Page 1 of 3

The Flex3Drive

Posted: October 7th, 2016, 3:24 pm
Hi guys,

I noticed there's quite some interest in this topic, so I decided to already start a thread where you can ask any questions, and I will post updates as I have the time.

From this thread:
Izzy wrote: Any chance of more photos of the flex3drive from all angles?
May have to invest in one to test it out and for flexable filaments, but may have to get a 2nd printer and fit it to that.
I can definitely take some pictures, no problem. About flexible filaments: I haven't tested it yet, but it's a good reminder because I do have a sample of very flexible ninjaflex and something that's a little bit stiffer laying around. Any suggestions as to what I should print, or what settings to use (temp, speed, etc)?

Re: The Flex3Drive

Posted: October 7th, 2016, 3:51 pm
by Izzy
Test piece wise I usually start with the UMrobot the treefrog & Benchy3D once I have the temperature dialled in.
Those pieces allow for testing of retractions, bridging, overhangs etc.
On flexable filaments I usually do 200um layers, at about 25-30mm/s.

Re: The Flex3Drive

Posted: October 8th, 2016, 4:55 pm
I took some pictures:
The motor is mounted on the back, over the hole where the old feeder used to be. The driveshaft of the motor is D shaped, and it's connected to the flexible shaft by the little screw you see. There are two of them, opposite each other. Note that this is not the original feeder motor, which is 400 steps/rev. This one is the same as the xy-motors, with 200 steps/rev. The motor is not included in the kit, because shipping it is more expensive than the motor itself.
Here it is from a different angle. I'm probably going to reroute the cables through the hole behind the bracket at some point, but I can't be arsed to take it apart at the moment :p Would look a little bit nicer though.
Here's how the ptfe tubes are connected to the print head. The filament tube is inserted all the way into the head, till just above the hobbed wheel. Note that the screws are different also, the originals aren't big enough.
The printhead is slightly bigger than the original since it has to have space for the gears to fit. They are located behind the 'ex'.
On the left there is a lever. It is used to tighten the bearing that pushes the filament against the hobbed wheel. You need to loosen it when changing filaments, and then tighten it after the new one is inserted. I tighten it by hand as much as possible, then give it about another turn or so with an allen key (2.5mm) until the plastic flows properly.
This is what the hobbed wheel looks like when you take out the lever. You need to pull out the bolt on the front left to do so. I took this pic when I had a nozzle clog because I was printing too cold. A little piece of filament got stuck at the top of the hobbed wheel when I removed the filament, but I could easily pull it out this way.
You can see I no longer use the PTFE coupler. This one came with the Flex3Drive. It apparently is made of graphite filled teflon. I originally was going to use the I2K insulator from 3dSolex, but I didn't need to anymore now :)
This is a picture of the coupler before I installed it. Unfortunately it's a bit fuzzy.
You can see the worm drive through this hole on the right side.
I weighed the print head before installing it. (I hadn't installed the new coupler yet when I took this picture). Unfortunately I didn't weigh it with the old setup. Still, 213 g is not very heavy :) The heater and thermistor are not installed, but that won't make much of a difference.

The entire printhead came assembled, but of course I took it apart out of curiosity before installing it. Regrettably I forgot to take pictures as I was eager to set it up. Of course I had to add the linear bearings, and the hot end etc myself.

If anybody wants more detailed pics of anything let me know.

Re: The Flex3Drive

Posted: October 22nd, 2016, 2:04 pm
Hey guys,

I had some time today so I wanted to show you some tests I did on my printer. Of course one of the most interesting things to test is the extrusion: how fast can you go? I printed a bunch of test cylinders, both with the UM2 with Robert's feeder modification, and with the Flex3Drive.

I got the tests from the Ultimaker forums: 3-10 mm³/s and 8-15 mm³/s. With the old setup, I only printed the 3-10 mm³/s test, because it would fail before reaching the end anyways.

MakerPoint Ultramarine PLA, printed at 230 °C.

