Notifications
Clear all

Comparing MMU2 and Palette+  

Stránka 1 / 6
  RSS
Peter L
(@peter-l)
Honorable Member
Comparing MMU2 and Palette+

With the (I hope) imminent shipment of my MMU2, I've started thinking about benchmarking the MMU2 and the Palette+. I've picked Gustav the Turtle ( https://www.thingiverse.com/thing:2316918 ) as a benchmark model, and my early comparisons are surprising.

Printed with no color changes and 20% infill at 0.15mm, Slic3r estimates Gustav will take 8.32m of filament and 3:27 to print on my printer (my print settings aren't quite stock so your results may be different).

A four-color Gustav with the same print settings will have a little over 400 color changes. (Normally I'd print a multimaterial model with thicker layers to minimize the changes and purge, but for this I want to use something closer to a worst-case scenario.)

Printing Gustav using the Palette+ takes 9:09, and requires 54.94(!) meters of filament. That's about 46.6m of purge for the 8.32m of actual model.

Printing Gustav using the MMU2 will take 14:37, and require 9.79m of filament. These are the estimates provided by Slic3r. That's 1.47m of purge.

It's worth noting that I am not using purge into infill because the Palette+ does not support that with gcode generated by Slic3r.

  • The difference in the amount of material is astonishing. I can vouch for the fact that the purge block generated by the Palette+ is probably 6-7x the weight of the finished model. I was not expecting the MMU2 to be that much more material efficient; in fact, it makes me wonder if the estimate in Slic3r is really accurate, or if it's somehow ignoring/undercounting the purge tower. Once I have my MMU2 in hand, I want to check the actual purge tower weight.

  • The difference in the print times is also surprising. The Palette's print time is the actual print time, and the MMU2's is estimated by Slic3r, but I assume it's fairly close to the mark. Material changes on the MMU2 are apparently very slow.

  • If the material estimate for the MMU2 is accurate, purge into infill will make a lot less difference than I thought it would. There just is not all that much purge, at least not relative to what the Palette+ requires.

  • It's also worth noting that printing Gustav on the Palette+ required two attempts because a splice broke the first time. This is definitely a weakness to the Palette's approach, and a broken splice is unrecoverable.
  • Napsal : 02/09/2018 2:47 am
    Peter L
    (@peter-l)
    Honorable Member
    Topic starter answered:
    Re: Comparing MMU2 and Palette+


  • The difference in the amount of material is astonishing. I can vouch for the fact that the purge block generated by the Palette+ is probably 6-7x the weight of the finished model. I was not expecting the MMU2 to be that much more material efficient; in fact, it makes me wonder if the estimate in Slic3r is really accurate, or if it's somehow ignoring/undercounting the purge tower. Once I have my MMU2 in hand, I want to check the actual purge tower weight.

  • I did a little more exploring and found that the material estimate in Slic3r does not include the purge tower. So it's not going to be possible to compare the amount of purge between the MMU2 and the Palette+ until I can print the benchmark on the MMU2 and actually weigh the purge tower.

    It also means you can't rely on the material estimate in Slic3r with MMU2. Presumably they will fix this bug (someday).

    Napsal : 02/09/2018 7:18 pm
    Peter L
    (@peter-l)
    Honorable Member
    Topic starter answered:
    Re: Comparing MMU2 and Palette+

    Now that I've had my MMU2 for a few days I've been able to do some direct comparisons of the MMU2 vs. the Palette+.

    For this test I used a 4-color version of Gustav the Turtle at double the original size, printed with 0.25mm layers and 20% infill. This print requires 484 material changes.

  • Print Time: The MMU2 required 22:28, and the Palette+ required 16:03.

