Too much Z acceleration with Slic3r?

I have an i3 reprapy-strapy thingy I made. It should print well and be properly calibrated except one problem.

Everything I slice with slicer seems to be ignoring my Z acceleration settings.

I print with pronterface from an ubuntu PC. using Slic3er for slicing.

my speed settings in marlin are:

// Homing speeds (mm/m)
#define HOMING_FEEDRATE_XY (50*60)
#define HOMING_FEEDRATE_Z (2*60)

//
// MOVEMENT SETTINGS
// @section motion
//

// default settings

#define DEFAULT_AXIS_STEPS_PER_UNIT {80.35,80.35,190,445} // default steps per unit for Ultimaker
#define DEFAULT_MAX_FEEDRATE {300, 300, 5, 200} // (mm/sec)
#define DEFAULT_MAX_ACCELERATION {3000,3000,5,10000} // X, Y, Z, E maximum start speed for accelerated moves. E default values are good for Skeinforge 40+, for older versions raise them a lot.

#define DEFAULT_ACCELERATION 30 // X, Y, Z and E acceleration in mm/s^2 for printing moves
#define DEFAULT_RETRACT_ACCELERATION 3000 // E acceleration in mm/s^2 for retracts
#define DEFAULT_TRAVEL_ACCELERATION 30 // X, Y, Z acceleration in mm/s^2 for travel (non printing) moves

// The speed change that does not require acceleration (i.e. the software might assume it can be done instantaneously)
#define DEFAULT_XYJERK 20.0 // (mm/sec)
#define DEFAULT_ZJERK 0.4 // (mm/sec)
#define DEFAULT_EJERK 5.0 // (mm/sec)


As you can see I have the accel on Z turned way down. Yet it still overshoots. Especially at the start of a print. Someone suggested it might be slic3r?

When I start a print it heats up the bed and homes. All good here.
then it lifts up and heats the extruder. Still good.
Then it homes again. Still good.
Then it lifts up a random distance goes to the start of the print, drops a little bit and trys to start the print at that hight. Usually a few mm above the bed...

I did have diferent z home max that fixed the start (250/1) (what’s with the multiplier there?) but I was still getting layer bands with low extrusion spread due to the nozzle being too high. and the Z screws keep getting out of sync after a few prints.

Mostly printing 0.2 layers, Smaller layers reduces the problem. but didn't get rid of it and I want to print with 0.2-0.4 layers sometimes.

I did get one great print after changing non-print acceleration on x y z to 30 from 3000. But then I thought I was clever and made more changes and screwed it all up.



here is a standard bad print. I'd show the one good one, but I was proudly showing everyone and can't find it for a photo...


I'm good at Electronics and Mechanics, but my coding sucks. Hopefully its something obvious?

Also if I slow the homing down too much it stops reading thermistors while in transit and takes 40 seconds to home z so ends up killing the printer due to heat error... So I would like to get this better.

Z seems to be fine doing 250mm/m with pronterface. How do I make it stick to that in a print?

Here is the latest fail in progress.


I had to pause and re home to get it to start on the bed...
As you can see its great when its not doing high layers... Although the acceleration is so low I printed most of that at 300% speed in pronterface (turned it up after the first few bad layers for shits and giggles.) and it still took 40 min...
The good one took about 22 min and I was happy with that. with 20% infil every 4 layers.

I got some improvement. Changed back to half steps. Set zstep to 390 (the proper number is 387.something, lost my notes)

Set XYZ accel to 100 for print and non print. I think I will dial it back to 85-90 next. It may still be missing small amounts of steps and its pretty quick at 100. Smoother speed changes won't hurt.

Set Z home max speed to 250/1. Still lost itself at the start, but missing stepps goind down rather than overshooting up and missing all the steps down.
Will turn this down to 150 next. I may have had it there before when it was good.

Found an escaping Z bearing rod. The one with the endstop on it.... That was definitely bad. No wonder I had to keep re-setting the bed gap...





This print started at 100% speed, went to 125% at the bottom of the X. then to 150% at the middle of the X then back to 85% for the top layers with the Z...
The slower bits are a little better. 100% is 60-80mm/s for most of it and its a wobbly old thing. I don't expect good results faster than 80mm/s...