Both were printed with the original feeder. Unfortunately I ran out of this filament, so I don't have any prints with the Flex3Drive. I printed this same test twice with exactly the same settings, because I wanted to see how repeatable the results were. As you can see, the results aren't bad at all. Up to 9 mm³/s the print came out very well, only at 10 mm³/s it started to fail. The second print is almost exactly identical to the first, so I'm quite confident that this is a reliable test.
MakerPoint Traffic Black PLA.
The left one was printed at 230 °C using the original feeder, the middle was printed at 210 °C and the right at 220°C using the Flex3Drive. I did have some trouble printing the large test. My first few attempts, I encountered a nozzle clog (see pic in previous post) and the print failed at 9 mm³/s. After I cleaned out the feeder, I decided to print it a little hotter, and it went without issues.
MakerPoint Signal White PLA.
Same story: left one printed with the original feeder at 230 °C, middle and right printed at 210 °C and 220 °C respectively using the Flex3Drive. The last one still shows some signs of under extrusion here and there, even though the last sections came out perfectly. Some time after I finished these prints I learned that the tension on the lever was probably too low.
MakerPoint Pearl Gold PLA.
With the original at 230 °C the print failed at 9 mm³/s. This filament displays under extrusion differently than the other ones, it seems to stick together more than the rest. With the F3D at 210 °C the test came out perfectly in one try.
MakerPoint Iron Grey PLA.
The upper three were printed at 210 °C, 220 °C and 230 °C respectively, using the original feeder. I printed it first at 230 °C, and was not satisfied with the way it turned out, the temperature was obviously too high for this filament. The surface finish was better at lower temperatures, but the test did not finish without under extrusion.

The bottom three were printed at 210 °C using the Flex3Drive. The leftmost cylinder was the first print I attempted after installing, and I was a bit dissappointed at first at how it turned out. It did look a bit weird to me though, the letters seemed a bit "fuzzy" compared to the other cylinders I printed. A while later I discovered I used a 0.6mm nozzle instead of a 0.4mm as intended.... :roll: The second print was done with the right nozzle, and it turned out fine. The surface quality isn't top notch, but there was definitely no under extrusion. The big cylinder came out well, probably could've been a bit better if I had tightened the lever a bit more.
MakerPoint White Aluminium PLA.
Printed with the F3D at 210 °C. This filament prints amazingly well. Definitely the best looking cylinder I've made so far, slightly better than the gold one even.

Comments on the tests
The small test runs very smoothly. It is very well made. On my old setup I used TinkerGnome, which allows you to monitor the flow as it is printing. I checked this test, and the output is exactly as indicated.

The big one has a few issues. The first section runs just fine, but from 9 mm³/s up the movement becomes very 'blocky' (can't think of the right word in English). You can actually see this if you look close, the surface is not smooth, it feels more like a 100 sided polygon or so. Also at most layers, some part of the circle prints faster than others, even if there's no text being printed. This is probably due to poor slicing, because every now and then there is a layer that does print perfectly smooth, so it definitely is possible.

Apparently there also exists a version that goes up to 20 mm³/s, but I haven't been able to find it. I've heard it was used at the Ultimaker HQ for testing purposes... Maybe @nallath could share it :-D ?

Practical applications
Of course printing cylinders is nice, but how does it work when actually printing something useful at high speeds? For that purpose I made a test object. I chose for a gear, because it has some interesting outside geometry that might print differently at different speeds. I made it so that half of it would print with and the other without top layers, so I could inspect the infill.
I printed them at 210 °C, at .25mm layer height and a 0.4mm nozzle. The speeds used were 50 and 100 mm/s resp. for the top two, and 150 mm/s for the bottom three. This corresponds to 5, 10 and 15 mm³/s output of plastic.

As you can see, the results did not turn out quite perfect first try. The one I printed at 50 mm/s came out pretty much perfect. Infill is solid, top layer is nice and smooth, and not under extruded. Only thing I might change is the infill angle, to close the little gap in the top layer.

The print at 100 mm/s came out quite okay too for the first part. The infill printed great, and feels solid. The top layer had serious under extrusion issues though. My first attempt at printing at 150 mm/s (bottom left) didn't turn out so good. The bottom layers were a bit 'see through': there are small gaps between the lines. The infill was mostly okay, but still there was some under extrusion here and there, and it did not feel solid like the other two prints.

Of course this was a little bit disappointing, so I contacted Mutley3D for some advice. After some help I found the problems: I had not properly tightened the lever, which at low speeds is not such a problem, but a higher speeds it is. Also I made an error in the slicer, and I had set the temperature to 190 °C for the top solid layers... oops. So it wasn't very surprising that they failed at all.

The fourth and fifth prints I had made sure the lever was tightened properly, and I made sure the temperature was set correctly. The fourth one was still slightly under extruded for the solid layers, the infill came out perfectly though. The last print I increased the flow rate, and it came out a lot better. The top layer is not perfect, I used too few solid layers and the flow rate needs to be just a little bit higher. Other than that, the print turned out fine.

Concerning the outlines, they did look a little bit better on the slower prints.

Since then I have printed some actual prints at 100 mm/s, 0.2mm layer height so 8 mm³/s flow. For the outlines I reduce the speed to 40 mm/s or so. I use an extrusion multiplier of 1.10, and the prints turn out just fine. Especially for prints with large flat surfaces this can seriously reduce print time.

