E3D V4 All metal hotend

[JohnStack]

Ok, I'm slicing my E3D in half the next time it jams. Not.

No ideas here.

I print PLA at 190 or 195. Some of them seem more gooey/stringy at temp than others. Blue and red at 190 and green at 195.

[mhackney]

I understand (and experience) your frustration. The Kraken has the same basic mechanical setup as the E3D (I have both) but with water cooling. I find it is less prone to PLA jamming but just when I have it dialed in, printing reliably and very nicely, odd filament starving issues pop up once again. I've slowed down, sped up, increased temps, decreased temps, polished, deburred, scrutinized, dissected, injected, rejected, ... you get the point (with a tip of the hat to Arlo Guthrie). I've not had a single jam with ABS with these hot ends.

I've also experienced the filament chewing phenomenon and worked on the extruder end of things - cooling fan on the driver, tuning the current setting, debarring the sharp knobbed gear, etc. Each time I do something things seem really good for a while. And when I say good, I mean the best PLA parts I've ever printed.

Just yesterday, after reliably printing dozens of parts, things started going astray again (filament starving). I thought I had addressed and at least thought about all of the problem possibilities and even combined, this is still occurring. Once I get into this state, I disassemble the hot end, clean, ream and reassemble and am good to go - for awhile. It is nearly impossible to determine where the issue is occurring and I've tried to look at the bore with a scope, the filament itself when I remove it, etc.

I am starting to suspect and think about the heat break geometry or nozzle itself. I do not think the brass-SS joint is at fault in my situation. I've polished both ends (the brass nozzle and SS heat break), assembled at 300°C and when I disassemble and look later this joint is mirror like with no evidence of leaking.

...

While I was writing this I was able to pull a complete filament out of a "jammed" hot end (Kraken) by chilling it down. This is a deburred and polished heat break and had been printing well for over a week (LOTS of parts). I used my little USB microscope to scrutinize things and do some more experiments. Here's what I captured:

Note that you can even see the .4mm tip that was in the nozzle. The dirt and stuff on the filament are my fault, they came from the brass threads (oxidation). A couple of points:

The .40mm tip section is pretty long, something like 2mm
You can also see a ridge where the plastic filled the chambers between the heat break and nozzle. You can see that much clearer here:

The other thing I was able to capture was where the cold break occurred.

You can clearly see the enlarged diameter of the PLA right up to thee top of the narrow heat break and where the water cooling would take effect. (Note also that I bent my heat break ) I have 3 others with the Kraken so I have more to play until a replacement arrives.

The other think interesting about this photo is that the melted PLA is clearly well below where the PTFE tube inserts into the top of the break. I had theorized that this is where the jamming might occur. I've since been able to dissect a second plug today and it shows the same thing.

And that led me to some experiments.

I removed the hot end assembly from the water jacket and heated it up to temperature. I inserted a 2" section of PTFE in the top of the heat break and then pushed some PLA down the bore. This was the heartbreak and nozzle shown above after pulling the plug out. As soon as I pushed, PLA flowed very easily for about 1cm and then more and more pressure was required. I tried this 4 or 5 times to verify. The temp was 195°C. I then started ramping the temp up in 5° increments. Long story short, at 220°C I was able to carefully push many cm of filament without the apparent plugging. At the lower temps, the extrusion would often come out curlycued, like there was some obstruction in the tip. I did a lot of permutations and even removed the heat break and used PTFE down to the nozzle. It really seemed like the jam was occurring down in the nozzle itself.

I then did some experiments with positioning the tip further away from the heated block - attached by only a couple of threads. Now, it was nearly impossible to push any filament until I got to 215°C. This might not be unexpected since there is much less metal-to-metal contact for the brass nozzle to heat transfer properly. That leads me to wonder if perhaps the nozzle is fluctuating in temp and when it cools, the PLA freezes and backs up and plugs. Almost anything can make the nozzle cool - the movement as it prints, air from a part cooling fan (or cold end fan), rapid transfer of filament (at high extrude speeds), retract, etc.

I'm not sure what to make of this yet. Options like insulating the tip with Kapton tape or a ceramic paste come to mind. Meanwhile, I just successfully printed a part at 220°C (for PLA!) that I was completely unable to print yesterday in 6 attempts using exactly the same hot end with no additional cleaning or polishing. I am going to attempt a print at 210, 200, ... and see what happens. This is very perplexing as I've printed this exact spool of PLA with this exact same hot end at 185°C on much more complex parts (lots of short hops and retracts) with no problems multiple times. Just odd!

