Strange and undesirable pressure advance behaviour

deckingman · 7 Jan 2017, 16:47

This one is for David (DC42)

David,

I started to print a test piece in order to refine my retraction settings now that I have initiated pressure advance and noticed very strange behaviour. It's easier to show on a video than put into words so here is a link. If you don't want to watch the whole thing, juts look at it from about 8 minutes onwards, concentrating on the extruder gear. The earlier footage just shows the same thing happening with smaller cubes.

https://drive.google.com/file/d/0B_MwtHtQR_ZvWXRQWnVuRTNkYjQ/view?usp=sharing

Basically, I started printing different size cubes so most of the moves you see are 45 degree infill. There is no retraction at the end of each move - only for the non print moves.
It appears that the extruder behaves in the way I have become accustomed to seeing it up to the mid point of the cube. Then, (this may be just a coincidence) as the moves start to get smaller, the behaviour of the extruder changes completely. Up to the mid point of the print, for each move, it looks like the extruder accelerates, runs at a constant speed, then decelerates, even moving backwards at the end of the move. This is how I normally see it behaving. Once the print goes past the mid point of the cube, where the moves start to get smaller, the behaviour of the extruder changes to what looks like an acceleration followed by a pause with no further movement of the extruder for the rest of the move. This continues more or less for the rest of the cube although it seems that "normal service is resumed" during the last very small moves. Then the printer moves on to the next cube and the pattern of normal behaviour up to mid point followed by abnormal behaviour from then on to the end of the cube, is repeated. Watch the video and you'll see what I mean.

Setting as follows:

Duet ETHERNET 1.0 +Duex 5
Firmware 1.18.1
Web interface 1.15a

relevant extracts from config.g

M207 S4.0 R0 F2400 ;set firmware retraction S=amount in mm, F=Feed rate mm/min(/60 to get mm/sec), R = additional length, Z = additional Z lift in mm
M572 D0 S0.50 ; set pressure advance coefficient
M572 D1 S0.50
M572 D2 S0.50

Tool definition used for this test
M563 P0 D0:1:2 H1 ; Define tool 0
G10 P0 X0 Y0 ; Set tool 0 axis offsets
G10 P0 R0 S0 ; Set initial tool 0 active and standby temperatures to 0C
M568 P0 S1 ; Enable mixing for tool 0
M567 P0 E0.90:0.05:0.05 ; Set mixing ratios for tool 0

Speeds and accelerations
M566 X1200 Y1200 Z20 E1200:1200:1200 ; Set maximum instantaneous speed changes (mm/min)
M203 X50000 Y35000 Z300 E3600:3600:3600 ; Set maximum speeds (mm/min)
M201 X1200 Y660 Z10 E2000:2000:2000 ; Set accelerations (mm/s^2)

Let me know if you need any other info or want me to run any other tests

Ian

dc42 · 1 Jul 2017, 17:44

Ian, please can you:

1. Run M122 after or during this print and see whether the Step Error count is nonzero after you observe that behaviour. That count gets reset to zero after each time you run M122.

2. Upgrade to 1.19beta8 and see if the problem goes away. Caution: the board MAC address will change, so if the Duet is getting its IP address via DHCP then the IP address will change.

deckingman · 1 Jul 2017, 19:45

David,

M122 shows step error count as Zero while the strange behaviour is being observed - tried it several times throughout the print.

Additional info - the behaviour happens on the first layer which is printed at 45 mm/sec as well as subsequent layers which are printed at 90 (as was the case in the video). So I'd say it's highly likely that print speed is not a factor.

I'm about to do the upgrade to 1.19beta8 so will report back on that shortly…........

deckingman · 1 Jul 2017, 20:28

David,

Quick update. Changed to 1.19beta8, started the print again and an thus far, no sign of the previous strange behaviour. Looks like the firmware update fixed it. Curious to know why or what the firmware change was that fixed it.

Ian

Whitewolf · 7 Feb 2017, 00:01

@dc42:

Ian, please can you:

1. Run M122 after or during this print and see whether the Step Error count is nonzero after you observe that behaviour. That count gets reset to zero after each time you run M122.

2. Upgrade to 1.19beta8 and see if the problem goes away. Caution: the board MAC address will change, so if the Duet is getting its IP address via DHCP then the IP address will change.

I have to second the question about if there are some improvements in 1.19 beta 8 to the pressure advance because I am currently running through a lot of retraction test prints to find the right balance of pressure advance and retraction and if there have been improvements then im thinking it would be smart to upgrade before doing any more calibration.

That is if there are not any known gotcha bugs for a cartesan printer on 1.19 that i should be aware of.

I am currently running 1.18

dc42 · 2 Jul 2017, 06:14

Yes, there is a bug fix to pressure advance in beta 8. I thought the bug didn't change anything except cause a step error to be logged, but it looks like it does.

Whitewolf · 2 Jul 2017, 07:33

@dc42:

Yes, there is a bug fix to pressure advance in beta 8. I thought the bug didn't change anything except cause a step error to be logged, but it looks like it does.

