PLA Heat Creep - SOLVED

[KelVarnson]

At least I think I have it solved.

Through my extensive trials and errors over the last few weeks, I have learned that adding more cooling air to the heat sink is not a reliable solution to the well-known PLA heat creep issue. The thermal interface between the heatsink and the metal filament tube is very poor, apparently by design. I can only speculate that the heat sink is intended for some other purpose than to actually cool the filament itself. In fact adding a second blower to the heat sink did nothing to reduce the incidence of heat creep in my machine. I even tried increasing the surface area of the fins by using slotted washers with thermal compound, and that actually made the problem much worse, I suspect by picking up additional heat rising from the heater block.

After careful study of the drawings for the heat sink and the filament tube, I decided that my best hope was to directly cool the filament itself, to prevent it from swelling above the brass insert and getting jammed. I took the stock Lulzbot extruder body and added a 20mm x 20mm x 8mm fan to the front. The fan blows into a cavity that feeds four longitudinal vents, equally distributed around the filament path. The vents are essentially slots that run parallel to the filament path. I added a little recession in the bottom to allow the air to flow past the top of the heatsink, and I also rotated and expanded the “D” detail at the bottom of the extruder body 180 degrees so that it ends up venting out through the slot in the mounting plate. The result is that the air flows along the entire filament path, and vents out of both the top and bottom of the extruder.

I have since been able to print several 4+ hour PLA prints with PVA support, something I had been unable to do for quite a while. The fan I selected has a low enough profile so that two of them will fit facing each other on the Dual Head V2. I have only installed one so far, but there is clearly room for the other, and I don’t think they will fight too much for air, as I don’t think a whole lot of airflow is required along the filament to cool it enough to prevent heat creep.

I’ve attached stp and stl files for the part. I will attach photos shortly, I don’t have any good pics yet because the one that I printed in ABS is installed in my printer. I have since printed two more in PLA with PVA support (support is necessary for the fine details of this part). The PVA is being dissolved right now, I will post photos when that is done. It’s probably easier to visualize the details of this modification by looking at the stp or stl files, it might be kind of hard to see in photos.

The fan I used can be found at Digi Key, and I wired it right in parallel with the 5V micro blowers that are blowing on the heat sinks:

http://www.digikey.com/product-detail/en/sunon-fans/MC20080V1-000U-A99/259-1544-ND/2757793

It is mounted to the front of the extruder using 2-56 screws.
Extruder Body, Vented, 2016-04-01 01.stp (1.66 MB)
Extruder Body, Vented, 2016-04-01 01.stl (395 KB)

[KelVarnson]

Here’s some photos of my modifications to the stock extruder. First, a side-by-side showing the front of the stock extruder vs. the modified one:

A close-up of the cavity shows the internal structure of the vented filament path. It’s a little hard to see, but it is basically a hollow column with four longitudinal slots, equally-spaced around the filament. The filament path has the same I.D. as it does in the stock extruder. The fan cavity wraps all the way around the back and left side of the column, hopefully providing airflow “somewhat” equally to all four flutes. These details require support when printing in order to turn out well:

Looking at the bottom of the stock (top) and modified extruder bodies, you can see the four flutes along the filament path. You can also see the depression that I added to allow airflow past the top of the heatsink, and that I enlarged the “D” shaped detail and rotated it 180 degrees. The plan is that the air comes down through the four flutes, gets past the top of the heatsink via the little channel, and then exits through the flat side of the “D”, which is oriented over the slot in the aluminum mounting plate. Before I added these details, the airflow out of the bottom was completely blocked, and heat creep was still occurring:

This photo is a little rough, but it shows the four flutes from the top of the filament path. It’s a good idea to use something to clear the flutes out after printing. I used a coping saw blade, but I haven’t done that on this specimen yet, so you can see that the lower right flute is slightly blocked. These are fairly fine details, although they seemed to print out pretty well here at 0.25mm layer height in PLA with PVA support:

Finally, here’s a pic of the fan just sitting in place on the front, not screwed down. The Sunon fan only has three screw holes, I put four on the extruder because I wasn’t sure what the best orientation would be. By the way, I made the decision to have the air blowing IN to the cavity, rather than drawing out. This was just based on gut feel more than anything, and also the thought that if any filament grinding does occur, it will be blowing air out of the top of the extruder body, perhaps reducing the likelihood of debris clogging these tiny little air channels. Also, I didn’t want to draw heat UP from the hot area below the extruder body:

