Jump to content

The crossflow fan approach


Recommended Posts

  • 1 month later...
  • Replies 142
  • Created
  • Last Reply

Top Posters In This Topic

Posted · The crossflow fan approach

After fixing some errors with my heated bed, I finally installed the cross flow fan and tested it. The setup is improvised but works for first tests.

First impression: The cross flow fan is great. I made some great prints with it. I totally like it, but my setup and settings leave room for more.

I printed out some 20mm cubes, the surface was perfect on all sides. Very fine print.

I printed out an eggcup in a round design with a slight overhang. Came out great. Best eggcup I did so far and I did a lot of them in the past.

The next test object was the maker faire robot. The one they used as test object in the latest 3D printer review of MAKE magazin. http://www.thingiverse.com/thing:40212

At the beginning the first one looked great. The legs are perfect, but when printing the first overhang, things got so so... The overhang facing the fan was very good and the ones on the side also. But the one on the backside was bad. It looked buckled.

Next try, I turned the robot 90 degrees on the build platform, with its face faceing the fan. Much better. Both arms looked very good, but the overhang on the backside, its "ass" was still buckled. This seems to be because of the lee of the model itself. The backside does not get cooled as good as the front and the sides.

All previous prints I did were with 20% fan speed. So next thing I increased the fan speed to the recommended 50%. The backside was a little better, but not yet perfect. Unfortunatedly at this speed the fan gets much louder, even a lot louder than the stepper motors.

Conclusion:

- Works good, but is still in process.

- Print quality depends on models and is also dependent on their orientation to the fan.

- Overhangs on the "backside" of models look a little messy.

Next things to do:

- Optimize the position and alignment of the fan, macbe add a duct which forms a small slit.

- Find a way to run the fan at very low RPM <30% for noise reduction, but with high print quality.

- Eliminate the problem with bad overhangs in the lee.

I got some ideas for fixing all this. Let´s see how they work.

I will add pictures tomorrow.

Markus, any news on your cross flow fan?

Regards,

Philip

 

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Hi fns,

    yes I think it´s a flow related problem, but air flow related. Which settings are you useing for the fan? As you said without fan the structure is strong, but the edges are ugly, with fan on (100%?) the objects are brittle. So the truth lies in the middle. I think the cooling is so strong that the layers do not stick together very well. They do not melt into each other.

    Did you try to print with only 50% fan speed or less? In the expert settings in Cura you will find a menu for Cooling, perhaps try 40% as fan speed min and 50% as fan speed max. I haven´t tried ABS yet, but for printing PLA with 200°C this works very good.

    Hope this helps,

    Philip

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @fns720

    Sorry to hear that. I've no experience with ABS yet. But many people say they print ABS without a fan. So I would assume the same reason as Duesentrieb. Too much cooling, no sufficient layer adhesion.

    @Duesentrieb

    Printing the maker fair robot right now. I'll report back.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Here we go. Printed with 50 mm/s, facing the fan. Printed one at a time (no cheating to optimize cooling :wink:)

    You can see a slightly lower overhang quality on the rear side of the arms but I'd consider overall quality pretty good.

     

    MakeRobot

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Thanks Philip. I tried with 40/50% fan speed but the result is too brittle. I'm trying now with 5/15% and will tell you the results.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Just a wild idea (and a bit off topic, sorry) But has anyone considered using an aquarium type pump with its associated small bore tubing to direct cooling air precisely to where it may be needed?

    Any pulses in the air flow (due to the pumps mains working cycle) could be smoothed out with a large chamber in the feed path. Like a capacitor to smooth ripple.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Hi Guys

    @fns720

    I have been getting problems with printing ABS with this fan.

    Initially I was getting very ugly overhangs on small prints on the non fan side.

    Turned up the fan and I got better overhangs but the model was splitting between layers.

    Turned up the nozzle temp to 170C and this fixed the problem.

    Then on a big print I can't get a good result.

    I am using a glue stick to hold the model to the hated bed which is working good.

    I'm am also using an enclosed chamber.

    Between adjusting bed temp, nozzle temp, no enclosure - partial enclosure - full enclosure, different fan speeds I can't get a good result with ABS.

    I either get unacceptably ugly overhangs and or splitting of the layers.

    I have printed a bit slower as well which made my big model about a 22 hour print.

    I haven't tried printing very very slowly though.

    I think this approach looks to be great for PLA but not for ABS.

    @ChrisR

    foehnsturm tried this previously and the results looked pretty good.

    I got an aquarium pump previously but when I switched it on the UM crashed (and it's a bit noisy) and then along came the Cross Flow fan idea so I have been trying out that.

    I am going to give the aquarium pump another go now though just have to stop the UM from tripping with a line filter or something.