Now I have over extrusion issues rather than missing layers its a very welcome change.
I can probably fix that with a more accurate z step number and some better calibration of temp and flow rate for a new nozzle/fan/ABS combo I have been running.
I haven't been able to get finer z step, e step and temp settings with the layering problem. Now I think I have it. I should be able to fine tune those steps again.
There may also be some small mechanical tolerances, its a wobbly piece of c**p

Why is it every time I have a problem for a few days I can't solve. That if I post here. I immediately solve it by myself?

Not that I am saying its solved yet. If I do that it'll screw up again...

What does the print look like at 20-30mm/s?

Going slower did make some small improvements.
Although reducing extruder temp and flow rate did a lot more.



The cube was started around 50mm/s and sped up from halfway up the X. The top half is rougher. Not sure going slower will improve results much more though.
The gear was sliced for 50-60mm/s but I slowed it down to 75% due to a disagreement between the print fan and the extruder heater in small layer times.
My first successful gear! Will be awesome in 0.1

Starting to look at the finer mechanical issues now I can find them

Unfortunately none of this helps me with Slic3rs super acceleration start. Its still skipping steps and needs to be re-homed after the nozzle heats.

Edited 1 time(s). Last edit at 09/03/2016 08:33PM by Mongrel_Shark.

Slower will help, i'd try 25mm all round, maybe even slower for perimiters, then perhaps any faults you then see will be the result of something else.

Slowing it down past 50mm/s makes little to no difference. As I stated the first time it was suggested. ( did try it and again today after the second time it was suggested.)

Mostly the printer goes best around 70-80mm/s. Probably because I have hardly touched the E accel settings.
Printing a long line slower makes the bead too fat at the ends and too thin (or even spotty) in the middle. Higher x-y speeds also seem to reguce flex deviations as the nozzle drags over any high bits (my flow rate may still be a bit high, need to sort those e accel settings next I think.)

Printing some little gears for an extruder at 25mm now. its a bit neater on really fine detail, but the acceleration on the start home is still way too high.
even if I set a max speed of 25mm/s in slicer and run pronterface at 50% speed, it makes no change to the start acceleration on that last home before the print starts.

the slice is clearly over riding the firmware.... Should I try messing with the Acceleration speed in Slic3rs advanced tab in print settings? Its currently set to 0
If I did what would be a good number? If I understand correctly that would be the max speed an acceleration move could go to, so setting it to 200ish should work?

I found some custom code in slicer printer settings that orders the lift after hotend heats and homes. It was originally set to 10mm lift at 5000mm/s

I changed it to 5mm at 120mm/s and that fixed the up overshoot and un-syncing of threads.

But then the first line of the slice (the bit that runs after the custom g-code for print start) was going back down too fast. Before leaving the home corner, so a manual pause and re-home is necessary. Or it tries to print a few mm above the bed on the first layer.

I changed the custom g-code in slicer to a 2mm lift and now its too short a distance to accelerate fast enough to miss steps.

So the start g-code in slicer's custom g-code tab was originally by default [G1 F5000 Z10] and has now been changed to [G1 F120 Z2]

This seems to have fixed it. Only done 2 prints this way. Will see if it gets out of sync after a few more prints.

Seems like a very kludgey fix...

Tempted to remove the line of custom code from slicer altogether. Whats it there for anyway? Seems it just makes it go back down right away anyhow. Seems like an unnecessary move to me. One thats causing problems, although its really highlighting another problem in the way slicer is slicing... Why would it override the firmware acceleration in the first place?

What would be nice is if it heated the nozzle above the home height, in the home corner, as it does, but then moves to the start of the print and re-homes z before printing.

This way the nozzle will be clear of the bed on the way from home to print start. Because if printing 0.1 layers and the bed is not perfectly flat, its dragging from home corner to start of print. Eventually this will wear nozzles and beds.... Getting the bed that perfectly flat is unlikely while I am using ABS juice on kapton..... Not very optimistic about finding glass that flat either. Even then ABS juice is adding significant variation of approx +/- 0.05mm.. I am currently getting around 0.1mm bed variations by homing Z 0.1 low (very firm on copy paper feeler gauge, bed springs with minimal pre-load) then printing a 0.3 first layer at 150% width that is actually printing at closer to 0.2 above the bed. Makes for a little trimming on some parts, like gears, due to first layer being 0.5mm wider like a brim. but its worth it for the better first layer consistency and adhesion. I just don't like the way the nozzle (lightly) drags over the bed in pre-print moves... I would also like to add some more pre-load to the bed springs as I think they are so loose its a source of slight play, and sometimes the nozzle pushes the bed down a bit and it binds on the mounting screw and does not spring back. But if I run more spring tension the nozzle can get blocked on the first layer (really nasty cheap Chinese hot-end)

