWARE_DATE: 2022-02-09 10:28:13

Phaedrux

@romeofox said in M592 activation issue:

However, I cannot see any difference between

How are you comparing? The effect should be quite subtle regardless.

RomeoFox

@phaedrux said in M592 activation issue:

How are you comparing? The effect should be quite subtle regardless.

I measure the filament feed on the feed side of the extruder by marking the filament with a pencil and steel ruler and extruding 100mm. At speeds below 5mm/s, the feed is exactly 100mm. At 11.7mm/s =700mm/min 4% are missing = only 96mm are fed.

I would have expected 4%/4mm to be easily compensated.

Phaedrux

What is the volumetric rate that you are trying to push? Non-linear extrusion tries to compensate, but it can't push unmelted plastic.

RomeoFox

@phaedrux said in M592 activation issue:

ate, but it can't push unmelted plastic.

At 1.75mm filament, that's 28.2mm³ per second with a Dragon High Flow Hotend and ABS+.

The extruder starts to slip through individual teeth between 900 and 1000 mm/min. Therefore, I assume that the 4% should actually still be possible at 700mm/min.

Phaedrux

As a simple test, does increasing the hotend temp allow for more expected results?

RomeoFox

@phaedrux

I tried today (+30°C) same behavior. From the answers I understand that it should have worked with the command M592 D0 A-0.11 B0.04 L2 alone, right?

Phaedrux

@romeofox said in M592 activation issue:

I understand that it should have worked with the command M592 D0 A-0.11 B0.04 L2 alone, right?

I can't say that for certain.

If you use this technique for finding your volumetric flow rate, what do you get?

https://duet3d.dozuki.com/Guide/Ender+3+Pro+and+Duet+Maestro+Guide+Part+4:+Calibration/40#s177

dc42

@romeofox to test that M592 is working I suggest you disconnect the hot end from the extruder so that there is no significant back pressure, and then measure amount extruded vs. speed.

timschneider

@dc42 said in M592 activation issue:

@romeofox to test that M592 is working I suggest you disconnect the hot end from the extruder so that there is no significant back pressure, and then measure amount extruded vs. speed.

Sorry for recycling this thread but I discovered the same behaviour since 3.4.

Our method to calibrate nonlinear extrusion was

heat up the nozzole to working temperature
mark filament at 120mm
extrude with G1 E100 F30 (same as e-step calibration) and mesaure
repeat steps 2. and 3. for F60,90,120,150,180
calculate A and B
set A and B values
check result with mark 120mm, extrude G91 G1 X100 E100 F30 - F180 measure and compare
if not ok adapt A and B and re-run steps 6. - 8.

But since 3.4 step 7. is not working anymore - there is no difference in the values measured in step 3. - 4. and 7 - they are just about the same. upto 3.2.2 we had a difference upto 10%. We cannot use 3.3 because it will flood us with errors.

I looked at the code and found that you @dc42 did alot of work in that area (InputShaping).
We use an CoreXY with Duet2 an RRF 3.4.

dc42

@RomeoFox @timschneider I confirm that this is a bug in RRF 3.4. I will fix it in RRF 3.4.1.

Temporary workaround: for Duet 3 multiply the nonlinear extrusion A coefficient by 750000 and the B coefficient by (750000)^2. For Duet 2 use 937500 in place of 750000.

Thanks @timschneider for identifying the change that introduced the bug.