  • Purge: For both prints the model weighed about 118g. The purge block for the MMU2 weighed 80g, and the purge block for the Palette+ weighed 119g

  • Print Defects: Neither print was defect-free. Both models showed some color streaking indicating insufficient purge when changing from the black filament to white or light green. The MMU2 appears to have failed to load one filament at one layer, leading to a visible surface texture defect and (probably) weak layer in the print. The Palette+ got out of sync with the printer in at least two places, leading to layers of the wrong color, one of which is very visible.
  • it's also worth noting that I've been working with and tweaking the Palette+ for six months, and I've only had a couple days to work with the MMU. So my settings with the Palette+ have been pretty well refined, but I'm only using stock MMU settings. These test prints were done without using purge-to-infill.

    Napsal : 09/09/2018 4:31 pm
    PJR
     PJR
    (@pjr)
    Antient Member Moderator
    Re: Comparing MMU2 and Palette+


    Now that I've had my MMU2 for a few days I've been able to do some direct comparisons of the MMU2 vs. the Palette+.

    Same slicer for both?

    Peter

    Please note: I do not have any affiliation with Prusa Research. Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage…

    Napsal : 09/09/2018 6:59 pm
    Peter L
    (@peter-l)
    Honorable Member
    Topic starter answered:
    Re: Comparing MMU2 and Palette+



    Same slicer for both?

    Peter

    Yes, both sliced with Slic3r PE 1.40.0+ and identical print settings.

    Napsal : 09/09/2018 8:27 pm
    toaf
     toaf
    (@toaf)
    Noble Member
    Re: Comparing MMU2 and Palette+

    both prints looked good even without the defects. interesting to see the differences.

    I have a Prusa,therefore I research.

    Napsal : 10/09/2018 3:48 am
    airsyspro
    (@airsyspro)
    Eminent Member
    Re: Comparing MMU2 and Palette+

    Thanks Peter for taking the time to do the comparison. I have the Palette + and was wondering if the MMU would be a worthwhile upgrade. Please continue to keep us informed on any updates you may have.

    Again Thanks.

    Larry

    Napsal : 10/09/2018 1:34 pm
    nuroo
    (@nuroo)
    Reputable Member
    Re: Comparing MMU2 and Palette+

    Hey Peter. Glad u posted your experiences between the palette and MMU2. I will be in your boat whenever the Palette 2 pro ships, ORDER #2890.

    So far mmu2 has been difficult and challenging but I'm learning/tweaking. At this point having had the MMU2 2days, I'd give it the "for advanced users only" rating.

    Anyway I hope to add to this discussion with some real data. I was planning on putting the palette 2 pro on my cr10s4, but maybe I'll put it on my other mk3 that doesn't have MMU2, 1st b4 moving to cr10. That would be more of a direct comparison.

    Prusa MK3 preassembled (R2/B6) > (R3/B/7)
    Prusa MK2.5 kit > MK3 > MK3+MMU2 (R3/B/7) 😀
    Prusa SL1 3D printer + Curing and Washing Machine (day1 order)
    Taz6
    CR10s4
    Delta 3ku

    Napsal : 10/09/2018 2:49 pm
    Peter L
    (@peter-l)
    Honorable Member
    Topic starter answered:
    Re: Comparing MMU2 and Palette+


    Thanks Peter for taking the time to do the comparison. I have the Palette + and was wondering if the MMU would be a worthwhile upgrade. Please continue to keep us informed on any updates you may have.

    One thing you need to know is that it isn't really practical to have both the Palette and the MMU attached to the same printer. So if you get an MMU you should plan to move your Palette to a different printer.

    Of the two, I think the MMU is a better solution. I give the Palette+ a "C" overall, having lived with it for six months. The Palette 2 should work better if (and only if) you use it with their Canvas Hub. The Canvas Hub should allow it to stay in sync with the printer much better, but without the Canvas Hub it will have the same issues. The best I've been able to get is about one visible synchronization defect every 500 or so color changes, which means that a large print is guaranteed to have at least one visible flaw.

    I'd give the MMU2 a "B-" today, but with some firmware tweaks I can see it getting to a "B+". I've had some problems with the MMU not properly loading or unloading a filament (so far about one error every 300 or so filament changes). Some of these can be detected and fixed by the operator, and I think some can be self-cleared if the printer is a little smarter about detecting the presence and type of failure.