[mhackney]

So, here's what I'm talking about. I printed three parts, 1 at a time, that all came out perfectly. Yesterday I attempted to print this part 6 times and got filament starving on all 6 parts. The ONLY think I did was chill the hot end so I could remove the filament intact as per the post above. I then reassembled with no additional polishing or anything else. BUT I printed all 3 of these parts at 220°C (they are PLA) and the 3 terminated parts yesterday were at my usual 195°C for this filament.

Here's a closeup where you can see the yucky filament starting.

I've printed several other parts tonight at 220° and all have come out good.

[Polygonhell]

My second one I could sometimes print at 195, with random jams, at 230 I could, print almost reliably, but as the print changed speed because of minimum layer time it would still jam, though only temporarily.

I figure it can't be anywhere inside the heater block or nozzle, because the filament should be completely molten at that point. I think raising the temperature really only extends the length of the transition zone.
What I can't see is what's so different about the one that doesn't have these issues.

[mhackney]

I've made similar observations. I'm leaning towards my nozzle "freezing" theory. Tomorrow I am going to insulate the nozzle with Kapton and try lowering the temperature. My hand extrusion tests really pointed me in this direction and seeing the curlycue extrusions. But it is a theory!

[McSlappy]

Well you've been far more thorough and methodical than I have been - but all of this replicates exactly what I've gone through with mine. Even down to being able to print some parts fine, then the next day the same parts wont.

I've got to install my new e3d and do some tests.

[mhackney]

I am very methodical. I will get to a state where I can print perfect parts - 100s of them - for weeks and then something happens and I get filament starving. Each time I made a change - cool the RAMBo drivers, debur the cobbed gear, polish the heat break, chafer the hole in the heat break, etc things would get back to perfect and I'd happily print for several weeks with no issues. This last go around of polishing the breaks made me the most hopeful since it is easy to imagine the correlation. But here I was again with more filament starving. And, with the tests and observations I made yesterday, there really did not seem to be a problem with pushing filament down the break, it really seemed (and observed) like there was some sort of blockage right at the nozzle that was plugging things up. When milton filaments squirts out of an office and curlycues, that is a good indication that there is a lot of back pressure and/or some sort of asymmetry in the nozzle. These could both be caused by a plug. The thing I don't quite understand - and I've tested this to verify - is that when this happens, if you stop, let things cool and then re-heat and attempt to extrude, you would think the system would reach equilibrium again and allow the material to pass through the now hot nozzle. I have not observed this though, what happens is I still seem to have a plug of some sort. That's what led to removing the hot end and freezing it yesterday. I wanted to see if I could remove the now frozen in place filament to see if I could find any issues. The entire previously-molten section of filament from the hot end easily slid out and even with close microscope scrutiny did not show any evidence of particulate plugging, no charring, or anything else that would lead to a plug. Putting things back together resulted in another plug (at 195°C) almost immediately. I was able to freeze and remove that one too. Then putting things back together and ramping the temp up to 220°C things flowed freely and I've since printed a dozen perfect parts without a hiccup. Something very odd is going on. The geometry of these hot ends is not that radically different from other hot ends. The one thing that is a little different is the process used to tighten the nozzle requires there to be a small gap between the nozzle and the hot block in order to provide clearance for tightening. That results in less efficient thermal transfer to the brass nozzle since almost all of the heat has to be transferred through the threads. Other nozzle designs are either machined into the hot end (J-Head) or tighten up against the hot end (SeeMeCNC and others) which would provide more heat flow. Again, this is just my latest theory in a long list of theories that seemed reasonable, seemed to result in positive change and then, in the end, starving still occurred.

[mhackney]

McSlappy, I forget, did you send your old one back to E3D-Systems? It would be good for them to have a known problematic one to look at.

[Tinyhead]

Just in the last few days I've got my E3D up and going running PLA and I'm getting starvation constantly. About the only thing I can do to keep it going is ramp up the temperature and speed. I'm going at nearly 60mm/sec (almost twice what I usually run) and if I slow it down it seems to starve. If it is choking, I will manually release the filament , pull it up a touch and shove it back down and it will typically start extruding again. I thought there was debris in the filament or something, but after reading your guys' posts, I'm thinking that's not the case and I see I'm not the only one with issues.