    I have seen a big ABS print from the UM2 that looked pretty good. (Partially enclosed dual fans)

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @foehnsturm:

    Thanks, that helped a lot. The first prints were all with 100mm/s. This time I lowered it to 35mm/s. The result looks much better. Did not know that it would influence the quality of overhangs that much, because the straight parts looked very good. The quality of the overhangs on the rear side is slightly lower. Yours seems to be a little better. So maybe it´s only a matter of tweaking the values.

    Maker Faire Robot Front

    Maker Faire Robot Back

     

    Sorry for the bad quality pictures, had to do them with my mobile.

    My print was done at

    0.1mm layer height

    0.8 shell thickness

    1.0 mm bottom/top thickness

    20% fill density

    35mm speed

    200°C printing temperature

    50°C bed temperature

    2.85 mm diameter

    100% flow

    0.4 nozzle size

    40.0 retraction speed

    4.5 mm retraction distance

    0.3 mm initial layer thickness

    150.0 travel speed

    20 bottom layer speed

    5sec minimal layer time

    expert settings:

    0 mm minimum retraction travel

    enable combing

    0 mm minimal extrusion before retracting

    0.0 Z hop when retracting

    0.5 mm fan on at full height

    40% fan speed min

    45% fan speed max

    10 mm/s minimum speed

    Which values are different from yours?

    @fns: would be interesting to hear which values work best with ABS. Still got 2kg ABS black but did not print it on the ultimaker until now. Are you printing on heated glass with a diluted wood glue? Did you modify the ultimaker for ABS in any other way. I think the cross flow fan will help to keep the upper part of the hot end cool and therefore avoid filament jams.

    @ChrisR: foehnsturm has tried a lot of different cooling ducts and concepts before comeing up with the cross flow fan. Not sure if he also tried the aquarium type pump. If not maybe you can give it a try and report.

    I was thinking about useing a pneumatic tube, with some very fine orifices aiming at the nozzle and a precision pressure regulator, flow tube, needle valve etc. to precisely control the air flow. Got all the stuff, but before testing it, I tried this approach and like it very much. It´s much simpler and the results are very, very good. Just optimizing for the last few percent.

    Regards,

    Philip

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Just make sure that the airflow is oriented horizontal or even a little downwards to ensure that the topmost layer receives uniform cooling everywhere.

    Generally, I'd say the crossflow fan provides way more cooling air flow right in the moment of extrusion whereas many printhead-attached fans (like UM2) don't reach the extruded filament at all until the head moves some 5 to 10 mm ... and only as long as it doesn't move too far away. But it provides little to now cooling air flow for the layers underneath on the lee. Additionally there will be some "cool -> not so cool" temperature gradient in the cross-section of the freshly extruded filament if it goes perpendicular to the airflow. So, if the freshly extruded layer isn't cooled down enough issues will arise.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @Duesentrieb: my heated bed looks like a sandwitch: I'm using QU-BD's silicone heater with a 4 mm alu plate and a 3mm plain glass on top of it. ALU plate is just to mount the entire thing on Ultimaker's Z stage and to spread the heat evenly. I did not want to glue the heater to the bottom of the alu plate so I'm using two pieces of 3 mm glass stripes on the very bottom to hold them together by using a regular binder (I need two pieces of glass to hold the bottom because there's a bump in the middle of the heating pad where the wires come out). Here's how it looks:

    Ultimaker-heated-bed-sandwitch.JPG

    This silicone heater is quite fast.

    I'm using concentrated ABS juice on top of the glass plate and with brim it works (it's still not enough for thinner objects without brim although I made a very concentrated juice). I did not want to add expensive & vulnerable kapton tape on top of the glass.

    I used almost a kg of ABS with very few success. First I tried to print objects without fan and completely opened sides, than with closed sides but opened top and finally with completely closed sides & bottom (=chamber). I also tried to print by using the crossflow fan, with 40%, 50%, 100% speeds, with opened top and closed top - no luck at all.

    If the crossflow fan is on the layers simply does not bond together: the object looks absolutely fine but if you try to break it it breaks quite easily.

    Without the fan the objects are rigid enough but sometimes the edges became ugly - look at the edges of that hex polygon inside:

    Bad-ABS-edges.JPG

    I think printing with ABS definiately needs some level of cooling - even if it is not too much.

    I' thinking about designing a temprary small fan mount (for stock 40mm fan) for this printer if I would like to print ABS.

    For the crossflow topic: I asked a laser cutting company to glue an acrylic box for me which holds the crossflow fan too - here they are.

    The left side which just directs the air:

    Ultimaker-chamber-left-side.JPG

    Right side with the crossflow fan:

    Ultimaker-chamber-right-side.JPG

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    fns720 -- looks lovely!