I have not upgraded to the new firmware yet but here is a side by side photo with the only difference in settings being pressure advance is off on the evenly spaced infill (top) and it is on in the oddly spaced infill (bottom). As you can see it is a repeated spacing error in the pattern when it is on.

dc42 · 2 Jul 2017, 07:40

I find that very odd. Is it reproducible? Does M122 report and step errors?

deckingman · 2 Jul 2017, 10:18

For info, I was printing cubes with 10% infill so they look very similar to the pictures above. Infill spacing looks fine. That's with 1.91.beta8. Unfortunately I never got to printing infill with 1.18.1 so can't say if the problem existed with that version of firmware on my printer.

Whitewolf · 2 Jul 2017, 18:40

Yes it is reproducable, both prints are 25% rectilinear infill. each print with pressure advance turned on has that odd spacing. I will rerun the prints again today and check for missed steps. Odd that it only occurs with pressure advance on so i doubt it is missed steps.

deckingman · 2 Jul 2017, 19:43

@Whitewolf. The other Odd behaviour I had with 1.18 didn't show any missed steps but went away when I upgraded to 1.19.beta8

Whitewolf · 2 Jul 2017, 22:10

Yeah, I plan to upgrade just running some prints right now so I have something to compare to…. my kids ran off with my prints as usual lol

GrodanB · 7 Mar 2017, 20:47

Bad picture but something strange with the infill…

1.19beta8

deckingman · 3 Jul 2017, 21:08

Errr. What are we looking at? Are we comparing one with another or what?

GrodanB · 7 Jun 2017, 08:52

Rightclick the image and view in window… Look at the infill... Before I had nice infill almost straight lines... now the are doubled up... But it might be stronger infill like this... But I do not like it at all...

also now have a strange outside:

I only had a crappy mobile phone to photo this time…

OK lets try again:
Pressure Advance ON:

Preasure Advance off: (better camera)

dc42 · 6 Jul 2017, 07:56

Pressure advance should not affect infill except at the ends, unless it is an indirect effect of reducing acceleration to stay within the configured extruder jerk. Looks like a layer registration issue. Are you also getting Z banding?

GrodanB · 7 Jun 2017, 09:53

No, not that I can see.

Do not consider this banding issues.

Once the part currently printing is done I'll check that part…
So far the part seem to print OK... Some issues like I do not get a good corner, I still do not know what to tune...

I do think I under extrude some but the result om the part is so good except the infill since it prints every second layer I assume the gap is from there

What makes me supriced is the triangles the infill has when I have this uncommented:
;M572 D1 S0.05

This makes no sense BUT the only difference between the above prints is this semicolon.

Same machine, same plastic.

GrodanB · 7 Jun 2017, 10:44

Hmm… investigation on Jerk...

Saw that M566 E setting was low... looking into this I set it to E300:300... And hope for the best.

Turn on pressure advance and see it my trouble disappears...

At least point out the M-codes would be nice here so I can find the information here in the forum easier...
https://duet3d.com/wiki/Setting_acceleration,_jerk,_and_maximum_speeds

And some hint that pressure advance need an acceleration that is bigger than X (I found 200) and a pointer to the page that should say how to set that acceleration.

Also I thought that a M122 during print would be smart to see the CPU/RAM usage… And maybe stepping... Well I actually tried M112.. and that had an effect on the print... So just to ask before I try again... Can I sen an M122 during print?

dc42 · 6 Jul 2017, 10:56

M122 shouldn't affect the print if you use firmware 1.18 or later.

GrodanB · 7 Jun 2017, 12:50

OK, So far I now have the same result with pressure advance on. So it seems you are absolutely correct in the analysis and identification of the problem. Will add photos when the printer has printed more…

But you ALSO cut 2 hours of the print time. (estimate of course...)

This is the setting so far:

 [[GCode]]
M201 X800 Y800 Z15 E1000            		; Accelerations (mm/s^2)
M203 X15000 Y15000 Z100 E3600       		; Maximum speeds (mm/min) 
M566 X600 Y600 Z30 E300:300            		; Minimum speeds mm/minute

I see many extremely higher numbers… like;

 [[GCode]]
M201 X9000 Y9000 Z150 E6000 ; Set accelerations (mm/s^2)
M203 X12000 Y12000 Z375 E12000 ; Set maximum speeds (mm/min)
M566 X9000 Y9000 Z12 E9000 ; Set maximum instantaneous speed changes (mm/min)

Are these sensible and I'm a chicken or are these just extreme…

	[[GCode]]
	M201 X800 Y800 Z15 E1000 ; Accelerations (mm/s^2)
	M203 X15000 Y15000 Z100 E3600 ; Maximum speeds (mm/min)
	M566 X600 Y600 Z30 E300:300 ; Minimum speeds mm/minute

	[[GCode]]
	M201 X9000 Y9000 Z150 E6000 ; Set accelerations (mm/s^2)
	M203 X12000 Y12000 Z375 E12000 ; Set maximum speeds (mm/min)
	M566 X9000 Y9000 Z12 E9000 ; Set maximum instantaneous speed changes (mm/min)