[Polygonhell]

That's petty much exactly what mine did.
It feels like a classic PLA jam to me when feeding by hand, the difference being if I retract at least 30mm then the jam pretty much always clears. Similarly if I bump the temperature it clears, my second one would basically clear itself after a few seconds, you push on the filament it would feel,jammed, you'd let off the pressure, and get this sudden rush of filament.
Printing at 230 with the second one you could literally see the Hotend jam and clear during the print, the filament would starve, the you'd get all the jammed filament in a blob.

I find it hard to believe that the filament in the nozzle isn't liquid, and therefore should not be able to jam in the conventional sense. And above the heatsink it's solid.
That basically leaves something happening in the transition zone, but other than surface finish, and possibly exposed length, I'm at something of a loss as to why some work and others don't.

[mhackney]

The experiment that was most telling to me was when I removed the heat break altogether and lined the hit end with some ptfe so the PLA wouldn't muck up the threads. I held the hot end in vice grips and pushed the filament thorough by hand. I would get several cm of good smooth flow with little pressure and then the pressure would start to increase until I couldn't push filament anymore - and this was at a typical flow (I tried to push the filament at the same rate it is when I print.Backing off did not clear the jam. Increasing the temp did. This really baffles me.

Just to check my sanity I ran this test again this morning with the other 3 Kraken hotends and got nearly the same results. 2 of these had not been polished or deburred, one had like the one I used yesterday. The filament would push easily for a few cm and then start to take more pressure and then stop. I tried backing off when I started to feel the pressure and that had no effect, the pressure required to push the filament kept rising. Just very strange.

Is the .4mm nozzle bore longer than other nozzles? I haven't checked any of my others but the E3D nozzle is a hair over 2mm of .4mm bore.

[Polygonhell]

What you describe sounds exactly like a conventional PLA jam. The rubbery portion of the plastic expands sufficiently to grip the outside of the Hotend, and as you continue to apply pressure you just keep expanding the blockage until it becomes impossible to push filament.
Increasing the temperature will often clear these jams, because it moves the transition point up the Hotend and the blockage becomes liquid.

Is the 0.4mm orifice, really 2mm long, that's REALLY long for a Hotend, and could certainly exacerbate any issue. The jhead is closer to 0.5mm, maybe I should measure the orifice lengths on the working and none working ones I have here! maybe that's the manufacturing variation.

[JohnStack]

I'm wondering if the Prometheus project on Indigogo might be the solution for PLA. So many people are using the older style Mark Ivs or similar and printing for a long long time - of course, they don't handle PLA well.

https://www.indiegogo.com/projects/prometheus-hot-end

Maybe it's the whole specialized head thing that we have to live with. Removable effectors here we come?

[mhackney]

Polygonhell, agreed, it sounds like conventional PLA jam but it is actually very different. You know that I have lots of experience with multiple hot ends and PLA jams. This is different, it is not a transition point issue. I am very confident in that bnased on experiments yesterday and today designed to remove that possibility.

Yes, the office really is 2mm long. You can see the little nipple in one of my photos yesterday. I thought it sounded really long too. I have some extra nozzles so I am going to bore one out a little deeper and try it. Also, if you look at the plug, the angle leading into the office is also very shallow. I haven't profiled other nozzles but I think I know how to do it and plan to this weekend. Basically, push a pice of PLA into the nozzle and heat the nozzle with my little butane torch while pushing the PLA until it just extrudes. Allow to cool, trim the excess protruding PLA. Then, put in freezer for a few hours. The PLA should pop right out like I did yesterday. I have nozzles from 6 different hot ends to check. All 5 of my .4mm E3D nozzles are close to 2mm long bore but there is some variation from 1.5 to 3. This may be "it" and worth checking out.

If this is the problem it is an easy fix and should result in a very reliable extruder. In any case, I would think you would want a short .4mm bore for best performance (lower friction) anyway.

[Polygonhell]

The trade off with bore length is extrusion pressure required vs die swell, and to a lesser extent how well it responds to retraction. It looks like the jhead nozzle length is longer than I thought, here's the post where reifsneider compares then from various hotends

http://jheadnozzle.blogspot.com/2012/05 ... ments.html

The Buda has extremely low extrusion pressure requirements, print really fast, but they are a pig to tune for retraction, though that's as much an issue with the larger melt zone.