    I've been lurking in this thread for a while because I've been hideously busy with a KickStarter project but it is of great interest to me. I'm glad to see this topic is heating up again :cool:

    Perhaps the solution to the issue of uneven cooling in the "lee" of the printhead where it blocks the airflow is a small fan on the printhead that blows directly downwards.

    So the setup would be that the fan is on the left side (the reverse of fns720's setup) and the fan is mounted to the printhead on the right side, blowing directly down, perhaps with a duct.

    My current setup has two fans on the printhead, one in the standard position and one blowing down vertically as described above, and seems to work well. You can see it here: http://www.thingiverse.com/thing:164499

    It is an accessory mount ring for the hot-end that makes it much easier to mount things to it, and it was designed with fans in mind.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @MadOverlord: this fan mount looks nice, I'll try it out. Do you print ABS with this config?

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Very interesting stuff guys!!

    A side note about using an aquarium pump:

    I've had to work with one (Innovatek HPPS, Eheim 1046 style DC pump) and I noticed that the thing is a HUGE inductive load.

    If you drive such a DC pump (not talking about 230V pumps here) directly from a laboratory power supply, then it will show something like 20V even though it's only set to 12V (inductive load...).

    The UM electronics will not like to be on the same power supply as such a DC pump!

    There is a simple solution: Use a http://en.wikipedia.org/wiki/Flyback_diode and a large http://en.wikipedia.org/wiki/Decoupling_capacitor to tame the pump.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Aww man, "Eheim 1046", that brings me back :D I still have one of those sitting in the closet together with the watercooling setup I built many many years ago (it even made the front page of [H] :p ). Those were the days :)

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @fns720: I only print in PLA.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Did some more test prints with the cross flow fan and they are absolutely awesome. Best prints I made so far.

    I printed out an ultimaker handle, with more than 7 hours and with almost 22cm a large object.

    https://www.youmagine.com/designs/ultimaker-handle

    No warping at all. The bottom is absolutely flat, the PVA could be spread more evenly, but that´s it. No problems on the lee, absolutely sharp contour, very good and smooth surface. Have a look yourself.

    Ulti handle

    Ulti handle

    Ulti handle

    Ulti handle

    Ulti handle

     

    It also works for very small parts. Printed the retraction clip. Perfect. Every detail was right.

    http://www.thingiverse.com/thing:46157

    A more complex part: Ultimaker minimum printhead.

    http://www.thingiverse.com/thing:154845

    Good quality. A little warping, turned out the heated bed at 3mm, that was too early. My fault. The detail grade is very high. No real defects, just some very small dents.

    Ulti Minimal Printhead

    Ulti Minimal Printhead

     

    Had the same idea like you Mad Overlord, but wanted to test it before posting. Yes, you have to cool the lee. A small fan is a simple solution. But I found an even simpler. I took a 8x8cm piece of 0.4mm Aluminium and folded it downwards. It´s attached to the printhead and directs airflow which is to high for cooling the part and has already passed the printhead downwards into the lee.

    Looks like this:

    Deflector

    With this setup I printed the Maker Faire Robot again, with exactly the same settings. This time the overhang on the backside looks very good. There are small, almost invisible imperfections, but that´s totally fine, I think. Have to test this setup for some more prints to make a real statement. What are good complex parts for test prints you would use?

    With the same setup I also printed out another hollow cube.

    http://www.thingiverse.com/thing:30171

    Hollow Cube

    Hollow Cube

     

    First it looked great, but failed on the bridging. Maybe I had the fan running to low for bridging. It was only at around 20%, which seems to be more than enough for most parts. Anyway I have to test this setup with bridging a little more intensive.

    The more I test this setup, the more I like it. Two positive side effects I have noticed are, the cross flow fan cools the print, but also the stepper motors. After 5 hours printing they were still very cool, only the extruder motor was hot. The upper part of the hotend is actively cooled, this prevents a clooging hotend and maybe enables higher printing temperatures for the hotend.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Interesting. Another possible approach would be a small printed duct that deflects the airflow into the lee. That might get you more airflow where you want it.

    But quite frankly a fan would be about the same weight and give you much more cooling air.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Yes a fan would also do. Maybe I will give it a try. Mhm I run the cross flow fan at very low RPM (only 20%), I think a small fan needs a percentage to deliver sufficient air... The best way would be to be able to controll the fans independently via software... I am sure there is a simpler solution. First plug both in, see what happens, test it...

    Foehnsturm, I agree. A cross flow fan offers a global cooling, whereas a normal fan only offers a local cooling. Another thing I learned with the cross flow fan is, that it even provides more than enough cooling air at very low RPM, even if it is almost stopping because of low power. So cooling is not about getting the strongest fan with the biggest air flow, but getting a solution which brings the cooling air exactly where you need it. Perhaps it is a good idea to test a setting with a cross flow fan for general cooling and some kind of construction that cools directly around the print head...

