[Closed] Speed/Quality of MMU vs Single  

Re: Speed/Quality of MMU vs Single

Hi root!

The MK3 is still on pre-order, so no-one has received a production unit yet.
However, speed is specifically one point that is supposed to be improved on the MK3 (new drivers, stiffer frame) and flexible filaments should be handled better as well with the direct drive Bondtech extruder. Print quality is of yet unknown, but I don't see why the MK3 should be worse than the MK2S. Maybe it will take some updates to the profiles, but the end goal is certainly the same if not better quality.

BTW, if you order the MK3 now you have to wait at least until January to actually receive it.

Re: Speed/Quality of MMU vs Single

Hi root!

The MK3 is still on pre-order, so no-one has received a production unit yet.
However, speed is specifically one point that is supposed to be improved on the MK3 (new drivers, stiffer frame) and flexible filaments should be handled better as well with the direct drive Bondtech extruder. Print quality is of yet unknown, but I don't see why the MK3 should be worse than the MK2S. Maybe it will take some updates to the profiles, but the end goal is certainly the same if not better quality.

BTW, if you order the MK3 now you have to wait at least until January to actually receive it.

Thanks! Yeah I know about the MK3 upgrades over the MK2. It looks like it will be a noticeable improvement in speed over the MK2. I don't have a problem waiting until January either. However the thing that finally convinced me to pull the trigger on the MK3 purchase was when I saw the MMU online. So I just wanted to know from anyone that has the MMU for the MK2, did you have to slow down your prints with it, over the stock/single extruder? Was quality any different? Those are my two biggest concerns, which I assume they aren't justified. I haven't read anyone directly comparing say single color print(s) on the stock/single extruder vs a single color print on the MMU and noting speed/quality differences, if any.

Thanks guys!

Re: Speed/Quality of MMU vs Single

I upgraded from the MK2S to the MMU upgrade and have been regretting it since. I have not been able to get the quality out of the MMU extruder and am about ready to rip it off and send it back.

Nathan

Re: Speed/Quality of MMU vs Single

I upgraded from the MK2S to the MMU upgrade and have been regretting it since. I have not been able to get the quality out of the MMU extruder and am about ready to rip it off and send it back.

Nathan

Yikes! I'm sorry to see that. Have you received any help from Prusa or the forums? That definitely seems unusually/unacceptably bad. I'm hoping that quality/speed isn't a common tradeoff with the MMU... I may not purchase the Prusa all together, and go with some dual extruder.

Re: Speed/Quality of MMU vs Single

I haven't read anyone directly comparing say single color print(s) on the stock/single extruder vs a single color print on the MMU and noting speed/quality differences, if any.

I have been printing with the MMU since the end of August. A LOT. My waste tower bucket is now about 1.5kgs.

In my experience, the MMU takes some dialing in, but the print quality is as good if not better with PLA. I have not done much else with the MMU (did some TPU and PETG prior to the upgrade). You can't get to the very low layer heights (due to the gearing of the new extruder gears) but relative to the same setting the quality is the same for me. Maybe a bit better due to the more positive extruder control.

As far as speed goes, in single color it is the same. In multi-color, you have the significant extra time of the purge tower. But then again, you get a four color print!

Re: Speed/Quality of MMU vs Single

I only see a huge difference in stringing, but maybe I can get that fixed. So no opinion as of yet, except for the fact that it's a lot more tinkering.

Re: Speed/Quality of MMU vs Single

I haven't read anyone directly comparing say single color print(s) on the stock/single extruder vs a single color print on the MMU and noting speed/quality differences, if any.

I have been printing with the MMU since the end of August. A LOT. My waste tower bucket is now about 1.5kgs.

In my experience, the MMU takes some dialing in, but the print quality is as good if not better with PLA. I have not done much else with the MMU (did some TPU and PETG prior to the upgrade). You can't get to the very low layer heights (due to the gearing of the new extruder gears) but relative to the same setting the quality is the same for me. Maybe a bit better due to the more positive extruder control.

As far as speed goes, in single color it is the same. In multi-color, you have the significant extra time of the purge tower. But then again, you get a four color print!

Great, thanks for the input! Interesting about layer heights, how low can you generally go with it for a high quality print? Yeah I've seen how much the purge tower wastes, definitely a major trade off for the functionality - but I'm okay with that.

Re: Speed/Quality of MMU vs Single

Great, thanks for the input! Interesting about layer heights, how low can you generally go with it for a high quality print? Yeah I've seen how much the purge tower wastes, definitely a major trade off for the functionality - but I'm okay with that.

The MMU can only go down to 0.15mm. (People have experiment with a bit lower.) The base unit can go to 0.05mm (which is only useful for very very small things - and really needs a smaller nozzle to be effective.) but the "good" quality is 0.1mm. The MMU does 0.15mm and 0.2mm which is fine for most of my printing needs.

Re: Speed/Quality of MMU vs Single

The MMU can only go down to 0.15mm. (People have experiment with a bit lower.)

It can go lower. I use variable layer height feature and with thiis it goes down to 0.07, which is, in most cases, no real difference to 0.05. I even printed with the 0.15 mm nozzle from e3d and I am getting a stunning quality.