It looks like the orifice length on the jhead was reduced to 0.5mm in 2012 to reduce extrusion pressure.

[mhackney]

Thanks for the length Polygonhell. I remember reading that page but forgot about it. Some good information. That's interesting about the J-Head reducing to .5mm. The method I'm using - if you can get the PLA plug out - not only allows accurate measuring of the orifice length but also shows the profile of the melt zone in the tip. I've been cleaning my garage and shop and moving piles of stuff from here to there and then moving smaller piles from there to here and, finally, moving these smaller piles into drawers. Thanks Carl! I'll start to make some plugs later tonight of my nozzle collection. I think this is an interesting line of exploration and the E3D tip certainly is at the long end of the scale.

Cheers,
Michael

[Eaglezsoar]

When one of you guys get a chance could you take a look at these prometheus drawings and give your opinion on how well these would work on PLA.
They show the bore length and other critical specs. https://www.flickr.com/photos/87676108@ ... 930743154/

[mhackney]

I'm no expert but I like the bore on this hot end. 60° cone leading up to a 1mm long nozzle bore. Don't know if 1mm is too long or not but it is in the ballpark of other hot ends. The E3D hot end has a very flat cone, it must be like hitting a brick wall when the melt comes down the hot end! Not sure how the cooling is going to work with SS up the entire bore of the aluminum heat exchanger.

[Eaglezsoar]

I'm no expert but I like the bore on this hot end. 60° cone leading up to a 1mm long nozzle bore. Don't know if 1mm is too long or not but it is in the ballpark of other hot ends. The E3D hot end has a very flat cone, it must be like hitting a brick wall when the melt comes down the hot end! Not sure how the cooling is going to work with SS up the entire bore of the aluminum heat exchanger.

Thanks for your opinion. It is becoming quite difficult to know which of the all metal hotends will reliably do PLA. You say you are not an expert but you are the closet thing to being one that we have on the forum. You and Polygonhell are experts on such matters as far as I can tell and still after all you have done the PLA printing problems rears it ugly head.

[mhackney]

Well, I could print PLA all year with the SeeMeCNC or J-head hot ends. They work fine. But the quality I was getting with the E3D and Kraken is noticeably better. The small melt zone and sharp transition to cold (especially on the Kraken) result in visibly better prints, less stringing, blobbing, etc. Everyone at the NYC show last week commented on the quality of finish on my gold PLA fly fishing reel. It really is nice. Now if I had a smoothieboard!

I think we are on to something with the long bore on the nozzle. It could explain a lot of the problems we are seeing. Today I cleaned and organized, tomorrow I'm going to drill an E3D nozzle to shorten the bore AND increase the cone angle to 60°.

[Broose]

A few thoughts...

mhackney- I like your idea of freezing the nozzle to remove the plug. Since the coefficient of thermal expansion of PLA is about 4 times greater than brass, when you lower the temp it should shrink more than the brass nozzle does.

I'm very curious to see if decreasing the nozzle hole length on the E3d will reduce your clogging with PLA. I suppose the tradeoff may be more oozing or retraction issues, which I've been very happy with on the E3d.

Too bad E3d doesn't show the internal dimensions of the nozzle in their drawing: http://files.e3d-online.com/Drawings/E3 ... Nozzle.png

I like the idea of a contiguous heatbreak/nozzle in the Prometheus to eliminate clogging, but I imagine stainless steel does not make the best material for the nozzle due to its low thermal conductivity. I wonder how it performs at higher print speeds

[mhackney]

Made some measurements of the bore length of some nozzles. First, here's the setup. I make sure the bore is clean then hold it in needle nose pliers. I have a 3" length of PLA at the ready. I then heat the nozzle with a little butane torch MILDLY. When the too is just hot enough to allow the PLA to flow (takes about 10 seconds or less) I push the PLA in until a little nipple comes out of the end like this:

Let this cool, do not touch the protruding nipple. When it is cool enough to touch, cut the protruding filament off with a single edge razor blade and place in the freezer for several hours. Put a pair of needle nose pliers in the freezer too. When ready, use the frozen pliers to pick up the nozzle - that way it does not warm up too fast. Grab the PLA and pull gently. Don't force it. It takes about 10-15 seconds for the brass to warm up from air temp and expand enough to let the PLA slide out. It will and will look like this:

As you can see, this J-Head is dead-nuts on at .5mm bore length. Interestingly, the cone is also very flat like the E3D.

