This has got me scratching my head



  • Really struggling to find the cause of this.

    https://drive.google.com/file/d/0B_MwtHtQR_ZvWjlZTEhIVTFnbW8/view?usp=sharing

    I came across this issue on another print so I created a simple test part to reproduce it. It's a simple square with 2 circular cut outs. It was printed with a Diamond hot end using a single tool with a mixing ratio of 0.50:0.00:0.50 so using 50% each of extruders 0 and 2 (and none of extruder 1). Blue filament is loaded into extruder 0 and white in extruder 2 (extruder 1 has yellow so we can be sure that it played no active part in the print). You can see that there are distinct colour changes. By observing the printer, I can say that these only occur after a retract - non print move - unretract sequence. It isn't random - does exactly the same thing every time. I thought it might be something to do with retraction/unretraction, - maybe some sort of pressure build up then release. But if that were the case, then I'd expect to see some odd effect which decayed as the pressure normalised so there would be a gradual transition back to "normal", but it's like something switches and stays that way until the next retract- move - unretract sequence.

    It's as if the mixing ratio changes. That is to say, one extruder runs slightly faster or slower than the other. It would only need to be 1 or 2 percent so wouldn't show up as over extrusion and it would be almost impossible to see by watching the extruder gears. I can't think of any mechanical cause that would be so consistent and reproducible. Skipping steps or such like would be much more random (and I can't see evidence of that happening while it's printing).

    All thoughts, ideas and suggestions welcome.



  • That is odd Ian isn't it? I've no experience with mixing hotends so this will be the outside perspective but what does the other side look like? That is does every layer have the same pattern? You said that mixing hotends don't completely mix rather create a toothpaste twist of filament are you seeing the effect of that in the print, seeing one side of the toothpaste stripe in one area and the other elsewhere?

    Be interesting (not that I imagine you need more complexity in your printer) to see the output from two filament sensors during the print see if both extruders are turning exactly as they should. Maybe you could mark the titan's larger gears with a few lines and print it again very slowly, see if you can count the revs on the extruders?



  • Hi DJ,

    It's difficult to tell if the first layer has the same effect. I use 3Dlac which leaves a bit of residue and I also print the first layer a bit thicker so it looks different than all the other layers in any case. Having said that, all subsequent layers have the same effect, just that the direction is at 90 degrees but it still changes colour when it moves to do the infill around the back of the holes.

    Ref the toothpaste effect - yes you can just about make it out when you look closely but for any given layer, the infill direction is exactly the same. So it's not light reflections or such like. The toothpaste effect is much more noticeable when you look at the sides of a tallish object, rather than looking down from the top. In this case it just seems that certain sections have a predominance of one colour over the other. - Very weird.

    I have already marked all the extruder gears with a "tipex" line on both sides at 90 degrees in a sort of cross shape. That's how I managed to deduce that there was an issue with recovery after retract that David has now fixed. If there is a difference in speed, it is only slight and I'll have to print a much bigger object to be able to measure it.

    More testing…......... (Could do with a few more hours in the day.)
    Ian.



  • Yeah I've a long list of things I'd love to spend time doing methodically but haven't the time. Can't shed any more light on the issue here, but I'm quite certain you'll work it out.



  • Just about to print 3 of those parts as a single gcode file. One part uses 50:50 extruder 0 and 1, another part uses 50:50 extruder 1 and 2 and the third part uses 50:50 extruders 2 and 0. That should alow me to see if it is common to all 3 or just a single extruder. Watch this space….....



  • Sorted !!. Well that is to say I now know the cause.

    Observing the printer while this is happening, what I see is as follows. Although the direction of infill is the same - for the sake of argument bottom left to top right, the print head moves steadily from left to right doing each line in turn. Then when it moves to go round the back of the holes, it is still doing infill from bottom left to top right but this time the print head is moving steadily from right to left doing each line, and that is when the colour changes.

    I think the mechanism that causes this is as follows. Imagine the filament coming out as stripey tooth paste which is colour A on one side and colour B on the other. When you do a series of parallel lines the filament may have a tendency to fold, either away from or towards the previous line, resulting in a predominance of one colour on top. Now when you do infill working from the other direction (but still at the same angle), the previous line is now on the other side of the filament so it will fold in the other direction and there will be a predominance of the other colour on top. Hence the difference.