try another slicer then, Matter Control, Baby stepping will help get the first layer down right, the gcode in the beginning will be different, I removed the homing, I position head at start position, turn on power, raise Z10, pump & prime, making sure filament is coming out ok, then lower Z & hit print, then if its dragging too much, raise Z 0.02mm with the baby stepping, repeat if needed..

Thats a vaguely useful kludge I will investigate. Not really intrested in trying to learn new software at this point. Its taken long enough to get my head around slic3r

I would like to learn to get slic3r working better though. As it says in the title.


Anyone else got some suggestions? Preferably someone that has the time to read my post's and the thread title?

Someone that actually uses Slic3er.

Getting pretty tired of repeating myself....

First time i've been bitten by a Mongrel Shark,
The title of the post Too much Z acceleration with Slic3r, though your post talks about many issues,

"i3 reprapy-strapy thingy I made. There may also be some small mechanical tolerances, its a wobbly piece of c**p
Someone suggested it might be slic3r?

I suggested changing software to see if that sorts it.
Matter Control is easy, I would be changing my Z settings more in the firmware in the IDE, so I knew changes where made,
but your on linux i missed that

"Z screws keep getting out of sync after a few prints. sounds nasty
I thought I was clever and made more changes and screwed it all up.
here is a standard bad print. I'd show the one good one, but I was proudly showing everyone and can't find it for a photo...
I'm good at Electronics and Mechanics, but my coding sucks. Hopefully its something obvious?
Also if I slow the homing down too much it stops reading thermistors while in transit and takes 40 seconds to home z
so ends up killing the printer due to heat error... So I would like to get this better.
Found an escaping Z bearing rod. The one with the endstop on it.... That was definitely bad. No wonder I had to keep re-setting the bed gap...

This way the nozzle will be clear of the bed on the way from home to print start. Because if printing 0.1 layers and the bed is not perfectly flat, its dragging from home corner to start of print. Eventually this will wear nozzles and beds.... Getting the bed that perfectly flat is unlikely while I am using ABS juice on kapton..... Not very optimistic about finding glass that flat either. Even then ABS juice is adding significant variation of approx +/- 0.05mm.. I am currently getting around 0.1mm bed variations by homing Z 0.1 low (very firm on copy paper feeler gauge, bed springs with minimal pre-load) then printing a 0.3 first layer at 150% width that is actually printing at closer to 0.2 above the bed. Makes for a little trimming on some parts, like gears, due to first layer being 0.5mm wider like a brim. but its worth it for the better first layer consistency and adhesion. I just don't like the way the nozzle (lightly) drags over the bed in pre-print moves... I would also like to add some more pre-load to the bed springs as I think they are so loose its a source of slight play, and sometimes the nozzle pushes the bed down a bit and it binds on the mounting screw and does not spring back. But if I run more spring tension the nozzle can get blocked on the first layer (really nasty cheap Chinese hot-end) "

NOW Breath...

the pic of the item you printed at 300%% speed seems good considering it must have been around 180-200mm/s
"The slower bits are a little better. 100% is 60-80mm/s for most of it and its a wobbly old thing. I don't expect good results faster than 80mm/s..."

"Slowing it down past 50mm/s makes little to no difference. As I stated the first time it was suggested. ( did try it and again today after the second time it was suggested.)"

So maybe the issues are not speed and use your electro-mechanical knowledge to figure out whats going wrong

"Why is it every time I have a problem for a few days I can't solve. That if I post here. I immediately solve it by myself?"


So.....what u waiting for...fix it!

Edited 1 time(s). Last edit at 09/05/2016 11:44PM by MechaBits.

Quote
Mongrel_Shark
Z screws keep getting out of sync after a few prints

I had this problem. It came down to a number of things.

