Jump to content
Ultimaker Community of 3D Printing Experts
Sign in to follow this  
rich

Experiments on Bridging

Recommended Posts

I had a problem bridging with my UM1 so I decided to experiment with variables such as extruder temperature, print speed, extrude rate, fan on/off, and separation between threads in the first layer.

The model used was a simple box with no lid. The bridging threads are approx 46mm in length.

Cura 13.11 was used to generate G-Code for the box and a spread sheet was used to generate code for the bridging layer.

The material used was ABS. For normal printing I have been getting good results with the extruder at 245C. Although other values were used for the tests (from 243C to 249C) the temperature changes did not have as much effect as some of the other variables.

A total of 48 threads were printed in each experiment and extrusion rate (ie dE/dL), print speed (F) and fan state were varied systematically.

The results seem to indicate that print speed has the biggest effect, extrude rate next then the fan. The image below shows the result of one experiment, with fan constantly at 100%, extrusion rate dE/dL (Vol/mm) = .0069 for first 24 threads then .0077 for the rest and Print Speed 20 mm/s for first 12 then 40 mm/s for next 12 then 20 etc.

 

BridgingExpt

Printing from lower right to left.

BridgingExptCut

The thinness at start of each bridge may be affected by overlap on wall.

The best results are for print speeds around 20 to 40 mm/s with extrusion rate around 0.0069 vol/mm (not sure what units of vol are used in Cura for this quantity).

There are several observations I would like to make.

1. For well spaced threads, those that fail to survive do so because their thickness drops to zero just after leaving the starting wall (not the destination wall). This seems to be reduced if the overlap on the wall is reduced.

2. For closely spaced threads (ie. where the hot nozzle may contact the previous thread) failure rate is very high at high print speeds but at low print speeds 2 or 3 adjacent threads tend to amalgamate to form one.

Having found some patterns in this behaviour I would like to have a little more control over Cura for bridging layers. Can anyone suggest how I might do this? For example is there some way to have Cura place comments in the g-code at the beginning and end of bridging?

Any comments please?

PS. I can send images directly to interested parties.

 

edit by gr5: I added your images inline - hope you don't mind. message me if you want me to return to original state.

 

Share this post


Link to post
Share on other sites

I'm very interested in this research, and the pics ;)

Why not create a gallery for them in the forum? Then you can put them into your post..

I've had problems with much shorter bridging parts, which is why for now I always construct my designs without any bridging, or with as little as possible...

 

Share this post


Link to post
Share on other sites

Great post - I rarely edit a post but I did yours - hope you don't mind. Let me know if you want me to revert to original. I did some bridging tests also recently but your zoom-in-photo is fascinating! It makes me think Cura should output a little extra flow for the first 2 millimeters maybe?

Currently post #17 (but that will likely change within a month):

http://umforum.ultimaker.com/index.php?/topic/1872-some-calibration-photographs/?p=25304

 

Share this post


Link to post
Share on other sites

1. For well spaced threads, those that fail to survive do so because their thickness drops to zero just after leaving the starting wall (not the destination wall). This seems to be reduced if the overlap on the wall is reduced.

2. For closely spaced threads (ie. where the hot nozzle may contact the previous thread) failure rate is very high at high print speeds but at low print speeds 2 or 3 adjacent threads tend to amalgamate to form one.

 

Wow. First of all, your photos don't show #2. Did you do many more tests? Can you upload more pictures to the same album - put notes with each picture that describes what it is showing (if you have these pictures already).

 

For example is there some way to have Cura place comments in the g-code at the beginning and end of bridging?

 

Oh! Daid please! Please add comments to show each individual bridge line! That way this guy can write a plugin to do some experiments to improve bridging string strength!

 

Share this post


Link to post
Share on other sites

Your post gives me all kinds of ideas. It's common to have under-extrusion during acceleration (start of a string) and over extrusion at the end. This is caused by the delay between when you ask for more filament and when it comes out (delay mostly caused by bowden tube and other stretchy/springy physical elements).

