Skeinforge 40 Changes to Dimension - Read this for problems with the new version of Skeinforge!

I just posted this in another thread here, but I figured a lot of people would have the same issue I did so I'll give it its own thread. As soon as I upgraded to Skeinforge 40, suddenly I was getting almost nothing coming out of my extruder. This is why:

Quote

The Dimension plugin changed for skeinforge 40. You need to change your firmware e_steps_per_mm to steps per mm of feedstock, rather than steps per mm of extrudate. Also, you should have your flowrate and feedrate set to the same value. Then measure your feedstock diameter and input that, and the filament packing ratio for ABS should be about 0.85 and for PLA should be 1.0. Also, you'll need to change the extruder retraction speed, as that is now in mm/s of feedstock as well, so it should be much smaller than it was before. Same goes for retraction distance.

The whole idea is that now skeinforge knows how much plastic is going into your extruder, and therefore knows how much is coming out. Also, you won't have to adjust your firmware or the skeinforge settings when using different nozzle sizes.

I just spent the past couple hours figuring that all out, can you tell? I had to track back from [fabmetheus.blogspot.com] to a post on a different blog to figure out what was going on.

Edited 1 time(s). Last edit at 03/02/2011 01:40AM by NewPerfection.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

Simple documentation is the answer....but that can only come from the author right?

If Enrique issued a tracking list of the features changed, enhanced, moved or deleted it would make the program a lot easier to follow and more people would give him the credit deserved.

---

Cheers,

Paul

That makes sense. I just kept increasing the flow rate until it came out right. It was never clear to me what the units for the flow rate is. It's not labeled. So, with the new method, is it now the same as the feed rate, mm/s of extruded plastic? What would you adjust if you change the layer thickness? In theory, since it has all the necessary information, you shouldn't need to change anything except the layer height. So now, do you change the feed filament size, packing ratio or flow rate to get the perimeter width to come our right?

Now when you change the layer height all you change is the width over thickness value. Packing ration, feed filament size, and flow rate should always be constant. The layer height and width over thickness are now the only values that control how much plastic is laid down. You can change the flow rate to adjust the amount up or down, but if you do you should just make that % adjustment to your e_steps_per_mm to keep the flow rate and feed rate identical.

This new way of doing it makes much more sense, since once you get your firmware and skeinforge settings tuned, the only values you should ever need to change for a given extruder and plastic type is layer height and width over thickness, even if you change nozzle sizes.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

Yes exactly as I have been doing from day one.

---

[www.hydraraptor.blogspot.com]

If someone has the time to put this info on the Skeinforge Troubleshooting section on the wiki, that would be great. Unfortunately I don't have time at the moment.

Also, any other simple tips and troubleshooting tips for Skeinforge would be welcome.

i haven't tried it yet, but if i did my calculations right, here are my results

*deleted* embarrasingly wrong math

probably should go to sleep.

*next morning*
ok well last night i ended up getting a number around 70 steps per mm in. Black PLA, Wades extruder, .9deg steppers at 1/8 microstepping (same number as 1.8 at 1/16)

Does this jive with anyone else? the gears are moving so slow you really have to watch them to see that they are moving at all. (i'm not even sure they are moving, or i just imagine they're moving) The only time the gears really visibly move is when it retracts. i have it retract 11 mm, but i understand that the retraction is now mm into the extruder, and so what i actually get is only about 1-2 mm retraction. seems about 10x too small.

Do i need to have "add flow rate" checked under speed? Shouldn't it just be an E distance now?

Edited 2 time(s). Last edit at 03/03/2011 11:42AM by Buback.

You have to have "add flow rate" checked, otherwise Dimension does nothing. Also, 1 mm movement in E is 1 mm of the feedstock, it has nothing to do with how much plastic goes in or out of your nozzle. This is the big change in 40 vs 39. Read this for a more thorough explanation.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

NewPerfection Wrote:
-------------------------------------------------------
> You have to have "add flow rate" checked,
> otherwise Dimension does nothing. Also, 1 mm
> movement in E is 1 mm of the feedstock, it has
> nothing to do with how much plastic goes in or out
> of your nozzle. This is the big change in 40 vs
> 39. Read this for a more thorough explanation.

sorry, that's what i meant when i said "into the extruder". i meant it would retract and then return, pulling the feedstock Into the extruder, but i see that that is a really confusing way to put it.

As i understand the change, we now have to set our firmware steps_per_mm as being steps_per_mm(of feedstock). then, we put some info into dimension regarding our feedstock properties, and then dimension calculates how many E steps are needed to produce the width over thickness that we put into carve, based on our layer thickness.

a wades hobbed 8mm bolt is approx 6.5-7.5 mm diamater, which has a circumference of ~20-25mm. so 1 rev of the big gear gets ~20-25mm feedstock fed into the extruder. so i'm coming up with around ~70 steps per mm fed into the extruder. am i way off?