    @fns720: Thanks for shareing your heated bed solution. Mine sucks right now, have a heated bed which should be a copy of the UM2 heatbed, and heat up very fast, but with 24V and right now I only have a big 12 PSU... The next heated bed is definitely a silicone heater... I haven´t found a good source in Europe, yet.

    I think it is good, that you tell us about your problems with the cross flow fan and ABS, because so far all of us have only used PLA, which works great, but ABS seems to be a completely different story. Perhaps the cross flow fan is not suitable for ABS. Looks like you need some cooling, which cools directly at the print head.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach
    @fns720: Have you tried to increase the air resistance e.g., by placing a filter in front of the fan to reduce the flow rate? It might do the trick for ABS
  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @tsp42: I didn't, the fan blows directly to the bottom. I mounted the fan using http://www.thingiverse.com/thing:21182.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Ok, so the resistance will have to be placed on top of the fan. I'm wondering if a vacuum cleaner filter could be used.

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    @tsp42: you gave me an idea :) I'll push a piece of sponge into the fan to spread the air and will run a testprint during the night.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    Did another test print. Mad Lords Voronoi Klein Bottle. Took 17 hours. http://www.thingiverse.com/thing:145694

    Voronoi Klein Bottle

    Voronoi Klein Bottle

    Voronoi Klein Bottle

    Voronoi Klein Bottle

    Voronoi Klein Bottle

     

    The result is good, but could have been better in detail. On the lower part there is no stringing. On the upper part there is. Why? Any suggestions? This print also confirmed that the lee has an effect on the overhangs and maybe the deflector plate improves the quality a little, but is not a perfect solution. So next thing to try is a small fan on the print head for the lee.

    First picture: Fan side

    Voronoi Klein Bottle

    Second picture: Lee side

    Voronoi Klein Bottle

    It´s not that clear on the pictures as it was in reality.

    @fns720: Hope you will solve the problem with ABS. If I understood the idea of the heat chamber right, it is kept at a temperature of around 70°C to optimize the merging of layers and improve overall print quality and minimize warping and irregular shrinking. So you need heat. On the other hand you need cooling for small details. The cross flow fan cools globally. Maybe a small solution which cools diretly at the nozzle with only a small amount of air will do the job. If you use the cross flow fan only to create an air curtain to keep the heat in the build chamber and therefore place it as high as possible, so that it DOES NOT or almost not cool the freshly printed ABS. I think the optimum would be to cool the ABS from more than 200°C to around 70°C, but not less. On Friday I am on a conference where they also demonstrate industrial grade FDM printers. Will talk to them and try to find out what exactly these machines do with heat management and if and how they cool. Let´s see.

     

  • Link to post
    Share on other sites

    Posted · The crossflow fan approach

    It was not a good idea to block the direct air flow - I made some pictures to demonstrate the difference. The object lies on its back now but it was printed in a standing position (it's a phone cradle).

    Direct (unblocked) airflow - a smaller problem on the inner edge:

    abs-print-direct-airflow1.JPG

    Direct (unblocked airflow) - the same small annoyance from the other side, all the other parts are more or less acceptable:

    abs-print-direct-airflow2.JPG

    Blocked airflow - disgusting back side:

    abs-print-blocked-airflow1.JPG

    Blocked airflow - ugly edges, bad "pillowing":

    abs-print-blocked-airflow2.JPG

     

  • 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

    • Our picks

      • Introducing Universal Cura Projects in the UltiMaker Cura 5.7 beta
        Strap in for the first Cura release of 2024! This 5.7 beta release brings new material profiles as well as cloud printing for Method series printers, and introduces a powerful new way of sharing print settings using printer-agnostic project files! Also, if you want to download the cute dinosaur card holder featured below, it was specially designed for this release and can be found on Thingiverse! 
          • Like
        • 10 replies
      • S-Line Firmware 8.3.0 was released Nov. 20th on the "Latest" firmware branch.
        (Sorry, was out of office when this released)

        This update is for...
        All UltiMaker S series  
        New features
         
        Temperature status. During print preparation, the temperatures of the print cores and build plate will be shown on the display. This gives a better indication of the progress and remaining wait time. Save log files in paused state. It is now possible to save the printer's log files to USB if the currently active print job is paused. Previously, the Dump logs to USB option was only enabled if the printer was in idle state. Confirm print removal via Digital Factory. If the printer is connected to the Digital Factory, it is now possible to confirm the removal of a previous print job via the Digital Factory interface. This is useful in situations where the build plate is clear, but the operator forgot to select Confirm removal on the printer’s display. Visit this page for more information about this feature.
          • Like
        • 0 replies
    ×
    ×
    • Create New...