    (An "A" in my book would be that the unit works perfectly, out of the box, with no failures, and minimal extra work as compared to running a single color print).

    The biggest advantage I see the MMU2 having over the Palette is that the MMU2 doesn't need to splice the segments of filament together. The splices that the Palette creates are (literally) the weakness of the approach. A broken splice is not recoverable, and it seems there's still some black magic to getting the filaments to splice perfectly every time. I'll have a print going just fine and then a splice will break for no apparent reason, and the splices before and after the break will be plenty strong. This happens maybe one in every 5,000 splices.

    Napsal : 10/09/2018 3:49 pm
    Nullzero
    (@nullzero)
    Trusted Member
    Re: Comparing MMU2 and Palette+

    Thanks for the comparison information!

    Quite interesting. For me, the MMU is the solid choice. This is mainly just because I am in the Prusa "ecosystem" now, and I assume Prusa is testing with their own hardware and will generally have better compatibility with their own printers. If I had a printer like a CR-10 or something, then the Pallete seems like a very worthy contender!

    The printing time difference is pretty big. That's unfortunate. But you can tell from some of the youtube videos of the MMU 2.0, that the material switch is a somewhat slow process. At least 1 minute per material switch. So doing close to 500 material switches, and you are going to use close to 8 hours of print time JUST to change filaments. It's unfortunate, but I think its just something we will have to accept. In terms of print time, the Pallete is the clear winner here. I don't think any amount of firmware or patching from Prusa will fix this. The MMU is built to do retraction, and that is always going to take more time than the splicing method for Pallete.

    The purge block weight is interesting. I imagine this will only get better with time for the MMU. Purge to infil will be great for some applications. I think the Pallete going out of sync is my biggest worry, over the MMU. If the splicing chain goes out of sync, even a little bit, it could really reduce the quality of the print.

    Overall, from the sounds of it, you are trading a bit of speed when using MMU. But I think you will also get some big benefits for that trade off down the line.

    - Support via firmware and updates, as Prusa has control over its own ecosystem, top to bottom
    - Less material waste in the purge blocks
    - 5 materials for the MMU vs 4 materials on the Pallete
    - No cloud connectivity required (a plus, in my book)

    But if you are outside the Prusa ecosystem, I think the Pallete looks like a great product!

    Napsal : 10/09/2018 5:02 pm
    Peter L
    (@peter-l)
    Honorable Member
    Topic starter answered:
    Re: Comparing MMU2 and Palette+


    Quite interesting. For me, the MMU is the solid choice. This is mainly just because I am in the Prusa "ecosystem" now, and I assume Prusa is testing with their own hardware and will generally have better compatibility with their own printers. If I had a printer like a CR-10 or something, then the Pallete seems like a very worthy contender!

    I think if you already own a MK3, then the MMU is clearly the better choice. If you already own some other printer then the Palette is the only choice.

    The interesting question is what do you do if you are looking to buy a brand new printer and want to make it multimaterial capable? The MK3 is a darn fine printer, but there are other choices out there some of which offer things the MK3 doesn't (especially larger build volume).

    The printing time difference is pretty big. That's unfortunate. But you can tell from some of the youtube videos of the MMU 2.0, that the material switch is a somewhat slow process. At least 1 minute per material switch. So doing close to 500 material switches, and you are going to use close to 8 hours of print time JUST to change filaments. It's unfortunate, but I think its just something we will have to accept. In terms of print time, the Pallete is the clear winner here. I don't think any amount of firmware or patching from Prusa will fix this. The MMU is built to do retraction, and that is always going to take more time than the splicing method for Pallete.
    The filament change takes one minute 8 seconds, not including the purge. I suspect this will improve as people get more experience with the MMU, but it's never going to be fast.
    The purge block weight is interesting. I imagine this will only get better with time for the MMU. Purge to infil will be great for some applications. I think the Pallete going out of sync is my biggest worry, over the MMU. If the splicing chain goes out of sync, even a little bit, it could really reduce the quality of the print.

