Multiple issues with Duet Ethernet since 2 weeks

V3DPrinting

I have done a second test print. Still the same L shape, same filament, same gcode.

The print is very similar to the previous one.

Except some under extrusion at the end of the print (the X letter is worse on the second print)

and the first layer have less wobbling, but it looks like the layers are a bit more than the last layers.

Both X and Y arms have the same pattern.

I will permute E0 and E1 drivers and reprint the same test.

dc42

@v3dprinting said in Multiple issues with Duet Ethernet since 2 weeks:

Some technical questions about the architecture of the board :
Is the clock global for the board on the processor ?
Same for the impulsion generator

The WiFi and Ethernet modules have their own clocks, but other than that there is a single 12MHz crystal oscillator, which the processor multiplies up to 120MHz. Step pulses are generated by the processor using a counter driven by this 120MHz clock with a prescaler of 128.

V3DPrinting

I have remapped the Y and Z axis to their normal assignation, as the problem was not linked to the drivers nor steppers.

I have swapped E0 and E1 drivers and done a test print.

From a dimensional perspective. no problem, it's very similar to the ones done since yesterday.

From the surface aspect perspective, still the same first layers that looks bigger and patterns, but the patterns are less prominent.

I will remap the extruders drivers to the normal and do a print using the second extrusion train (which I only use doing specific materials or nozzle diameter.
So if there is a mechanical issue on the extruder it will be revealed.

Dougal1957

Just a thought but are you using a recent version of Cura by any chance I have heard on another forum that a bug has been introduced that sets the filament diameter to 2.85 and one version didn't activate the change if you set it to 1.75) and also I believe a few are having variable extrusion issues with the latest one. this is hearsay mind as I don't and probably never will use Cura

V3DPrinting

@dougal1957 said in Multiple issues with Duet Ethernet since 2 weeks:

Just a thought but are you using a recent version of Cura by any chance I have heard on another forum that a bug has been introduced that sets the filament diameter to 2.85 and one version didn't activate the change if you set it to 1.75) and also I believe a few are having variable extrusion issues with the latest one. this is hearsay mind as I don't and probably never will use Cura

No, I use S3D for those prints.
Though, last week, I have done a test with Cura 3.6, but the result was exactly the same.

I haven't upgraded to S3D 4.1 for the moment, as I would like to understand were the issue is.

Dougal1957

@v3dprinting it was just a thought

V3DPrinting

@dougal1957
I use Cura for my 3 Ultimaker (3 and S5) and I don't have issues with the extrusion setup.
Works pretty well and for some features way better than S3D.

Dougal1957

@v3dprinting I believe it was V 3.5.1 that had the issue this details the bug

V3DPrinting

I have done the last test, using the second extrusion train which is not used usually.

The test print is good from a dimensional perspective.

I still have some patterns on the print, like in all the previous tests.
But on top, there is some sort of periods of the print without the pattern.

There is an inconsistent extrusion across the print

I have printed all the tests with different spools of filament, having the issues. All from quality manufacturers (FormFutura, ColorFabb). All from spools already used with successful prints.

The Z axis issue haven't been reproduced since yesterday, but might reappear as no explanation was found.

So I believe there is something wrong with the board.

I've got an hypothesis : the oscillator is having some derivation and / or the processor is not generating the right form factor.
But it is not constant over the time, from the very first symptoms, might be periodic.

Given the steps per mm M92 X160.47 Y160.11 Z1600 E505.7:505.7, a 10% deviation of the oscillator (which is huge !) will have an impact equivalent to :
160 steps on Z so 100 microns
50 steps on E0 or E1

So it would be very visible on Z and Extruders, but not that much on X and Y.
If there is a period or so in the deviation, it might not create an issue in X an Y.

Thanks for your feedbacks !

V3DPrinting

@v3dprinting said in Multiple issues with Duet Ethernet since 2 weeks:

I've got an hypothesis : the oscillator is having some derivation and / or the processor is not generating the right form factor.
But it is not constant over the time, from the very first symptoms, might be periodic.

Given the steps per mm M92 X160.47 Y160.11 Z1600 E505.7:505.7, a 10% deviation of the oscillator (which is huge !) will have an impact equivalent to :
160 steps on Z so 100 microns
50 steps on E0 or E1

So it would be very visible on Z and Extruders, but not that much on X and Y.
If there is a period or so in the deviation, it might not create an issue in X an Y.

Thanks for your feedbacks !

As the Z axis only goes on the same direction during the print, we cumulate the errors.
For X and Y axis, as it goes forwards and backwards, errors might be minimized for that test print.

V3DPrinting

I have done a real print yesterday night. Hopefully not for a customer.
It is a print I've already done successfully a year ago with the same material and same S3D factory file, just updated it for the filament diameter.

From a dimensional standpoint, no major issue.

From a surface finish standpoint, many many extrusion issues:

Patterning is still present
some sections with over extrusion and under extrusion from one layer to another

Definitely there is something wrong with the board.

I will do another print of the same object, the printer is now powered on since 2 days.