This "error" is reduced if you print at slower speeds but should also be reduced if you greatly lower acceleration parameters. So I'm thinking that simply lowering the speed when you get near a bridge would help. Also lowering the XY acceleration for the section of print that involves a bridge might help. You want to lower it a LOT. I would lower it to about 1/10 of the current value for a test.

I have many experiments on my list right now so I might never get to this one, but this is very intriguing!

 

Share this post


Link to post
Share on other sites

Jonny - thanks for your comments re using the gallery.

gr5 - mind? Thank you. I do have more images and there are a few more experiments to come. I haven't looked at acceleration settings - excellent thought. My comment #2 is illustrated in #2 image. Counting from the right, threads 7 & 8, 9 & 10 then 11 & 12 are amalgamated (see bottom near wall).

 

Share this post


Link to post
Share on other sites

threads 7 & 8, 9 & 10 then 11 & 12 are amalgamated

 

I see that now. I thought maybe you printed closer together in another experiment (.4mm apart) to be more realistic with what Cura does. Didn't realize they would cling together like that even though so far apart.

 

Share this post


Link to post
Share on other sites

@rich and gr5:

When you did your bridging tests, did you ever have the problem that the first few lines of the bridge layer did not stick to the bridge head but curled along the nozzle?

I'm just on the way to get really nice results with bridging and this is (hopefully) the last obstacle... it occurs after I cleaned the nozzle from old PLA...

 

Share this post


Link to post
Share on other sites

Never. The first bridge line was *always* perfect. But then the second line sometimes destroys the first line.

I would look at your nozzle tip under a powerful magnifier (or just reading glasses). You may have somewhat damaged your tip.

edit: they cost < 5 euros including shipping so don't get too upset if you did.

 

Share this post


Link to post
Share on other sites

Never. The first bridge line was *always* perfect. But then the second line sometimes destroys the first line.

 

I an idea I had last week from another post about bridging. To keep subsequent bridge lines from destroying the prior lines, how about lowering the print bed slightly after starting the bridge, then dropping back down at the other side to make the connection. This would help keep the nozzle away from the prior laid lines.

Also, why doesn't the first bridging layer always pick the shortest possible unsupported span?

 

Share this post


Link to post
Share on other sites

Also, why doesn't the first bridging layer always pick the shortest possible unsupported span?

 

I think it does. Cura detects bridging and usually seems to do the shortest span - I'm not sure if that's always true but it was for my part. The part above was printed with excel, not Cura. Re-read that post.

 

To keep subsequent bridge lines from destroying the prior lines, how about lowering the print bed slightly after starting the bridge, then dropping back down at the other side to make the connection.

 

I was thinking the same thing. This might help the shape a bit also. Kind of like pre-stressed beams.

 

Share this post


Link to post
Share on other sites

OK, you know how it is... if something doesn't work as you think it should, you have to stick with it until you got the case solved...

For me, this was the case today with bridging. After I found out that the nozzle was still ok (I will anyway organize some spares, you never know what's happening next...)

and the curling of the extruded filament was due to insufficient adhesion of the bridge layer on the previous one, I found a way to produce nice bridging - even with the standard Ultimaker fan!

gallery_18010_308_68121.jpg

gallery_18010_308_4558.jpg

The model is the well known one from Thingiverse (no. 12925) with 5mm instead of 10mm height (as I wanted to see my Ultimaker doing brigdes not struts... :wink:).

Your post gives me all kinds of ideas. It's common to have under-extrusion during acceleration (start of a string) and over extrusion at the end. This is caused by the delay between when you ask for more filament and when it comes out (delay mostly caused by bowden tube and other stretchy/springy physical elements).

This "error" is reduced if you print at slower speeds but should also be reduced if you greatly lower acceleration parameters. So I'm thinking that simply lowering the speed when you get near a bridge would help. Also lowering the XY acceleration for the section of print that involves a bridge might help. You want to lower it a LOT. I would lower it to about 1/10 of the current value for a test.

gr5, you mentioned two ideas in your post. You were perfectly right with both of them (thxs). I reduced the x/y-acceleration by a factor of nearly two magnitudes for the bridge layer. And I gave the extruder an extra kick at the beginning of the bridge layer by adding a small amount of extruded material. I did it with Cura 13.11 by using an experimental TweakAtZ plugin as very special conditions are needed for the bridge layer.