Before this, I had gotten to a point where I was barely able to print at all with PLA due to heat creep. It didn’t start out that bad at first, but it got progressively worse. I am not sure what to attribute that to, perhaps buildup on the inside of the metal filament tube reducing its ability to conduct heat away from the PLA? I don’t think it was due to a decline in the micro blower performance, because I added a second micro blower in an attempt to alleviate the problem, and it had no effect whatsoever. It had become clear to me that my best shot was to cool the filament directly. My first pass had the flutes over the entire length, but was not venting out of the bottom, and as I said above, that did not work. Adding the channel on the bottom has resulted in 100% success, at least so far. I used the unscientific method of holding a piece of vinyl tubing up to my ear and using the other end to sample airflow out of the top and the bottom of the extruder, mounted in the machine, and the flow seemed quite robust. So I think it is working as intended.

Thanks for looking. I welcome all comments and suggestions.

[kcchen_00]

That’s an interesting mod. Where does the airflow exit once in the hotend?

[KelVarnson]

To make sure I understand the question, I might need to fully understand the terminology used around here (I’m kinda new at this). I understand “hot end” to mean the heater block and the nozzle. Do I have that right?

The air coming out of the bottom of the extruder body exits through the open side of the slot on the aluminum plate that the heatsink slides into and that the extruder body mounts to. There shouldn’t be any appreciable amount of air flowing down into the filament tube and into the heater block and nozzle, if that is what you mean. That path is effectively blocked by the filament.

Do you have a concern that the exiting air could influence the print? I just want to make sure that I understand the question.

[kcchen_00]

The hotend comprises of the heatsink, heat block and nozzle.

I’m just wondering about the flutes you’ve made along the side of the filament. I assume they serve to cool the filament and provide airflow into the hotend assembly. Once in the hotend assembly, where does the air flow? If there’s no exit, how does new / cool air circulate?

Just re-read your reply…

So the flutes only serve to cool the filament at the exit of the extruder body, and flows out the aluminum plate slot… Essentially, “super cooling” the filament to combat heat creep in the hotend.

Interesting design.

[KelVarnson]

Thank you for the clarification of the terminology, and the compliment. Yes, the air flows along the filament only when it is in the extruder body region, but it should be cooling it along that entire length.

I am counting on the paths of least resistance being the top and bottom exits of the flutes, making it impossible to build up enough pressure to force any air into the hot end. And as I said, I have unscientifically verified that there is appreciable airflow from both top and bottom exits.

I’d like to think that this approach will work 100% of the time while printing PLA, but the only way to be sure is to get significantly more prints through it, and/or for others to give it a shot and share their results.

[killacycle]

Eva (my wife) and I have had real problems with glow-in-the-dark PLA and other filled PLA jamming (grinding).
After months of occasional failed parts, I think we finally have figured it out.
We too used a 20x20mm fan, but as a substitute for the tiny blower on the finned section of the hotend.
We installed this nifty new duct and 20mm fan and we haven’t had a jam since. Works great. Awesome prints.
We posted the duct I designed on Thingiverse here:
http://www.thingiverse.com/thing:1461335
(We bought our 20x20mm fan on Ebay for ~$3 which is one third the Digikey cost.)

The problem seemed to be that the PLA filament partially melted and swelled a bit just before it entered the heated section, while it was still in the finned section of the hotend. The filament would then grind (strip) during retraction. The E3D hotend doen’t do this, and it has a similar cooler design, but uses a 30x30mm fan because the finned section is longer. I used a 20x20mm fan and designed a duct to fit the shorter Hexagon hotend. So far, it works like a champ.

Bill D.

[KelVarnson]

The part you wrote about it jamming during retraction from the heat sink area is particularly interesting to me, because of what I just experienced today. We started seeing skipping/grinding on a print again, even with the fan installed (the subject of this thread). But, it wasn’t a 100% failure like I had seen before, it did manage to keep printing.

Before, there was obvious swelling of the filament, with a distinct “step” left in the removed filament, where it jammed against the top of the brass bushing.

So I may try your duct and fan idea, thank you.

[killacycle]

We often found that we had to extract the jammed filament out of the extruder with a pair of pliers, carefully leveraging against the plastic extuder frame. This led us to discover the bulged/swelled filament, just before the melted end. In turn, we noticed the jams were mostly happening during retraction.

We print with Inova and PLA mostly, and we had no jamming problems with Inova. We have been hunting down this PLA jamming problem for months.

I designed and printed the small duct and put it on the printer, and it was like flipping a switch. Problem solved.