1. One of my couplings was slipping. I used to have Nophead's design then I switched to these. I tried the split aluminum ones for a bit but didn't like them. I suspect that since it all hangs from the motors on my setup, that there was an amount of bounce with them. Switching to the rigid has taken care of that.

2. Make sure your vertical guide rails are absolutely parallel to each other. If they're not, they'll bind up and get out of sync.

3. Gotta get your Z steps per mm right. You shouldn't have to guess. You know what motors, microstep, and what the pitch of your screws are. Use the Prusa Calculator.

4. Z speeds in Firmware. I've found that you have to be pretty much dead perfect with the above to even get close to 5 mm/sec for the Z. This being too high also caused my coupler to slip. The motor would just turn in the coupler since it physically couldn't turn the screw that quickly. I'd suggest:
#define HOMING_FEEDRATE_Z (.5*60) <-- That's point 5 = a much more reasonable 30 mm/min
#define DEFAULT_MAX_FEEDRATE {300, 300, .5, 200} // (mm/sec) <-- Again point 5
#define DEFAULT_MAX_ACCELERATION {3000,3000,100,10000} <-- 100 works here

Id say flash that setup
If you've got eeprom enabled, send M502 followed by M500

As far as Slic3r goes
What's your start and end gcode?

This line is usually default
G1 F5000 Z5

While I feel that going to 5mm above the bed may be a good idea to avoid hitting clips holding glass and whatnot, they've chosen a stupid, arbitrary high feed rate of 5000 mm/min
I wouldn't call it kludgey to set that to a reasonable value your machine can do. If you want it to do that. Like you said, I found it to be unnecessary. I removed it entirely.

Edited 2 time(s). Last edit at 09/06/2016 01:01PM by FA-MAS.

Ok so I have the Z accel sort of sorted. There is indeed electromechanical issues. Mainly a slipping clamp. Or rather a heatshrinked rubberhose joiner with no clamp.. Facepalm. Will print a new clamp soon and maybe get z max feed back up around 30mm/s. for now its only getting out by a tiny bit (aprox 0.1 every 50mm or so) and not causing serious problems. Aside from the odd printer kill due to no thermistor reading during z home travel.

The speeds I thought I was doing were no where near as high as I thought due to really low accel settings in firmware. Setting perimiter to 180mm/s was only printing at maybe 40-50mm/s due to not enough room to go faster... Setting it to 40mm/s = a 20mm cube with 2 shells and 2% infil taking 40-50 min to print! Slic3rs accel override highlighted the problem for me.

Had a play with slic3rs accel override settings. While it was useful to do. It mostly screwed a lot of stuff up, as no way to set different accel for print and non-print. So I spent 2 days trying to get rid of huge dags due (I think) to slow x-y&retract accel combined with extruder temps between 270 and 290 (did I mention I'm using abs?). The temp was around 240 as of my last post but this caused major warping and splitting. Still experimenting here. Going to learn to manually edit g-code soon and do a heat tower. I think my nozzle temp may be somewhat lower than heatblock temps... Its hard to k-probe accurately and its a craptastic Chinese hotend...

Printing faster is better than slower with the temps everything seems to work best at. If I slow down too much I get drooping, sagging and blockages. (blockage may be retraction, working on that now.)

Just turned Slic3rs accel control off and improved my firmware settings about 14 hours ago. I don't know what I did the first time, but its not super accelerating anymore.
Still running 2mm lift in custom g-code. Just so the nozzle heats off the bed if it needs to wait. I would still like to work out how to maintain this lift to start of print. As is, it drops down in home corner and shoots to start of print (fairly fast with latest x/y accel settings) at 0.35 or whatever my first layer hight is. It wont hit anything doing this, but it still makes me nervous.

I don't know what I did to my old firmware, but starting with a freshly unziped version and entering in my settings stopped Z exceeding the 5mm/s max feedrate.
So it was probably a PEBKAC error all along. Likely due to poor understanding of the maths involved (I'm good at maths but marlin is too complicated for me, I have no idea what library does what, or what lines of code are referring to another folder.)

I still have trying other software on my to do list. Got a copy of matter control downloaded. Just need to install. Have cura installed, but need to either change firmware baud rate or work out how to export g-code to be able to use it.(I'll work it out when I get time to have a play don't really like cura much)
I suspect that will offer some improvement. Although I have seen a few other people mention similar problems with slic3r and think it would be useful to solve.