    Well that is the theory. It explains what I am observing and we know that the filament does indeed come out like stripey toothpaste. Does that sound reasonable?

    Ian.

    P.S. Of course, there is no cure, or at least there is nothing I can do about it. It's another "feature" of the Diamond hot end



  • Oh man!
    I love you way you bite into a problem and do not give up until you get it. Well analysed!
    The only way to solve it would be if the Diamond nozzle had a better mixing strategy. But space is so limited in there that I can't see that improving.



  • @Zesty_Lykle:

    Oh man!
    I love you way you bite into a problem and do not give up until you get it. Well analysed!
    The only way to solve it would be if the Diamond nozzle had a better mixing strategy. But space is so limited in there that I can't see that improving.

    Thanks for the compliment. It's kind of in my nature to dig deep until I understand the cause of something. When I had my own consultancy business, I spent a few years commissioning test facilities on behalf of a Japanese company that supplied analytical equipment. It was no good just identifying if something wasn't working right. I had to pinpoint the problem, whether it was hardware or software (or in some cases user error), so that the really clever people could fix it. I think that kind of background helps.

    Ref the Diamond - yes I kind of wonder why I persevere with it. I guess it's because, despite it's faults, there is nothing better out there at the moment (but I'm working on it). If you know it's faults and limitations, you can often tweak the design of what you are trying to print to make allowances - which is why I needed to know the cause of this little issue.

    Also, I think I must have a masochistic streak.

    Cheers

    Ian

    P.S. I'm coming to Cyprus for a week at the end of April - what will weather be like then?



  • So I wasn't too far off the mark? Just thinking about something I use at work that might apply here. We use a lot of two part adhesives and impression materials which are delivered from two barrelled syringes into a mixing nozzle with small fins inside that forces the material to blend as it travels down the nozzle. https://www.google.co.uk/search?q=dental+mixing+nozzle&espv=2&biw=1920&bih=925&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjnpKCt0-XRAhWmAsAKHU_CCh0Q_AUIBigB Now I'm not suggesting anything so extreme for a diamond hot end, but a little bit of disruption to the flow of material as it travels down the nozzle might force a level of mixing that currently doesn't occur to take place, provided it did not resist the flow of the material any more than absolutely necessary to mix it?



  • Hi DJ.
    No you weren't far off the mark at all. I know what you mean about 2 part adhesives - I've used this sort of thing before http://www.screwfix.com/p/fischer-fis-vt-vt-vinylester-chemical-mortar-resin-380ml/14364?kpid=14364&cm_mmc=Google--Product%20Listing%20Ads--Sales%20Tracking-_-sales%20tracking%20url&gclid=CjwKEAiAn7HEBRDHwNqitoWqsQcSJAADWmI2k_jQSdbmkNsKw_VUCMi0K6jrFUu-N9N_AW8SvyiS8xoC55Lw_wcB. Same principle just on a bigger scale. IMO it is what is needed but as the Diamond hot end stands at the moment, it would need to be scaled down to 2mm long and 0.4mm diameter. Or maybe something grafted on? Either way some fairly precise machining. Ian


  • administrators

    For history of the work already done on this worth reading these links:
    http://blog.reprap.org/2012/04/colour-mixing.html
    http://www.reprap.org/wiki/File:RepRapColourMixingReport-jmc.pdf

    Rich also has a good overview:
    http://richrap.blogspot.co.uk/2012/08/3-way-quick-fit-extruder-and-colour.html

    Passive mixing does not work well with hot plastic. Active mixing works but is mechanically difficult.



  • @T3P3Tony:

    For history of the work already done on this worth reading these links:
    http://blog.reprap.org/2012/04/colour-mixing.html
    http://www.reprap.org/wiki/File:RepRapColourMixingReport-jmc.pdf

    Rich also has a good overview:
    http://richrap.blogspot.co.uk/2012/08/3-way-quick-fit-extruder-and-colour.html

    Passive mixing does not work well with hot plastic. Active mixing works but is mechanically difficult.

    Thanks for the links Tony. I'd already seen richrap's blog but not the others. I'd probably temper your conclusive comment to something like "passive mixing has not yet been found to work well" because, to the best of my knowledge, there is no fundamental reason why it shouldn't be possible.


Locked
 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.