The new Taz6 comes a 30mm (or 40mm?) hotend muffin fan and a duct similar to mine. They dumped the small blower, likely because it caused some cooling problems (and cost a bit extra for that tiny blower.) There are a number of 30mm muffin fan Taz hotend retrofits on Thingiverse. I think my design and yours are the only ones that uses a 20mm muffin fan. I figured that the 20mm fan seemed to deliver the CFM needed and tucked out of the way nicely under the mounting plate.

Bill D.

[KelVarnson]

Thanks for the detailed info, Bill, I appreciate the response. Yesterday’s failure was particularly disappointing for me, because it had been working much better than it had in a long time.

Virtually all of my PLA failures had exhibited the swelling and the “step” I described, which was an indentation from the brass bushing. Once in a while I would get one where the filament had to be pulled with pliers, exactly as you described, but most times the filament would still retract, but then would jam again as soon as it got to the “step”. So I would say generally that most of our failures did not involve retraction. By the way, I typically use around 16mm for retraction when switching between heads.

I’m glad you had good results with the larger heatsink fan. I may try it, but I am a little skeptical, because a.) adding a second blower did not fix the problem for me, and b.) looking at the drawings for the heatsink and the filament tube it looks like the designers were trying to LIMIT thermal conduction between those two parts. I really do think that it’s possible that I am not fully understanding their intent with this setup, if anyone thinks they can clarify this for me, please step up.

Since most if not all of my failures involve swelling of the filament in the area right above the bushing, and I have had at least partial success with my redesign, I may change it to focus even more airflow on that particular area. For instance, I am thinking maybe removing the flutes that exit out of the top, and increasing the clearance in the area right above the top of the heatsink. What would be really slick is if I could use the same fan to accomplish this while simultaneously cooling the heatsink, but looking at the assembly this seems like kind of a stretch.

I am glad you found a solution to your issues. Thanks for sharing the information.

[kcchen_00]

I found this video from the E3D guys pretty interesting to help understand hotend design:
https://youtu.be/mk30MPLoYf4
Note - our hotends are not E3D, but I haven’t come across similar information for the hexagon. The design principles should apply regardless of the hotend.

At the 10min mark, they start talking about the filament transition from solid to viscous state in the heat break.

I’ve always tried to ensure the small blower points a bit down towards the last fin (thick) in the heatsink assembly. Mitigate the heat propagation closest to the source (heat block). I think the hotend could benefit from a greater heat break distance or a heat shield before the heatsink.

Interesting to note that the original Hexagon design (and the E3D for that matter) has a longer heatsink… and both have 30x30 fans. Keeping the filament cool close to the heat break seems to be the key.

[killacycle]

The fan duct surrounds and cools only the finned section of the hotend. It does not surround or direct air around the lowermost section of the hotend. The finned section is supposed to dissipate the heat from the tip of the hotend. The small blower helps do that, but not evenly or nearly enough for some filaments, such as PLA.

The newest Taz6 Hexagon hotend has what looks to be a similar duct and fan to mine:
https://devel.lulzbot.com/TAZ/Mangrove/photos/Mangrove_Photos/mangrove_toolhead.JPG

It looks like, on at least this prototype, that they are using a 30mm, perhaps a 40mm fan. My hotend cooler uses a 25mm muffin fan. (I notice that my previous post said “20mm,” that was a typo.)

They also are putting on a second post cooling fan which improves the print quality quite a bit. I have done this (or the equivalent) on my printer extruders.

Bill D.

[KelVarnson]

I think I need to make a correction to something I have been saying all along, my statement that the thermal interface between the cooling block and the filament tube seems to be poor by design. On a second look at the drawings, I see that the OD of the tube is 4.0mm, and the ID of the cooling block is 4.1mm. I had the erroneous impression that the gap was much larger.

I do wonder if there is any thermal compound between these two components. It would seem like filling that gap with heatsink compound might make a significant improvement.

[killacycle]

Here is a link showing a Hexagon hotend disassembled:
https://tamarintech.com/article/hexagon_std_assembly

As you can see, the finned section is separate from the heated section. The “heat break” is a separate threaded stainless steel tube. Stainless is a poor conductor of heat, so it isolates the finned section from the heated block section. The heat break is screwed into both the the heater block and the finned section. A set screw holds the heat break into the finned section. There is a small center section on the heat break that has the threads removed. This is to further insulate the finned section from the heater block.

The idea is to keep the finned section cool and the heater block hot. My fan duct cools the finned section better and more uniformly than the stock blower manages to. It is a smaller version of the duct that E3D uses on their very nice hotend.

I would have used the E3D duct and fan, but their hotend is longer, as is their duct. Thus, I made my own for the Hexagon hotend.