At this point I think I have the unwanted Z accel problem fixed/patched. Was a combination of the custom g-code and something silly somewhere in my firmware.
the out of sync problem was mostly the poor clamp, and maybe a little bit the longer wires leading to one motor. I'll report back once the clamping is sorted. If wire length is a problem that will be easy to fix once its the only problem.


I could use some more speed advice though. My homing feedrate seems fine at 150/1 I still don't understand why a multiplier was used here. I think it was multiplying with other settings. I would love to better understand this if someone can point me to an article or any other parts of the firmware dealing with speed/accel. Same for the other accel rates. It seems a big part of the problem was me messing with stuff I understood poorly.

My current firmware is as follows.

Home max feedrate 150/1

steps per unit x&y 80.3, z379, e427. These are very close. within half a percent over 100mm. I'll get it better later. Half a percent too big is what I'm aiming for. Leaves a bit of meat for finishing...

Max feedrate 400, 400, 5, 300.
Max accel 3000, 3000, 5, 10000. I have been thinking about trying 1 on z. .5 could be interesting too. Although I think unnecessary with the multiplier removed from the max home rate. That may well be the source of the problem. I don't understand it enough to be sure. Is it an "accelerate too this speed" figure, or a multiplier ie max feedrate*accel rate?

Default accel 1000
max retract accel 3000
travel accel 2000

I also turned z jerk down to 0.05. I think that may have helped too. Need to go through a bunch of prints just changing one thing at a time in firmware to confirm exactly. Not sure I will find time. Would like to though so I can give something back to the community. Seen slic3r accel issues mentioned a bit in my googles, mostly bad layers like I had, and no solution I could find.

Right now I am looking at 2 other issues. I can probably work it out, and will start another thread for each if required. Although a quick tip may save me a few hours.

A. I am having trouble balancing flow rate between perimeter and solid infil/top fill. If I go up around 440 e steps. I get great solid layers, but over extrusion on the walls, which was the cause of the banding in the cube above (the one with the small gear in the pic). Turning e-steps down to 425-430 makes lovely walls, but a slightly gappy solid fill. Tried adjusting temp and speeds (one at a time) and had some results. But this leaves the solid infil so slow its causes blockages or the perimeter so fast its under extruding after retractions.

Not too worried about it at this point. Working on retractions first as it seems related.

B. retraction. I think my high temp is part of the problem. Its either leaving big dags everywhere with less than 2mm retract. or sucking up air with retracts over 2.25mm and then, either under extruding on start, blows big bubbles (they pop and crackle, at first I thought the plastic was boiling, then I realised it only happens after retracts.) or sometimes just over extrudes. My problem is it seems to be random. Some retracts suck air and some don’t. If it does suck air sometimes it blows a bubble and sometimes it doesn’t. I have looked for temp variation here, and there is a correlation with extruder fluctuation, but I don't think thats all of it. The type of layer it was doing and is about to do seem important too. Things with lots of retracts and short extrusions seem to suck the most air to the point it stops extruding at all. If it gets some good fast extrusions in over a bit of distance then it goes well with 2.75 retract and 0.25-0.275 extra length.
I have gone from 0.5 to 5 retract in 0.5 incriments. The magic number is around 2-3. Hopefully closer to 2. I was running 8 for some reason??
I have also tried a variety of exra length on restart settings. First none then progressing up to 0.3 in 0.1 steps, then some fine tuning back and forth with 0.01 and 0.005 steps, but more by intuition/guestimation rather than systematicly. As the 3 cubes I was printing took 2 hours a print and I was dialing in flow rate as well (I know, one setting at a time) I don't think I have tried all the extra lengths with all the retract lengths. Mostly been doing extra lengths between 2&3mm retract. Its not needed below 1.75 retract. I think the magic number for extra length is going to be somewhere around 0.2-0.25

I'm still in the early stages of calibrating this with my improved firmware. About to load up a faster retraction calibration test. Maybe one with bridges too as I got them pretty good on the 3 cube print yesterday (1-2%infil on xyz cubes).

I can probably work out the flow rate and retraction problems with a few more prints and some googling. Lots of good guides etc out there.

