foehnsturm

@ JohnnyBischof, Very good advice! I measured the other end: approx 3.3 mm ID . Apologies for my hasty conclusion Sander! So I will mount the original tube to my new printhead and see if retraction improves.

As promised. Disclaimer: Results cannot be generalized because printed with a 4 mm inner diameter bowden tube. I will repeat the test with the original tube mounted.

Hi Sander, Thank you for your instant reply. As I never managed to find a source for imperial sized PFA tubes here in Europe I'm quite sure it's the one supplied by Ultimaker. Don't get me wrong, I've been very happy with overall print quality. But never got rid of considerable stringing with PLA/PHA so started to investigate. My understanding is that 4.0 ID mm gives about 1.8 mm bowden lag ((4-2.85) * Pi /2 ) vs 1/8" gives about 0.5 mm. This defines the timespan until the filament actually starts to retract at the print head and the nozzle happily oozes.

Until today I assumed the inner diameter of the bowden tube delivered by Ultimaker to be 0.125". After some issues with optimum retraction settings I took a close look. Mine is 1/4" (6.3mm) OD but 4.0 mm ID (I know this size is produced as well). As I'm not fighting retraction length (easy to increase) but retraction lag (small bumps which are produced when the nozzle sits still and waits for retraction to start) this might be the reason. Did Ultimaker occasionally sell 1/4" - 4.0 mm tubes?

That print looks clean! Would it be possible to share the stl of the green part. I can swap UBIS and Merlin within minutes now and would like to give both a try.

Really not aiming that high as discussed here. But perhaps this https://www.youmagine.com/designs/ulitimaker-orig-modular-printhead-merlin-ubis could serve as an inspiration. I just swapped my Merlin and UBIS within two minutes.

I added the UBIS mounting parts for the modular printhead.

@Nick I hat noticeable stringing, especially with PLA/PHA. Could you give me an example of your settings (material, temp)? I want to try again when I've got my modular printhead ready for the UBIS.

There is a new, https://www.youmagine.com/designs/ulitimaker-orig-modular-printhead-merlin-ubis online now at YouMagine. It comprises a main part which holds the linear bearings and two interchangeable parts for mounting various hotends (Merlin realized so far, UBIS planned).

It's more or less a matter of acceleration like discussed http://umforum.ultimaker.com/index.php?/topic/4235-ghosting-comparison-with-um-original-and-um2/?p=35094

Speed was set to 50 mm/s. But think it's more related to jerk (20) and/or acceleration settings (9000) and belt tension. Nozzle was 0.5 mm.

Update: As my original version uses a different bowden fitting I uploaded an alternative version which incorporates Nick's profile for mounting the original bowden fitting with horseshoe clip. But this one is not tested yet.

Of course, I currently printing a second iteration which will have interchangeable adapters for Merlin and UBIS. The first iteration is on YouMagine now https://www.youmagine.com/designs/ultimaker-orig-printhead-for-merlin

Well, as it's the classic brass - PEEK - PTFE setup I'd stay below 250°C. For sure it features a comparatively small hot zone, so max extrusion rate will be limited. I will run some tests to find out. On the Merlin website they show prints with ABS, PLA, HIPS and PET. I think it's fine for layer heights below 0.2 mm.

... inspired by the UBIS thread. I've been happily printing with the UBIS for two weeks now. However one thing bothered me. With PLA/PHA, stringing was even a little worse than the Ulitmaker hotend (due to a longer hot zone I suppose). So I made an adapter and tried the new http://merlin-hotend.de/. I'm stunned. Stringing and underextrusion after retractions are gone with PLA/PHA. Overall results are as good as with the UBIS. The Merlin is tiny and uses minuscule airbrush nozzles (comes with 0.5, 0.3 and 0.2 mm). So there is very little additional heat input to the part. As it performed so well, I made a new printhead (still under development, by far not as slick as Nick's). A short video (air is coming from the right).

First prints, temperature is still a little too high. The standard 100 k thermistor table works fine (tested with water at 100°C). The hot zone might be a little longer compared to the UM original hotend, so I'm printing approx. 10 °C lower. The second part was of course printed without support. Just wanted to evaluate the small structure (3 mm diameter) overhang issue. Layers: 0.1mm, printing speed: 50 mm/s, temperature: 200 - 205 °C, Ultimaker red PLA

Something weird: Withworth BSW 1/4"-20 BSW (W 1/4"), 6.35 mm outer diameter, pitch 1.270 mm Finished the hotend swap today and started printing. First impressions: - heats up in no time - no noticeable temperature fluctuation - first prints look very clean.

I have male and female Molex connectors and the crimp parts lying around (ordered 20 each because of the shipping costs).

However, there is one difference worth to mention as far as i know: Bowden tube and fittings are imperial. Metric dimensions will only give you 3 mm or 4 mm inner diameter. The latter is quite loose, the first is too tight. Ultimaker uses an imperial size tube with a 3.somerhing inner diameter.

If I'm seeing right: yellow violet gold gold = 4.7 NOT 4.7K

Besides extra torque I would like to know if there is any difference in accuracy and ringing . Theoretically, there should be an advantage, as with the standard setup the part of the belt which pulls the head in one direction is three times as long as the part that pulls in the opposite direction.

Nope ;-) the four rods are rotating as well, while the other two don't. Linear ball bearings are designed for linear movements only.

I don't think that's the point here. The crossflow fan provides enough cooling to achieve the same printing speed more or less. If others e.g. can reliable print lots of long meshmixer support pillars with let's say 3 mm diameter it's indeed less important. As far as I know that's not the case. In fact, a discussion related to this issue drove my attention to it.

The print speed even of big prints is quite often limited by the print quality of small islands (raised edges issue) as discussed here. Take the knocking over of meshmixer-generated supports as a typical example. It's just about partially remelting tiny structures when printing 1st and 2nd perimeter and eventually infill within a second or less. Given the maximum available cooling, you have to stay below a certain print speed to get things right. But the rest of the part could be printed considerably faster. Would it be possible to add a "minimum perimeter time" which causes the print to slow down, if a perimeter line would be finished (and the next line laid down aside) below that time?

I only removed 3 support structures that KISSlicer added (lower right arm, left boob and neck). I'm way too lazy and impatient for considerable post-processing ... I used "well-seasoned" white PLA from German RepRap at 205°C, with PLA/PHA or faberdashery the surface might even become a little smoother. Print speed was 40 mm/s.