The Flex3Drive makes it possible to print much faster compared to the original feeder. However, this doesn't mean that you can just turn up the speed and expect miracles. You need to adjust your settings, and use the correct tension on the lever, both of which take some experience to get right.

Next time: prints with a lot of retractions ;)

Re: The Flex3Drive

Posted: October 22nd, 2016, 2:45 pm
by Neotko
Thanks for sharing all the info men!

Re: The Flex3Drive

Posted: October 22nd, 2016, 2:59 pm
Of course! The best way to learn is to share experiences if you ask me. You take some and you give some.

What I forgot to add above by the way:
Printing PLA at 230 °C is not a very representative test if you ask me. That is not a temperature you would use when actually printing something. Overhangs and bridges would probably suffer greatly, and small parts would be very difficult. I think 210 °C is much more realistic.

Re: The Flex3Drive

Posted: October 22nd, 2016, 3:13 pm
by Izzy
It's worth double checking the actual print speed / volume by checking the time.
I used the same origional volume test piece and calculating the layer thickness, nozzle diameter and the time taken for each revolution (timed over 10 levels) the actual extrusion rate is a little under the stated rate, (checking a square 50x50mm, 50mm/s wasn't correct) but having said that it is still a good indication of the approximate rate, and a like for like comparison across filaments some like ColorFabb, Faberdashery, RepRapper have good PLAs that will get to 8,9 or 10 depending on colour at 210-215'C, when I tried's PLA 3-4 was all it could manage, although their HT PLA was 8-9.
I found that some filaments could print at higher rates but lacked good layer adhesion so a good actual top volume with good layer adhesion was a bit lower. Although I wouldn't push PLAs much above 220 as some tended to be brittle on thin walled pieces.

Re: The Flex3Drive

Posted: October 22nd, 2016, 3:26 pm
I've never tried timing the test, and I'm not sure how useful it would be since it prints the text quite a bit slower than the rest of the cylinder. I could try printing a simple cylinder without text and see how it turns out. Might be interesting...

A square is a little bit trickier than a circle, since the print speed will be limited by the acceleration and jerk settings of your printer. That's why printing at higher speed will make the most difference when you print large objects. On smaller details, your printer can only go so fast (and they will probably turn out better at lower speeds anyways).

I haven't tried many other filaments yet. I do have a roll of PET, might be interesting to try that one.

Re: The Flex3Drive

Posted: October 22nd, 2016, 3:57 pm
by ivan.akapulko
Looks like during increasing the speed the printing quality becomes worse (as if shifting axis). I mean that at low speeds the quality of the figures is much better than the high, and adhesion between layers is slightly lower. Or it could be effect of camera shake of mobile phone.

Re: The Flex3Drive

Posted: October 22nd, 2016, 5:02 pm
by Izzy
As you can see from the photo I did a fair few of cylinder flow test and temperature test on various PLA and ABSs :lol:

Re: The Flex3Drive

Posted: October 22nd, 2016, 6:23 pm
Ivan, I don't really understand you. What image are you referring to, and what do you mean by shifting axis?

Re: The Flex3Drive

Posted: October 23rd, 2016, 5:14 am
by ivan.akapulko
MTVDNA wrote:Ivan, I don't really understand you. What image are you referring to, and what do you mean by shifting axis?
I myself have never tried this test. Today in the evening will start, and write about the results. In the photo I circled the points that I wrote in a previous post.

Re: The Flex3Drive

Posted: October 23rd, 2016, 9:43 am
Ah yes, you are right about that. On the small cylinders the text is much crisper at lower speed. On the big cylinder the text seems to be printed at the same speed every time, so the quality doesn't really change.

Re: The Flex3Drive

Posted: October 23rd, 2016, 9:49 am
by gr5
That's normal overextrusion and expected on this print. This print was designed specifically for the older UM2 with the original black feeders that would skip a lot and where it is visible when it skips as a perforated ring when you are printing too fast for the feeder. The UMO and UM2+ and UM3 don't skip so you must weigh the cylinders if testing those printers - it's really not the best test for those as there are much better tests.

Anyway about the lettering - Marlin slows down for corners - those letters have right angles and if your jerk is at the standard "20mm/sec" then it will slow down to 14mm/sec for each corner and doesn't have enough time/distance to get up to speed until all the letters are done. Because you slow down so much and because the bowden and nozzle combine to store so much energy/pressure you get heavy overextrusion on the upper parts of the cylinder where it is printing fast.

Re: The Flex3Drive

Posted: October 23rd, 2016, 10:50 am
Could you point me to those better tests? I'm very curious to try them!