SoliDualCool -> Floating Perimeter Bridging -> Massive Overhangs!

Update: new mount file attached.  It doesn't bump into things now!  Although if you use it, you need a party tray for your front door of the enclosure.

I have been working for a few days on a project to see just how far I can push overhang printing on my S2 Pro.  It all started when my package of 40mm fans finally came in (after 27 days...) from Hong Kong. 

As I mentioned in another thread, I was looking for a way to create an effective fan mount for my J-Head MK-V and Lawsy's MK IV.  Having tried a ring duct nozzle (and one before when I used the original Solidoodle hot-end), I was not satisfied with the relatively weak current of air emanating from it.  DePartedPrinter suggested I drop the ring duct completely, and to go with a direct duct instead.  I decided on a somewhat different course: to just eliminate the ducts altogether.  This is somewhat similar to g00bd0g's fan mounts on Thingiverse.  What I did is design a mount that works for two simultaneous fans focused on the nozzle and the print.

The resulting hack is here:

Note: I have included an "FAQ" section at the bottom of the post to explain my decisions and address some of the possible questions people have.  Please be sure to read it!

Now, here are the results of the mod!  In the below video, I demonstrate the printing of a 2cm x 2cm square region that is essentially floating, except for four support pillars.

You can see a bit of failure at the beginning of the print, but more than 80% of the (completely unsupported) overhang region is flawless.  I call it "Floating Perimeter Bridging" since it relies on the adhesion of bridge material to floating perimeters.

To prove the quality of my approach, here are some underside shots of three replications:

You can see that there is very little sag.  It is starting to approach the quality of an ordinary print surface.

Now, the PhD student in me insists on proper scientific methodology, so here is the comparison of the same print without any cooling:

So it seems clear that it has a substantial effect.  Of course, it may be possible to achieve the same overhangs without my ridiculous dual fans.  So I encourage others to try it with an ordinary fan duct.  Maybe it works great, who knows.  I just want to help push things forward!

Conclusion / Recommendations
I think that we can be close to achieving some big improvements in minimal-support print techniques. 

What I can suggest from the experiment is this:  with proper cooling, any polygonal surface in which all, and only the vertices are supported, should be printable.  (given reasonable distances) 

I have attached the STL file for the test print, and the STL for the fan duct mount.  The test print is a slice off the top of the BallCageTwist from thingiverse (http://www.thingiverse.com/thing:42074). 

For printing the fan mount, flip it by 180 degrees so that it prints nicely.  Be sure that you use cooling or dummy prints to let the fan mount section cool off properly as it prints.  To mount it, you can glue 1/4" magnets into the mount holes (two in each hole).  It might tend to flip up a bit when mounted to the Lawsy MK IV mount.  I just glued a little beam on top of the body of the MK IV that pushes down on the fan mount to hold it down.  You can see it in the first picture of this post.

Future work includes:
-Bowden-Doodle for better fan placement
-Characterization of long rectangular objects to assess warp
-Other polygonal surfaces such as triangles or pentagons.

*****FAQ Section*****

There are a number of questions that probably come to your mind, and for good reason.  These might be:
-Why do you need two fans?
-Doesn't that cool the bed down?
-Doesn't that cool the extruder down?
-Won't that cool the part unevenly?
-Won't you get brittle parts from too fast of cooling?
-Won't that abomination of a fan mount hit the Solidoodle frame?
-Why don't you just print the part upside-down?

So, allow me to address these.  First, though, I will explain my fundamental reason for going this route.  Without going into too much detail, the approach is to forcibly neutralize any thermal gradients on the part as it is printed (with the exception of those produced by heat conduction through the bed).  By using two fans, the amount of airflow is staggering.  Most of the part is close to room temperature at all times.  Thus, there is only one temperature gradient, and it is in the vertical direction. 

Now to go through the concerns:

-Why do you need two fans?
Wider coverage of the surface area, and more intense cooling.

-Doesn't that cool the bed down?  Will this make the part pop off?
When you use an enclosure, the case reaches something of an equilibrium where the temperature of the bed is not affected very much by the fans.  It actually has the effect of increasing the ambient temperature due to pushing heat off the bed more rapidly.

-Doesn't that cool the extruder down?
Yes, but who cares.  The resistor puts out so much heat that it is easily able to maintain the setpoint perfectly.  Actually, the fans help since it kills the overshoot by a lot.

-Won't the placement of the fans cool the part unevenly?
For sure, it could be a lot better.  However, there is so much airflow that the part tends to just cool instantly anyway.  I am contemplating going to a custom Bowden setup so that I can move the fans to a more symmetric orientation.

-Won't you get brittle parts from too fast of cooling?
No, because if your plastic is hot enough, it will still have time to melt the adjacent layer and form a good bond.  I have noticed no difference in strength so far.

-Why don't you just print the part upside-down?
Because I like to make myself suffer unnecessarily.  No, but really, I'm just trying to address the problem in a pure fashion.

Implementation details: (updated 5-8-13)

Mount: Lawsy's MK IV
Nozzle: J-Head MK-V-BV @ 228 C (not solidoodle hotend temperature) (also, updated)
Case: Fully enclosed with party tray in front side
Surface: Glass w/ Aquanet Unscented

Important Slic3r settings:
-5 perimeters
-40 mm/s perimeter speed
-50 mm/s bridging speed
-180 mm/s travel
-2000 mm/s^2 bridge accel
-3000 mm/s^2 perimeter accel
-2000 mm/s^2 infill accel
-0.8 bridge flow ratio
-0.2mm layer height
-0 mm lift z
-Turned off "cooling" in slic3r (my fan runs all the time)
-Used 2mm of brim for this part, but not necessary for larger objects.