NewPerfection Wrote:
-------------------------------------------------------
> Now when you change the layer height all you
> change is the width over thickness value. Packing
> ration, feed filament size, and flow rate should
> always be constant. The layer height and width
> over thickness are now the only values that
> control how much plastic is laid down. You can
> change the flow rate to adjust the amount up or
> down, but if you do you should just make that %
> adjustment to your e_steps_per_mm to keep the flow
> rate and feed rate identical.
>
> This new way of doing it makes much more sense,
> since once you get your firmware and skeinforge
> settings tuned, the only values you should ever
> need to change for a given extruder and plastic
> type is layer height and width over thickness,
> even if you change nozzle sizes.

Actually, with 040, you do have to change the flow rate when you change layer height just like in the older versions. The only difference that I can see is that if you don't change the firmware value to steps per mm of feed from steps per mm of extruded filament, then the flow rates are larger by the ratio (feed diameter/nozzle size)^2. Thus the flow rates I'm now getting are 36x larger than what I used to get.

If I change my firmware values by some factor like (2.8/0.5)^2, then I would have to change the flow rate and retract distance by the same factor.

So, except for getting flow rates that make more sense, changing the firmware is not really necessary.

Why would you want to change the width over thickness ratio when you change layer height? I just leave it at 1.5 regardless on whether my thickness is 0.4 or 0.3 mm.

Someone please correct me if I'm wrong on this.

Quote
brnrd
Why would you want to change the width over thickness ratio when you change layer height? I just leave it at 1.5 regardless on whether my thickness is 0.4 or 0.3 mm.

That's not too good of an idea. Ideally you want to be moving at the same speed that you are extruding. If you increase layer height but leave width over thickness the same, you have to increase flow rate more than feed rate, which causes squishing of the filament output. This makes overhangs and bridges really difficult, if not impossible, to print.

Ideally, you should always have feed rate and flow rate set to the same value. To do this you always want your output cross section to be the same as your nozzle cross section (allowing for die swell/shrink). So if you increase layer height, you want to decrease width over thickness, and if you decrease layer height you want to increase width over thickness.

Edited 1 time(s). Last edit at 03/05/2011 01:22PM by NewPerfection.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

I have always done the same as brnrd: left the w/t the same when changing layer height and adjusted the flow rate by the square of the change in layer height. I can see that this is sub-optimal for bridges (and indeed with 0.25 layers, many of my bridges snap on the first crossing attempt), but without bridges, the print resolution is improved by extruding a smaller cross-section. I tend to use 0.25mm layers for pulleys and gears which dont tend to have much in the way of bridges (other than for set screws) and 0.4 for everything else.

I have also always just used 1.8 as PWOT... my fine stuff at .25mm layers have smaller width strings... and I like the detail.. I just adjust my e-steps per mm till I get good bottom fill and the perimeter is about right (always a bit thicker).. and I get reasonable loose fill...

I'll try to do it this other way and see what benifit I get...

Al...

---

[araspitfire.blogspot.com]

my experiments with SF 040 thus far:

with my e_steps_per_mm at ~70, i have to set my flow rate at (feedrate * 4) to get anything near usable. the fill is barely there. i'm going to try (feedrate * 5) next, but i think my e_steps_per_mm rate is way off if flowrate is now supposed to be (feedrate * 1).

also, when you change the e_steps_per_mm, most softwares are still going by extruded length (not the feedstock length) if you hit the manual extrude button.

-----
-Later-
I don't know if i have a decimal point misplaced or what, but i'm getting good results at flowrate = (feedrate * 10)

Edited 2 time(s). Last edit at 03/10/2011 11:27PM by Buback.

The steps per mm will be approximately micro steps per rev * gear ratio / ( pinch wheel diameter * pi).

For a Wades and x16 driver it would be 3200 * 39/11 / (5 * pi) = 722, so yes it seems your figure could be 10 times too small.

---

[www.hydraraptor.blogspot.com]

Thanks for double checking my math. it must be that i forgot to include the microstepping, and for some reason used full steps. my prints are much much better now; close to 039 already.

I took most of the week off from printing because i knew i'd have to print lots of calibration pieces to figure out my problem, and it would be a night lost, but the skirt plugin is great and well worth it.

The only unknown variable in the calibration now is the effective diameter of the pinch wheel. That can be measured by extruding a fixed length of feedstock and measuring how much filament is dragged into the extruder. That should be the only calibration needed to get good prints (apart from temperature). No need to print any calibration pieces at all.

---

[www.hydraraptor.blogspot.com]

I spent some time last night trying converting to the sf40 way of doing things.. I'm still struggling to get where I was with sf39 (on a Mendel with RAMPS)

I very carefully measured my extrusion distance per rotation with an M8 hobbed bolt (in a Rhys herringbone extruder).. came up with 20mm per rotation, which is 529 steps per mm (1/16)

Carefully measured and input both PLA and ABS diameters over the evening..