However, the sticking together of the various strings is indeed a problem and for my cooling fan setup unavoidable as the standard fan has to blow in lateral direction in order to cool the strings in both directions. It would require a different cooling setup (two fans and/or special fan duct). Thus, I lowered the print temperature to the pain level, i.e. 185°C, but only for the bridge layer as I got underextrusion for the rest when printing everything with that temperature.

@rich: what cooling fan setup did you use for your tests?

 

Share this post


Link to post
Share on other sites

I don't think you can change the acceleration that precisely. Marlin plans out many segments in advance. That's why when you change the speed setting on the UC it takes a little while before anything happens. The same is true with acceleration settings. So you have to adjust accel something like 30 line segments ahead of time.

Share this post


Link to post
Share on other sites

What do you mean by precisely? Precisely in time or precisely in value?

I just had another look at a video I made during the print and I do not see a huge difference in the different movements of the bridge lines which means the acceleration changes more or less instantanously. The same thing when changing back to 'normal' acceleration: the printer runs racily again at once...

Marlin plans out many segments in advance.

I'll never play chess against my Ultimaker... :wink:

 

Share this post


Link to post
Share on other sites

>Precisely in time

Precisely in time. In other words if you send a gcode to change the acceleration (or do it from the front panel) it seems to happen about 10 line segments *after* that gcode because Marlin has already planned 10 line segments in advance.

I don't know if it's exactly 10 or if it's 20 or what. But I would guess between 10 and 30 from my experience with how long it takes to speed up or slow down when you mess with the feed rate. It's the same amount of delay.

Edit: changed "cura" to "Marlin". I meant Marlin.

 

Share this post


Link to post
Share on other sites

Now, you confused me... :wacko:

Who is the big planner: Cura or Marlin?

 

I had some further looks into the GCODE generated by Cura. The first bridge line is the second segment of the bridge layer; there is just a very short segment before on the bridge head.

This is the video I made during one of the various test print runs (don't worry about the fan noise; there are other devices than the Ultimaker running in the same room):

 

You may see or rather hear that the acceleration is indeed not excactly constant for all the bridge lines. However, the differences are quite small as the acceleration change from 9000 to 100 and back is a huge leap.

 

Share this post


Link to post
Share on other sites

Very impressive video Dim3nsioneer.

I've not been able to get the g-code M201 to have any effect on my printing. I haven't used plugins - for my test setup I add them to the g-code in a spreadsheet. You are clearly modifying Amax X and Y. I tried lowering acceleration to 800 but the printing is unaltered. Neither have I had any success with M203 - I would like to try to increase the maximum extrude rate because the bridge threads at 150 mm/s are thinner (0.1mm) than those at 20 mm/s (0.3mm) and a bit of arithmatic says the default value of Vmax e of 25 mm/min will not supply the material required at the higher speed. I'd be interested in pushing as much plastic as possible through the nozzle.

Any suggestions on why these g-code are not being executed? What version of Marlin are you using?

Re the fan .. I use the standard fan. I also use a 50W quartz halogen desk lamp just above and to the side of the printing. I've not succeeded getting it on film but with the naked eye I'm convinced it keeps the bridge threads straight and in the nozzle plane. When the model is cooling the threads buckle out of that plane and move significantly with even minor air currents.

 

Share this post


Link to post
Share on other sites

I use a Marlin version downloaded from http://marlinbuilder.robotfuzz.com/

I made the download on the 14th of October 2013 so it must be the Marlin version from the 7th of August 2013 as it is still this version there.

The plugin does nothing else than just to add 'M201 X100 Y100', as if you would insert it manually... as Cura plugins do not change the world but make it more comfortable... :cool: So the difference has to be somewhere inside Marlin... which is absolutely not my area of expertise...

If you have an Ulticontroller you could try to alter the Amax X and Y there...to see if your Ultimaker just ignores you or not.

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

×

Important Information

Terms of Use Privacy Policy