The Thing I mostly want help with is understanding those speed/accel maths better. Then I can probably mark this thread as solved and stop drifting off topic

If I don't learn, I'll break it again... I couldn't learn to drive a manual gearbox until I re-built one... Same for most other things I have operated over the years. If I don't know how it works I end up crashing or breaking it..... Still can't drive an Auto gearbox (well I can for a bit, then it breaks or falls off the car)!


P.s. Mechabits. Sorry about my tone earlier. I was very frustrated and lacking sleep (still not had much). I should have thought that through a bit more. You are only trying to help. Thank you for being so patient and tolerant!

P.p.s. I think writing out my problem and re-reading it is a big part of the reason I make big improvements after posting here

P.p.p.s Attached are pics on my last two prints and the bucket of bolts that is my printer. The Wii box in the background is a functioning part of the printer. Blocking out drafts from the building layout and the ATX fan. It also has bits of soda can, used plumbing and old pallets as functional parts lol. 4 weeks ago I would have said this machine would never print this well. Now I think it can do a lot better. Its amazing something so wonky can work so well. Seriously, apply about 5 grams of pressure to any part and it will deflect a bit....
I was going to fix a bunch of stuff on it, but then I realised if I fixed everything up good. the only original parts left would be 5 motors, the hotbed and the arduino (RAMPS board had a small fire and is missing some traces). So once the flow rate and retraction issues are sorted. I'm welding up a better frame and printing a better printer At which point I'll frame this as a memento or burn it. Haven't decided which yet. Don't tell Phil lol.. He still thinks he's getting it back lol..

This page has some information on how the planner works that helped me understand accelerations. http://www.extrudable.me/2013/04/02/the-myth-of-z-speed/

Thanks for that link FA-MAS. It was an unteristing read, but unfortunately didn't get me any closer to finding the maths I need or how to edit the maths in firmware.

Probably not an issue anyway.

Examining the g-code from slicer quickly revealed the problem.

Every Z move is set at stupid high speeds. F10200 for my Slic3r settings!! If the move is under 2mm the acceleration in firmware efectivly slows it down enough to not stall out the motor. But its still pretty dangerious.

If I change non-print move speed in slicer then it slows X-Y and I get huge dags everywhere.

I changed travel (non-print) speeds in slicer to 50mm/s and it outputs F3000 for all x y z moves.

I'm assuming this is steps and its assuming the a4988 is set to 16th steps like most x-y axisis.
Except all the setup guides say to use quarter or eighth steps on a screw driven Z.... Which makes sense, because no one needs 50nm steps. Thats just silly.
No wonder this is such a common problem!

So one kludge/fix would be to run 16th steps on z.

Another Kludge would be to set the accel rate for z so low that greater distances can be ordered at high speed and it wont get to that speed.

Ultimately though. There needs to be a way to set a separate Z axis speed in Slic3r, or ideally a way to tell slicer what step rates are acceptable.

The other thing that is bothering me. Is Why is the gcode overiding the firmware maximum? Why even have a firmware max if its not going to be observed??

I guess I should take this over to the slic3r chat on github or somewhere? So the developers can try to fix it? Where would be the best place?

I would also like to stop slicer from exceeding the max speeds I put in firmware.

Here is the start of the G-code:

; generated by Slic3r 1.2.9 on 2016-09-12 at 13:34:40

; external perimeters extrusion width = 0.40mm
; perimeters extrusion width = 0.40mm
; infill extrusion width = 0.40mm
; solid infill extrusion width = 0.40mm
; top infill extrusion width = 0.40mm

M190 S130 ; set bed temperature
M104 S255 ; set temperature
G28 ; home all axes
G1 F120 Z2 ; lift nozzle

M109 S255 ; wait for temperature to be reached
G21 ; set units to millimeters
G90 ; use absolute coordinates
M82 ; use absolute distances for extrusion
G92 E0
M106 S56.1
G1 E-2.00000 F8400.00000
G92 E0
G1 Z0.200 F10200.000
G1 X84.484 Y84.756 F10200.000
G1 E2.00000 F8400.00000
G1 X86.659 Y83.041 E2.04924 F1500.000

I'm not seeing anything there that should cause an override. Perhaps I did something silly in firmware? Its not making any sense.