I'm using a tonokip's fork (tesla's) and I quickly found that I'm over-extruding for fill layers for .4mm layers at 1.8 PWOT.. and even the perimeter is a bit over... bits of perimeter filament buldging over 4 to 10 mm lengths as if it want's to flip out.. (too much flow IMHO)

I've been using feed = flow for some time. and am used to adjusting the E steps per mm on the fly in tonokips... I am used to getting quite good looking prints.

I changed my first layer flow rate fill multiplyer and the infill width over thickness to get my prints under control.. but I really think I should just turn down my E steps per mm a bit more...

I measured all the measurables.... and I extrude a block.. it's overstuffed... so what is the 'safe' parameter to adjust?

I would really like to know what I should change to make it so that when I set my layers to .25 (next step for me once I get .4mm tuned) I don't have to fiddle with E-Steps per mm to make it right..

I really want to try to do sf this time the 'way it should be done'.. as opposed to making it work with fudges that I know have un-intended consequences that I may not notice till I print something were that factor is important...

Also.. when I change my layer height... does sf reduce flow by itself? same for PWOT? It's not clear how this is supposed to work for me..

(some examples of my prints from sf39 [araspitfire.smugmug.com])

Al...

---

[araspitfire.blogspot.com]

I made a wiki page to help people get started with the new settings : Volumetric_Dimension_settings. Please feel free to correct anything I got wrong.

It might be a good idea to post links to this new page from the extruder wiki pages, as well.

nophead Wrote:
-------------------------------------------------------
> No need to print any calibration
> pieces at all.

That's the idea. Once I switched to SF40 and got my e_steps_per_mm right, my first print was better than I had ever gotten before. I love it.

---

Help improve the RepRap wiki! Just click "Edit" in the top-right corner of the page and start typing. Anyone can edit the wiki!

So when you are measuring your pinch wheel diameter, is it to the outside of the teeth or the inside?

I think it should be the outside...

cheers

It depends how far the plastic sinks in. I think with ABS it is the inside of the teeth but PLA sees a larger diameter. This difference is more extreme the smaller the pinch wheel is.

The only way to get an accurate measurement is to measure how much a given number of revolutions feeds into the extruder.

---

[www.hydraraptor.blogspot.com]

Ok.. I checked again.. I'm moving exactly the right amount of feedstock with the given E-steps 1 exact rotation takes 529 steps. Repsnapper when told to feed 20mm rotates the big gear 1 rotation, exactly, and feeds 20mm....

But I'm still too high with the flow... what should I tune in sf that won't wreck the way it's supposed to work now?

I feel like I'm missing something here..

Al...

Edited 1 time(s). Last edit at 03/11/2011 06:44PM by araspitfire.

---

[araspitfire.blogspot.com]

Have you entered the correct filament diameter? Apart from that I can't think of anything else that needs to be set. I don't use this version myself but my own software works this way, i.e. uses maths to work out the flow from the feed rate.

---

[www.hydraraptor.blogspot.com]

In response to my asking whether the inside or outside of the teeth is used in determining the diameter

nophead Wrote:
-------------------------------------------------------
> It depends how far the plastic sinks in. I think
> with ABS it is the inside of the teeth but PLA
> sees a larger diameter. This difference is more
> extreme the smaller the pinch wheel is.

I would think that if the teeth do not slip then although it is not feeding from the outside of the filament, but maybe 1mm in, the linear motion of the feedstock is still that of the tip of the teeth... or am I missing something?

You could imagine so, but you could also consider the outside of the filament runs in the bottom of the teeth so the linear distance that moves would be less than it is for the inside of the filament. In practice I suspect it is somewhere in between the two, that is why I maintain that you need to measure the feed rather than the diameter to get an accurate figure.

A few people have said that ABS changes in density as it is extruded but PLA does not. I can't find any evidence for that, but what I can say is that the same pinch wheel extrudes them at different rates. I can explain this by the fact that PLA sits high on the teeth and ABS sinks in. I find that easier to believe than some of the ABS magically disappears or its density changes.

I have been doing some work on this recently, so I should blog it as there seems to be a lull in demand for printed parts at the moment.

---

[www.hydraraptor.blogspot.com]

wohoo - looking forward to another of your quality blogs Nophead.

Stupid question for those in the know - How do I go about instructing my extruder to do a full rotation? How do I go about sending 200 step instructions (or multiples thereof to ease measurement)?

Cheers

This is not working for me.

I calculate my steps per mm at 79.98

I am using the standard reprap firmware and a 7.96mm diameter pinch wheel. gear ratio is the old Ardian's ratio of 55/11

Under 39 my stepper calculations were based on half stepping so that is what I used here.