However, it took quite a long way until there: After the MMU Upgrade I had to cope a loss of print quality:

- rough/uneven vertical surfaces
- stinging

For example I tried printing a little M5 screw. On the Mk2 it was no problem with 0.05 mm standard settings. On the MMU, also 0.05 mm standard settings, the screw is not that good, but still okay.

The real problem is stinging in my optinion. This has increased dramatically since the MMU upgrade. However it is not a big problem because nearly every object is surface finished anyways. A little lighter or boiling water for PLA and stings are gone. Same after ABS smoothing with acetone.

The solution I may will work on when redesigning the extrudder of the mk3 is using an additional direct drive extruder so that I can print flexible materials and rreduce retraction to 0.8 mm again. Less stinging expected. In my optinion, that is the way Prusa should go with a may coming MMU 2.0 Upgrade.

Re: Speed/Quality of MMU vs Single

The solution I may will work on when redesigning the extrudder of the mk3 is using an additional direct drive extruder so that I can print flexible materials and rreduce retraction to 0.8 mm again. Less stinging expected. In my optinion, that is the way Prusa should go with a may coming MMU 2.0 Upgrade.

great post. and that quoted part is really what should PR do in the future with the MMU

Re: Speed/Quality of MMU vs Single

The solution I may will work on when redesigning the extrudder of the mk3 is using an additional direct drive extruder so that I can print flexible materials and rreduce retraction to 0.8 mm again. Less stinging expected. In my optinion, that is the way Prusa should go with a may coming MMU 2.0 Upgrade.

I'm also working on this, with my not-MMU i3MK2. See here, but I'm no longer trying to do it with a camstruder. Not enough space.
I tested a modified version of the MMU filament tip shape gcode with great results: Also the MK2 forms a usable tip that can be reloaded. But it will take some time of the winter, as many system components are needed. I want to use an Arduino to control the filament change system to keep it simple. So all the FW has to do would be to send a number which correspondents to the wanted filament and wait for an "OK" signal.

Re: Speed/Quality of MMU vs Single

I'm also working on this, with my not-MMU i3MK2. See here, but I'm no longer trying to do it with a camstruder. Not enough space.
I tested a modified version of the MMU filament tip shape gcode with great results: Also the MK2 forms a usable tip that can be reloaded. But it will take some time of the winter, as many system components are needed. I want to use an Arduino to control the filament change system to keep it simple.

You are talking about a custom MMU Upgrade I guess. I have read your linked post and needed a little moment to understand the solution with the camstruder/camshafts. Pretty interesting idea I did not think of before. However, I think it may could be even simpler (if Prusa MMU is given):

My approach would be mirroring the stepper-input before the multiplexer to a Bondtech extruder installed right between the Y-splitter and the PTFE tube from the hotend. I guess there will be enough space, even if it will be a little challenging. If the retraction distance to the metal pipes remains the same, no FW change is needed at all 🙂

Cons (or problems);
-> The only real problem I see at the moment is the signal mirroring for the steppers. A fast Google-search did not bring any valuable results. If that does not change, some mods on the Einsy or the FW may be needed.
-> More weight on the extruder head. But it should be only a little more than with a Standard Mk3. No big deal.

Pros:
-> Printing ALL materials just like the Standard Mk3, including flexibles!
-> Much higher pressure on the nozzle is possible! I am thinking of a 0.1 or even a 0.05 mm nozzle... 😀
-> Less retraction needed. Less stringing expected.

However, I will not have time in Jan/Feb due to my exams. Fortunately, the MK3 Upgrade Kit and the MK2.5/3 MMU Upgrade Kit will arrive just right (end of Feb, beginning of March). I will then have a stepper/extruder and some additional spare parts from the upgrade leftover needed to make this solution work. So I won't begin with it before February due to missing parts, but I am willing to help if someone uses the same approach in the meantime.

Re: Speed/Quality of MMU vs Single

Yes my device would be a custom thing.

My approach would be mirroring the stepper-input before the multiplexer to a Bondtech extruder installed right between the Y-splitter and the PTFE tube from the hotend. I guess there will be enough space, even if it will be a little challenging. If the retraction distance to the metal pipes remains the same, no FW change is needed at all 🙂

If I understand it right, always two steppers would drive the filament: One in the X axis (direct drive) and one at the frame (MMU style). I think that might lead to some problems, beyond the reversed running direction. This system is over-constraint. All the little imperfections you have in real world systems will result in the one extruder pushing a little bit more filament over time than the other. As the filament can't go somewhere, you will end up with high stresses inside the filament up to a point where one of the steppers skip a step to release the forces again.
This approach would be extreamly easy, but I don't think it will work. Maybe for small parts with short used filament lengths, but not for long time printing.

Also, you need to provide 2x the electrical power from the stepper output if you want to drive two steppers - like on the Z axis.

To work around this, my system is designed to push the filament into the direct drive extruder, but as soon as it catches the filament change motor is unclutched from the system. For filament change retraction, first the extruder stepper reverses the filament all the way up and out of it's range, than the filament change stepper grabs this filament and retract it behind the splitter. This way, always one motor is pushing the filament.