    The problem with going out of sync is a huge issue in my book. If you need your print to be free of visible defects, then you're going to have a hard time getting a large print with 1,000+ color changes to work. It doesn't matter if the Palette is faster if you have to throw out two-thirds of your prints.

    On the other hand, if a few defects are acceptable to you, then the speed might make a difference.

    But keep in mind the Palette 2 in combination with their Canvas Hub should make synchronization defects a lot less common with the Palette.
    Overall, from the sounds of it, you are trading a bit of speed when using MMU. But I think you will also get some big benefits for that trade off down the line.

    - Support via firmware and updates, as Prusa has control over its own ecosystem, top to bottom
    - Less material waste in the purge blocks
    - 5 materials for the MMU vs 4 materials on the Pallete
    - No cloud connectivity required (a plus, in my book)

    It's amazing how many people on the Palette support forums have been complaining that the new Canvas slicer will only work in the cloud.

    For the record, the Palette 2 should still work without using the cloud-based slicer. But it looks like most of Mosaic's development effort will be going to their cloud software and I'm worried that their desktop software (Chroma) could be left behind.

    The single biggest thing that will improve performance of multimaterial printing is clever ways to minimize the number of material changes, and we're barely beginning to scratch the surface of that one. For example, some slicers are starting to experiment with printing several layers in each color at a time, actually lowering the nozzle below already-printed layers to allow some batching. If you can get only one extra layer per change then you cut the number of material changes in half. This will be complicated, but I can see how there may be a lot of potential here.

    Napsal : 10/09/2018 5:57 pm
    arthur.c2
    (@arthur-c2)
    Trusted Member
    Re: Comparing MMU2 and Palette+


    The single biggest thing that will improve performance of multimaterial printing is clever ways to minimize the number of material changes, and we're barely beginning to scratch the surface of that one. For example, some slicers are starting to experiment with printing several layers in each color at a time, actually lowering the nozzle below already-printed layers to allow some batching. If you can get only one extra layer per change then you cut the number of material changes in half. This will be complicated, but I can see how there may be a lot of potential here.

    Really good point, S3D already allows to do it. With several models you can choose sequential printing and set the maximum height difference between models.

    The problem here is that models touch each others, so there's a problem when the nozzle go down from several layer heigth and print the permiter touching others models.

    I'm sure we will find some tricks !

    Napsal : 11/09/2018 8:55 am
    PJR
     PJR
    (@pjr)
    Antient Member Moderator
    Re: Comparing MMU2 and Palette+

    For me, KISS has a much better method to define the filament with just 2 settings and will optimise filament order based on purge requirements to minimise waste. My KISS prints usually are also (on average) 25% faster than Slic3r prints with MMU2.

    Peter

    Please note: I do not have any affiliation with Prusa Research. Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage…

    Napsal : 11/09/2018 9:58 am
    Nikolai
    (@nikolai)
    Noble Member
    Re: Comparing MMU2 and Palette+


    For me, KISS has a much better method to define the filament with just 2 settings and will optimise filament order based on purge requirements to minimise waste. My KISS prints usually are also (on average) 25% faster than Slic3r prints with MMU2.

    Only because you're a developer of KISS 😉

    Often linked posts:
    Going small with MMU2
    Real Multi Material
    My prints on Instagram

    Napsal : 11/09/2018 7:05 pm
    PJR
     PJR
    (@pjr)
    Antient Member Moderator
    Re: Comparing MMU2 and Palette+


    Only because you're a developer of KISS 😉

    Now, I have no idea where that came from. Totally wrong, sorry.

    KISS happens to have been my slicer of choice for the past 4+ years; I do more "work" here and have had some involvement in the development of Slic3r for the benefit of those who chose to use it.

    Peter.