??!! I hardly get anything extruded.

What is the default step ratio (full, half 1/8, 1/16) for the standard reprap firmware with a G3 board??

Also, where is this 'add flow rate' option NewPerfection is talking about that has to be checked? [edit2] found it - on Speed tab[/edit]

[EDIT] - erm... set the steps for the wrong motherboard so was actually operating with 2.22 steps per mm!!

still getting rubbish results though seemed to do the first layer of the raft well enough, but then the motor started jittering etc. and feed was really bad [/EDIT]

Edited 2 time(s). Last edit at 03/12/2011 06:20PM by AgeingHippy.

Ok.. I've got this working.. and I am sold...

For the info of people scratching their heads.. a brief tutorial follows... I'll post something with pictures on the Mendelparts forum in a couple days for the plug and play people...

I entered very carefully a measured E steps per mm using ABS, as I mostly print in ABS. I also measured the diameter of feedstock very accurately..

The way I measure the steps per mm of feedstock is make a mark on it more than 100mm up from the top of the extruder block... tell repsnapper to advance 100.. measure how much it moves... take current E-steps per mm, divide by how much it moved... multiply by 100.. this is your new E steps per mm of feedstock.. I recommend you set the firmware to 80% of this and do it again to get better accuracy as the initial movement will be very small...

Also.. best to do this with the hardest PLA if you are using both.. I'll discuss this below.

I was getting too much flow initially and it was because I was using ABS and left the Filament Packing Density (ratio) in dimension at the default .85... of course, because I had just measured the actual distance... this ratio needs to be 1 (you just measured it...)

So... start with PLA... and the ratio can be 1 as instructed. Then stick ABS in and make a mark 100 mm up... extrude it 100mm in repsnapper (or whatever host you like) and measure how far it moved.. distance moved/100.. is your Filament Packing Density for that hardness of ABS (and soft PLA)

BTW.. I did this through the heated hot-end.. in case there was any slippage effect from pushing against pressure... I don't collect any plastic on my hobb.. but it's a good practice I think.

So.. Now I have the 3 numbers in sf40 that describe my filament volume to skeinforge...

I STILL had too dense infill on the solid fill layers...but... my perimeter width, as measured on a hollow test cube, was very good on the calculation: Layer Height times Perimeter Width Over Thickness (ratio)

So.. I raised the Infil Width over Thickness Ratio from the default 1.5 to 1.7.. this made the solid infill paths further appart... and cured my over-stuffed full fill...

I also changed Additional height over layer thickness in bottom to .7 from .5 and activated it, made sure to activate Limit and set my maximum Z speed and set maximum initial feed rate to equal my feed rate, or slightly lower

I made some other minor changes to suit my preferences, but none that effect the main print geometry... I set my retract to 2mm and speed = 30ish (as fast as possible and as short as necessary is best)

I can report that I only change the layer height and print... the amount of plastic extruded is indeed calculated by sf and works... I have printed the same objects in .4, .3 and .25mm layers... I need to increase PWOT for the .25 layers, because, as mentioned... my strings on the sparse fill get fragile...

So.. tuning SF has gone from an art, to engineering, in one version... measure your input parameters accurately and print....

If you have trouble.. things seem too sparse... try again after changing the feed to a lower number ... (say 30 rather than 40) (of course flow equals feed now)

if it all cleans up and gets better.. you were overdriving your hot-end... trying to push too much plastic... lower your layer height and you can speed it back up... (less plastic being extruded)


I'm a happy bunny... I used to spend some time tuning my E-steps per mm (I set it with a gcode in tonokip's fw) for every layer height change, and even feed/flow changes put my prints off and needed to be tuned...

Now I have a tuned setting and can change layer height with one or two numbers and print... easy..

Hope this helps some people...

Al...

---

[araspitfire.blogspot.com]

I think that certainly will be a big help Al.

I mentioned earlier that my extruder would start jittering/stuttering after a while.

I found that my replace.csv file was not being actioned and therefore the power to the extruder was being taken from the M113 0.75 code rather than the pot (an M113 code without a parameter, which is what my replace file replaces the M113 0.75 section with).

It turns out when I upgraded to Sf40 the name of the replace file (on tab export) was changed to 'value'.

I have changed it back to replace.csv, and along with the required updates to my start.gcode file (to prime the extruder with the correct amount and at a much slower F (feedrate) value) it seems my test cube is pretty good.

I will use Al's technique to accurately measure my filament to determine the e_steps_per_mm tomorrow, but we are getting there

(touch wood) my extruder at least now seems fairly reliable given I have replaced the thermal break with an inner diameter of 3.5mm with a thermal break having inner diameter 3.1 which is the same as my barrel.

now I need to figure out why at the start of each thread it extrudes I get a blob - anyone have any ideas?