Re: Speed/Quality of MMU vs Single

This system is over-constraint

Definately. However, I appreciate the simplicity. I would want to make a test print to see if there are the pressure-issues. As you said, it should be no problem for small parts so I might guess it would go okay for printing in MM-Mode more than printing in single mode. Damn, now I am more willing to build it than before just to try it out in practice 🙄
There would be another "solution" (yes, I know this will be more like a duct-band solution): To keep it simple, my workaround would be making sure the x-extruder has a higher torque than the extruder on the frame. This way, I would take the bitter pill of a may skipped step from time to time, but at least it does not affect extrusion rate that much (only if the filament gets rough from skipping, but let's leave that aside).

I also remember printing TPU with the MMU and it tangled around the gears in the frame-extruder due to a blockage. So instead of a skipped step, that may happen too... . On the other side, flexibles may have the "ability" to compensate the little difference considering the pressure due to their elasticity. I just don't know what will happen.

❗ However, I really hope that the new Mk3 feature of detecting skipped steps does not apply on the extruder but that the filament sensors are used alone to detect a filament blockage. Otherwise this approach will also get more and more complicated.

To work around this, my system is designed to push the filament into the direct drive extruder, but as soon as it catches the filament change motor is unclutched from the system. For filament change retraction, first the extruder stepper reverses the filament all the way up and out of it's range, than the filament change stepper grabs this filament and retract it behind the splitter. This way, always one motor is pushing the filament.

That would be the mechanical correct way, ensuring the system is not over-constraint. It is also more efficient considering electrical power. But not only is it more complicated, you also lose the advantage of about double the pressure on the nozzle.

I think a practical example is needed. Now it's waiting time...

Re: Speed/Quality of MMU vs Single

I agree with marvin in the simplicity of his solution, but… agree with sebastian in that it’s constrained.
I think the “solution” of one additional direct extruder (only working after the bowden extruder has pushed the filament and catched it) is more reliable and has other benefits:
- only one motor is consuming energy at once
- only the direct extruder must do pression through the nozzle
- the bowden extruders can be motors with less torke (an less comsumption, and cheaper)

The main problem are the changes in firmware.

We must consider the direct extruder motor as if it’s any of the other extruding motors in “printing” phase, (the other motors will be off during this phase)
During the “pushing” phase the selected motor will be on, and the direct extruder motor will be off.

The question is how determine in wich phase are de print job:
pushing: only while tool change (pull or push filamente in steel tubes, etc.)
printing: while extruding material or retracting…

ps. sorry for my “spanglish”

Re: Speed/Quality of MMU vs Single

I want to keep it simple, therefore I'm going this way:
- The filament change motor will be no stepper at all, as it's not necessary to do precision moves with it. I will use a standard geared DC motor to drive all loaded filaments during change, while a camshaft selects the driven filament.
- A seperate Arduino Uno or simmilar will control the servo for the camshaft, DC motor and necessary sensor switches. Only a very basic change to Marlin / Prusa FW is then necessary: Unload filament with extruder stepper, call the next filament number, wait for "Go" from the change unit. Continue printing.
- To get the feeding and unclutching process smooth, the DC filament change motor is connected to the drive shaft by a torque limiter. This limiter is adjusted so the maximum force to the filament can be pushed will be around 200 gram (2N). This way, the DC motor keeps pushing the filament tip against the steppers hobbed bolt for some seconds while the "OK" signal is sent to Marlin / Prusa FW after the filament change and the torque limiter keeps the drive power to 2N. When the steppers hobbed bolt grabs the filament tip and starts pulling it, it can easily start printing during the unclutching process of the change unit because the stepper can pull much harder, which will open the torque limiter. Hope you get what I mean 😉

Re: Speed/Quality of MMU vs Single

When the steppers hobbed bolt grabs the filament tip and starts pulling it, it can easily start printing during the unclutching process of the change unit because the stepper can pull much harder, which will open the torque limiter. Hope you get what I mean 😉

I get what you mean. But with this in mind, I see less difference to my solution considering the mechanical-system constraints: Your torque limiter at the clutch can be seen as my improvised torque limiter aka "weaker stepper motor" which skips steps if the force is too high.

Re: Speed/Quality of MMU vs Single

True, but to fine tune the steppers torque you need to run it from an independent driver as an independent axis. If it's just a slightly wearker stepper like a pancake one, I guess it's still too stronge. But we will only know fore shure if someone test it 🙂

Re: Speed/Quality of MMU vs Single

I'm confused - Does the MMU replace the direct drive head and turn it into a (four-way) bowden drive or do the four motors just load and unload the filament from the direct drive head? If it's the latter, why would there be any significant change in resolution / quality or ability to print with flexible material?

I am eagerly awaiting my MK3 and I did order the MMU a bit later, but after reading this thread I am kind of reconsidering the MMU.

One other basic question but - I am less interested in multi-material prints than just being able to software select single color prints without having to manually change out the filament (yes, I am lazy!). Is there any reason that this is impractical? What I mean is - with the current software can it heat up and use one filament without a great deal of overhead or waste / priming of the unused colors?