    Please note: I do not have any affiliation with Prusa Research. Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage…

    Napsal : 11/09/2018 8:32 pm
    Nikolai
    (@nikolai)
    Noble Member
    Re: Comparing MMU2 and Palette+



    Only because you're a developer of KISS 😉

    Now, I have no idea where that came from. Totally wrong, sorry.

    KISS happens to have been my slicer of choice for the past 4+ years; I do more "work" here and have had some involvement in the development of Slic3r for the benefit of those who chose to use it.

    I remember you wrote a post processor for KISS MM. And Joan titled you as Master of KISS (whatever background it has 😉 ). And you're also a moderator in the KISS forum. Sorry, if my conclusion was wrong.

    Often linked posts:
    Going small with MMU2
    Real Multi Material
    My prints on Instagram

    Napsal : 11/09/2018 10:32 pm
    PJR
     PJR
    (@pjr)
    Antient Member Moderator
    Re: Comparing MMU2 and Palette+


    I remember you wrote a post processor for KISS MM. And Joan titled you as Master of KISS (whatever background it has 😉 ). And you're also a moderator in the KISS forum. Sorry, if my conclusion was wrong.

    I am also a moderator here and have been a software developer for almost 40 years. I have more than 10 times as many posts on this forum as on the KISS forum.

    When I purchased my first 3D printer, I had the choice of Cura or Slic3r. I just could not get my head around Slic3r and Cura was a whole load easier (remember that back then there were no profiles available, so the learning curve was very steep).

    Cura had some major flaws and I stumbled upon KISS. After playing about with it for a while, I realised that it was simply so much better (in terms of use and result) than anything else available. Unlike some other slicers, the G-code generated by KISS was so efficient and easy to work with, so writing a post-processor was rather simple.

    The biggest thing with KISS was that it was an excellent free program with the option to purchase; I purchased the "PRO" version a year before I needed to. Most of the other slicers have continued to develop, but still do not have the capabilities of KISS. Yes, there are ares where KISS lags behind, but I have not found them to be a deal-breaker.

    With all that being said, I realise that we have managed to hijack this thread, so let's move back to topic (please).

    Peter

    Please note: I do not have any affiliation with Prusa Research. Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage…

    Napsal : 11/09/2018 11:17 pm
    toaf
     toaf
    (@toaf)
    Noble Member
    Re: Comparing MMU2 and Palette+

    i guess I will be curious too see if any of the (counting filament steps for changes) will fallow to the mmu. I can see the up sides and down sides of it. it does seems like it should work in theory. but then again. if the ETA of a print cant be estimated, no way they can guess the amount of filament.

    KISS army !!

    I have a Prusa,therefore I research.

    Napsal : 11/09/2018 11:26 pm
    Vojtěch Bubník
    (@vojtech-bubnik)
    Member Admin
    Re: Comparing MMU2 and Palette+

    > For me, KISS has a much better method to define the filament with just 2 settings and will optimise filament order based on purge requirements to minimise waste.

    Would you please expand on that? Thank you.

    Napsal : 27/09/2018 4:48 pm
    PJR
     PJR
    (@pjr)
    Antient Member Moderator
    Re: Comparing MMU2 and Palette+


    > For me, KISS has a much better method to define the filament with just 2 settings and will optimise filament order based on purge requirements to minimise waste.

    Would you please expand on that? Thank you.

    Vojtech, you have the program; you could have a look. And I tried to explain to Lukas as well in the early alpha I received.

    Simply this:

    Ramming and purging speeds are based on max volumetric (90% thereof, I believe).
    Ramming volume is defined for the printer (at present, but may change).
    Each material profile has a "Base Purge Volume" and a "Bleed" value
    Extruders are ordered so that waste is minimised
    Purge octagon is automatically positioned and each layer is printed in opposite directions.

    Peter

    Please note: I do not have any affiliation with Prusa Research. Any advices given are offered in good faith. It is your responsibility to ensure that by following my advice you do not suffer or cause injury, damage…

    Napsal : 27/09/2018 9:26 pm
    Stránka 1 / 6
    Share: