Monitor - Measure - Improve 3D Flexible Filament Printing Through a Bowden Cable

Recently I decided to delve deeper into my Flexible Filament extrusion system in order to gather some empirical data to try to understand what is actually going on inside that PTFE tubing system - I had two main issues which required answers, these were.

Firstly can I further improve the overall system in regards to print time and quality - an issue which we all probably seek.

The other was to try to understand a problem that I was continually experiencing- Why do some prints fail at lower speeds while others continue onto completion with no problem even when processed and printed at much higher levels.

This Instructable has been broken down into the following parts -

Step 1 - How to set up and gather data (Monitor)

Step 2 - 5 - What my data showed me (Measure)

Step 6 - What I did with my data to produce better prints at faster speeds. (Improve)

Measuring the most critical component of FDM 3D printing system - That is the amount of force that is applied to the printing filament in order for material to be extruded from your hot end. If you can obtain good data you can then start to understand how much force is produced, how it is managed and what happens to it within your system - Data is King!

I found a novel way to monitor this force through the use of a simple load cell (pic 2) an amplifier and an arduino it was based upon these great articles (among others) airtripper and thingiverse 2429390.

In reference to these ideas I placed the load cell at the point where the filament exited the extruder housing and where it enters the PTFE tube, the main pic shows the position of load cell on top of extruder body. Once you work out how and where to attach the load cell to your particular configuration you then calibrate your set up - I found a great set up and calibration video here.

As the filament is placed under load, the increase in pressure moves the PTFE tube slightly away from the extruder body thereby producing an electrical signal through the load cell -- For every action there is an equal and opposite reaction --.

I have built two systems, one a dedicated platform which strives to keep all the variables constant in order to test individual components like hot end / PTFE tubing etc. (pic 3) and another load cell is directly attached to a printer which provides real time data during print procedures.

To record data I used a cool little program called 'coolterm' which allows you to collate the data in excel and then produce graphs.

The first type of analysis I conducted was to work out the maximum output or failure point of the extrusion system in other words at what point do things go pear shaped and break, better known as print failures.

In order to establish this point I ran the simple gcode script as follows (you may adjust values as required to suit your individual system).

<p>G92 E0
G1 E10 F100
G92 E0
G1 E20 F200
G92 E0
G1 E40 F400
G92 E0
G1 E60 F600
G92 E0
G1 E80 F800
G92 E0
G1 E100 F1000</p>

As this script is intend to jam your extruder you will then need to take apart your system, rectify and reset.

For good scientific rigor run the test script a total of three times for each different type of flexible filament you use. This process may be a little time consuming however it will definitely save you time long term.

The results can be plotted in a graph as above. My graph demonstrates that my upper limit for ninja flex semi flex (92 A shore hardness) is on average between 1700 and 1900 grams and for ninja flex (86A) was about 15% lower.

I set a safe working load limit on this machine for this filament of 1750 grams.

Now that you have set your upper limits you can start to set speed parameters.

It is time to start exploring what is going on inside your Bowden Tube and the problem of hysteresis and in particular 'runaway' hysteresis so you may gain a better understanding as to which speed you should slice with.

Explanation of my results: You will notice in the above graph that required force initially builds quickly and then continues to increases at a reducing rate as it moves towards (hopefully) equilibrium.

At the two lower speeds, force was kept in check, at the second highest speed force started to run away but the experiment ended before failure occurred. At the highest speed the failure point was quickly reached resulting in extruder jam and. print failure.

What I surmise is happening here is - As the flexible filament bunches up more force is required to overcome the extra friction generated, this additional force results in even greater hysteresis being created which then increases friction, this then compounds around in circles leading to 'runaway hysteresis' which quickly exceeds the failure threshold.

What can be concluded is that when printing small cross sectional areas you can run at higher speeds and when printing larger cross sectional areas care must be taken to not allow build up of friction and pressure.

Now to put those initial observations into real world practice.

I printed this bumper bar for a remote control car (the file can be found on thingiverse here) sliced and printed @ 50 mm / sec. This is my go to model whenever I do a test, it is a very typical of what I am asked to produce commercially, sort of anyway, well in the way that there are multiple small sections and lots of retraction a little overhang etc, but you can use your favorite calibrate thing to test.

What the graph shows is that the build up of force is kept in check due to multiple retractions and printing small cross sectional areas, the required force does not exceed 680 grams.

What this graph also shows is that the amount of force is not lineal it can rise and fall along peaks and troughs as it traverses through different print phases.

This part was sliced with the same parameters as the bumper bar and printed at 50 mm / sec, it is a basic flat plane square 60 mm x 60 mm x 4 layers.

You can distinctively see the 4 layers - The first base layer was printed slow @ 30% (the first flat line) the second and third printed at full speed and the forth printed after a retraction (obviously the prominent dip).

What you can gain from the data is that 'runaway' hysteresis would never happen at the very low speed of the first layer but neither would any thing be printed in a sensible and viable time frame.

The second two layers printed at full speed really pushed the limits, peaking at 1380 grams.The dip and the end of the second layer is when the perimeters were being printed (which are at 50%). This allowed the force to recovery somewhat. The retraction at the end of layer 3 also did a lot to alleviate pressure.

It can be assumed if another one or two large sections would need to have been printed then the failure point may have be reached. If however smaller sections were required next then a recovery in pressure would have most likely occurred.

Things to consider when setting print speeds

1. Size of cross sectional parts x how many layers.
2. Is there smaller cross sectional parts to relive pressure in between large ones
3. Set retraction, also a good idea to incorporate retraction on layer change (helps to settle things a little).
4. Consider infill pattern which are to cover large cross sectional areas that contain changes in direction, for instance lately I have been experimenting with more honeycomb and even Hilbert curve type patterns. While these infill patterns take longer to print it allows for a constant speed through the process.
5. Consider slowing down perimeters to relive a little bit of stress.
6. Can you monitor your print and manually adjust speed during printing and when required.

I have a load cell permanently set up on one of my machines which allows me to continually monitor the printing process thus enabling me to obtain the most from the unit, what I hope to do in the future is feed the output data from the load cell to the printer control board so as to automate the whole system but this may be a subject for another Instructable at a later date.

For more in on the Flex Wheel extruder check it out here.