Reversing extruder feed and direction

In an attempt to reduce the stringing of filament at the end of a line, I have been experimenting with reversing the extruder drive. It is not easy to get a tidy end to the line and some initial experiments suggest that there might be benefit in back-tracking at the end of movement, so for a line from (0,0,0) to (x,0,0) the extruder runs normally to (x-a,0,0) when the extruder is switched to reverse. The optimum value of 'a' was a 5mm in my experiment. Movement continues first to (x,0,0) and then back to (x-a,0,0.25) when the extruder drive is switched off. Increasing z in the final move avoided the nozzle disturbing the existing filament but combined the two sections of filament to produce a tidy end without a blob. The speed was constant in my experiment. There was still a small string of ooze when moving away but much better than I normally get. I am using a BitsFromBytes lasercut RepRap and HDPE.

The gcode for the tests was produced by a small program and before putting in a lot of time to try it on a real product I would like to hear from anyone who has tried anything similar.

Jon

So are you saying that the length from x-a to x is initially thinner than it should be because the extruder is reversing and then going back over it adds just enough to make it right?

An interesting technique. Any chance of a video or a picture?

---

[www.hydraraptor.blogspot.com]

The tracks in the image below were printed from left to right. The extruder went into reverse at 13mm on the ruler and the ends are 3mm, 4mm ... 9mm long.



With a 3mm end there is clearly still lots of ooze. In this run, 6mm or 7mm looks optimal, after which the end is getting rather thin.

Jon

I think you are on to something there. When the extruder is stopped the flow ebbs with a negative exponential because the longer it oozes, the less the pressure. If however you reverse the extruder the flow rate will fall to zero more linearly assuming the pressure drop due to the feed moving backwards dominates the drop due to ooze.

That being the case if it takes 2a for the the flow to drop to zero linearly then folding it back on itself for a distance 1a will give exactly the right amount of plastic all the way along.

When the extruder starts again there will be a deficit so perhaps a double back there would help as well.

I have wondered before if we had a drive that could feed much faster whether we could stop then flow almost instantaneously by reversing very quickly and then start cleanly be fast forwarding the same amount before resuming the normal slow feed rate.

Perhaps we could use a solenoid to lift the motor assembly just the right amount.

BTW, what are you using as the base material?

---

[www.hydraraptor.blogspot.com]

I have also tried a startup pattern reversing to the start point of the line as you suggest. The initial attempts look good but I have not yet had time to get any experimental results.

The base material in the photograph is a kitchen chopping board. I would normally add a raft but that would have made it more difficult to see the test results.

Jon

Hey Jon,

This is a great technique. Seems like it should work for both starting and stopping. You should be able to get the ends the same thickness by slowing down the head a bit. This would also make it easier to start/stop.

-Geert

I have found that a startup sequence is essential after a reverse drive on the previous extrusion as it takes longer to get the material flowing again compared with just switching off the extruder motor.

Getting the parameters correct for a mixture of path lengths is proving difficult. A fixed time for starting and stopping can be determined by experiment provided the extrusion runs for over 30mm, but for short lines something more subtle is required, and more experiments are needed. Does anyone have a way of measuring the flow from the extruder second by second?

I have an idea, but not sure if you like it: Fixed webcam & recording software... If you know the distance from the surface to the extruder, you can calculate the distance in pixels. You know the time (x frames per second), and with this you can calculate flow speed Crappy solution, but works

Edited 1 time(s). Last edit at 02/11/2009 04:31PM by Mr. Seeker.