[CoParaTech]

I’ve recently been experiencing heat creep on my Mini with several different filaments and been having jams/stripping with the prints that have a lot of retractions. I found that the little fan was clogged and working intermittently. I noticed that killacycle’s fan mode will also fit the Mini so I ordered a 25mm fan and am going to give that a go and see if it fixes that.

[CoParaTech]

Thanks killacycle, your fan mod worked on my Mini! Heat creep solved and no more grinding or jams!

[KelVarnson]

Killacycle, thanks for the great info and the shroud design.

I have a dual head, it’s a little more crowded and your exact design won’t work, but I am planning on modifying the approach to make one that fits. I’ll post it here when I get it going.

Thanks everyone, for the help.

[killacycle]

I almost have a dual head duct for a 25mm muffin fan. I have gone through several design iterations, each getting the fan closer and the duct more compact.

The duct fits, but it rubs up against the heated block. I don’t like that so much. It will melt and then it blows directly on the heater block. It also will stink when it melts. Going to design a smaller one next, but I’m not sure exactly how I am going to keep it from contancting the heated block. It is a very tight fit for the muffin fan on the dual head.

I was also considering designing a duct for the stock blower. A duct could really help confine and direct the air from the blower around the heatsink. Should be nearly as effective as the 25mm muffin fan, but it has slightly lower CFM. Blowers typically can tolerate a bigger back pressure that you often get from a duct, so the tiny blower could work equally as well as a 25mm muffin fan when pushing into a duct.

Bill D.

PS
I am glad to hear that my first duct design works as well for you as it did for me. We are happy we could find a solution that really works and solves a very difficult to diagnose problem.
Please post a “I Made One” on Thingiverse to help spread the word.
Lots of folks post things that don’t work and often can’t even be printed, so a confirmation “Made” photo really helps.

[nopick]

The newest Taz6 Hexagon hotend has what looks to be a similar duct and fan to mine:
https://devel.lulzbot.com/TAZ/Mangrove/ … olhead.JPG

I have been using the Taz 6 style toolhead with a 40mm 5V Noctua fan for a couple months. It cured the issues I had with the small squirrel cage stock fan. I think that small fan works fine for ABS but is just barely adequate for PLA printing. Consequently, when it has any reduction of airflow due to a sticky bearing or crud in the fan, it allows some heat creep and causes sticking in the cold section of the hot end.

[KelVarnson]

Hey Killacycle, I just read about your shroud design for a 25mm fan for the dual head. I have designed a shroud for the same 20mm fan I was using on the extruder body, and even with the smaller fans it is pretty danged tight! Although, I am using a separate 20mm fan and shroud on EACH extruder heatsink. I have it installed and running, and it seems to be working well. I completed an 11 hour print with PLA/PVA last night, and I am 5 hours into what is probably going to be a 30 hour print right now using the same materials. So far, no hiccups - fingers crossed. By the way I still have the extruder body fan installed that was the original topic of this thread, but added a switch to disable it. For a true test of whether the heatsink fans and shrouds are going to fix this, I have the extruder body fan turned off for these prints.

So here’s what I have done. Shown are a pair of shrouds, with a reversible mounting clip that I designed at the same time. I printed them in ABS. The holes are sized for 2-56 hardware:

I added a layer of copper foil on the surface nearest the heater block, just for good measure:

Here’s how I have them installed:

I scrapped one of the extruder bolts and nuts, the one underneath the stepper motor. I replaced it with an M4 x 35mm machine screw with a 7mm hex head (incidentally this solves the problem of trying to fish that nut in there during reassembly, which was a huge pain). Underneath my custom mounting clip is a 6mm tall spacer with a 4mm clearance hole. The screw comes down through the spacer, through the clip, and another nut goes on the end to hold it all together and hold the extruder down (I printed the clip at 100% infill so that it could withstand the force needed for that task).

I had originally intended to place the fans both on the same side of the head, which is why I made the clip reversible. However, I found that it works better for me as shown, with the fans diagonally opposed, mainly because the shrouds don’t interfere with the thermistor wires. Everything just barely fits, absolutely no room to spare! There is about a 2mm gap between the bottom of the shroud (copper foil side) and the top of the heater block. No sign of the shrouds melting or discoloring as of yet.

Here’s the whole thing put back together. I also added a couple more machine screws through each of the print cooling shrouds to hold them up, they were beginning to sag (not surprisingly):

I have attached STL files for both parts.
Shroud 2016-04-19 01.stl (80.4 KB)
Clip 2016-04-19 01.stl (25.3 KB)