V3DPrinting

I have tried to reprint the part.

First try, the print has started pretty well, but doing the first layer infill, ticking noise from the extruder, then under extrusion. Skirt and perimeter were ok.

I've checked the hot end, not jammed nor clogged
I've continued the print and then clear over extrusion and back to nearly a normal extrusion.

Then under extrusion. I've stopped the print

The spool, ColorFabb PLA/PHA had printed successfully many parts with these settings.

I've tried to reprint the same gcode file, but no success, I had nearly no extrusion, though the nozzle was fine, just checked before starting the print.

I've done a last try and had the same scenario as above in this post.

The printer had been switched on for 2 days.

So I am sure the board is faulty :
Z axis height not consistent
Extrusion not consistent

BTW I have noticed when using a daughter board, the CPU is partially masked from any blower, so maybe the cooling is reduced.

Feedback appreciated.

V3DPrinting

I have let the printer switched off a couple of hours

I have upgraded to firmware 2.01 (last stable release) with DWC in order to check if the problem was linked to the firmware.

I've done a new test print with the L.

Dimensions are good.
No sign of under or over extrusion.

I got significant pattern on X and Y, like before.

So the issue is not linked to the firmware.

Attached the new M122
0_1544811139548_M122 131218.txt

V3DPrinting

I have reprinted the same test print in a row.

I have a good first layer for the skirt and perimeter.
Then the ticking noise from the extruder and over extrusion and then under extrusion.

The problem is due to improper retraction at a certain point.

I've stopped the print with severe grinding and the E3D V6 jammed with the filament retracted very high after the heatbreak.

So the extruder is not properly controlled by the driver / board when having used.

It confirms the board is faulty.

I need to replace it.

V3DPrinting

I have well received my new Duet 2 Ethernet.

I have replaced it very easily as it is the same layout.

The old one had some oil on it on the side with the chips.
I don't understand how oil have managed to get on the side which was facing down and no oil at on the other side.
Anyway I will clean the board with isopropyl alcohol.

In order to avoid some configuration steps, I took my old USB card and put it in the new board, as it was already with firmware 2.01.

I made a first test print, L shape 10 mm height. Same slicer and parameters.
I got the same issues as weeks ago, when I had the first issues. First layers with a higher layer height than the selected resolution

I've done the test a couple of time with the same issues.

So I switched the usb card with the one supplied with the new Duet board.
I've done the upgrade to firmware 2.02 and copied only the configuration files (config.g, homexxx.g ....).

I've done a new test print which was perfect !

All the prints done were good too.

Until today, when I've decided to upload the gcode files from the old usb card to the new one. I've done it using FTP as it's quicker.

The print done just after had the first layers height issue.

So I've removed the gcode files from the usb card and the print were good again.

I have checked the old usb card with fsck, no issues detected.

So the problem is not on the board itself and the culprit is the usb card, more precisely the amount of available disk space.

Is there any memory swap or some temporary system files when running on the usb card, so a minimum available disk space is required ?

That said, cleaning the old board is mandatory, but I believe it would also work.

Thanks for the feedback.

dc42

What a strange problem! While printing, the firmware writes nothing to the SD card unless you have logging enabled; in which case is will write any error messages to the log, along with a few other messages such as print started and print stopped.

Do you have logging enabled?

Do you have bed compensation enabled; and if so, did you set a taper height?

V3DPrinting

@dc42 said in Multiple issues with Duet Ethernet since 2 weeks:

Do you have logging enabled?

No

Do you have bed compensation enabled; and if so, did you set a taper height?

Yes.
M376 H2 ; Set bed compensation taper

dc42

@v3dprinting said in Multiple issues with Duet Ethernet since 2 weeks:

@dc42 said in Multiple issues with Duet Ethernet since 2 weeks:

Do you have logging enabled?

No

Do you have bed compensation enabled; and if so, did you set a taper height?

Yes.
M376 H2 ; Set bed compensation taper

I think the problem is that you are applying a large amount of bed compensation and you have set a small taper height. So:

Check that your height map doesn't have a significant offset. It should have close to zero height error (green in DWC) close to the centre of the bed.
Use a larger taper height, e.g. 10mm, or no taper at all.

V3DPrinting

Thanks Dave
I'll have a look as soon as my printer is free.

V3DPrinting

Just an update on that old topic.

I had many issues with that printer, always the same type of steps not being consistent.

I changed many things on the printer :

PSU (400W 24V TDK) replaced
Duet 2 Ethernet replaced
Probe (capacitive sensor) replaced by an IR Sensor
Opto coupling board for the capacitive sensor removed
Duet 2 grounded to the casing
Casing grounded

Stepper motors were checked for each wiring impedance, but not replaced.

It had changed nothing to my issues, after a certain time (couple of days) being switched on, the steppers were not having consistent steps.

Until two weeks ago, were I've discovered that the ground provided by the mains in my building (I'm in a business center) was faulty with a constant 105V - 30 microA AC current....

It has been fixed this week. I hope it was the root cause.