I also measured three other nozzles and they were all within .5 to 1mm in length. Only the E3D has been anywhere close to 2mm.

Today I am going to chuck one of the E3D nozzles in my lathe and rebore the 2mm lead up to the nozzle bore. I'm shooting for .8mm. I don't have any data that substantiates this but I'm thinking a length to diameter ratio of 2 is a reasonable ballpark. I've googled to try to find any information and there is very little discussion about bore length in the RepRap world, everyone is focused on bore diameter. But, at these small diameters, there is a significant friction component and the longer the bore, the more friction. Stay tuned!

[mhackney]

Too early to claim victory, but...

First, I setup and printed a part that has been causing problems the last few days (a mini Kossel effector). I have been able to print it at 220°C but any lower and I get filament starving. One comment on this - I like the KISSlicer curved infill. It really puts the printer through its paces though and seems to exasperate this filament starving problem. Printed with straight infill, I can print this part at 205°C. I verified that the problem is occurring today. I then immediately took the nozzle off the machine and pulled the execs PLA out.

Down in my studio/shop/man cave I used a 2mm drill and twirled it between my fingers to remove any excess PLA - some did in fact come out. I then measured the length of the drill and then the length of the drill inserted into the nozzle. The difference, at least to a reasonable approximation, should be the length of the nozzle bore. I got 1.96mm (in reality, since the bit is pointed, it inserts a little into the bore of the nozzle so the measured length of the bore would be expected to be shorter than actual). I then chucked the drill in a pin vise (a handheld drill bit holder) and used it to gently lengthen the hole. I wanted to try this for several reasons - 1) it wouldn't require a lathe to fix the nozzle and 2) it gives very precise control. With just a gentle twirling of the pin vise between my fingers and a little pressure, the brass was easily bored. I frequently removed the bit, cleaned off the brass bits, and remeasured the bit and nozzle as above. I was taking about .1 to .2mm off at a time, so very precise control. At some point I decided to shoot for .5mm rather than the .8mm I mentioned in the last post. When I was close to the mark, I decreased pressure on the pin vise, twirled faster but for a shorter time. This was to take a fine cut and leave a better surface finish. I was able to get right at .5mm This entire process took a little less than 10 minutes and resulted in what I believe to be a reasonably accurate bore length.

Then, I did my frozen PLA test to capture a plug to see how long the bore actually is. I've gotten the technique down pretty well now so with a minimum of heat to melt the PLA and then dropping in a glass of ice water I can almost immediately remove the plug without freezing for 2 hours. Success, the nipple measured as close to .5mm as I could measure. At least it was significantly less than the 2mm stock bore!

Now, armed with a modified nozzle, I cleaned out the inside with a stream of isopropyl alcohol from a syringe with a 26 gauge needle to clean it out. Very little material washed out. I then mounted in my Kraken, rezeroed my Z axis, loaded my gcode, lowered the print temp to 185° where I used to print this particular PLA with a J-Head or the SeeMeCNC hot end (by editing the gcode, thus ensuring that everything else was identical to earlier prints), said a Hail Mary and pressed Print. Actually, before I did that I did extrude about 100mm of filament into the air to prime the nozzle and see how it would behave at 185°C. It seemed to flow fine and in fact, almost seemed to start faster when I hit the extrude button.

As I've been writing this (20 minutes or so) the part has been printing. Here are my observations:

• I'm successfully printing at 185°C (so far).
• The first layer went down incredibly smooth. One of the nicest first layers I've seen in a long while.
• I'm about 1/2 way done with the part. Earlier, I would get to about the 4th layer and have filament freeze issues. I'm at layer 22 of 41 and the part looks very nice. Very smooth layers, no indication of starving, no stringing, Absolutely no evidence of filament starving.

It is way too early to claim victory but I am very optimistic about this. It makes sense in retrospect and it is an area most of us completely overlook - the bore length. This is not to say that improvements like polishing the heat break bore, etc are not also contributors to filament freezing but I've systematically eliminated all of those to get to this point (no pun intended).

[Polygonhell]

Very interesting.

I think the real litmus test would be to establish that there is some significant variation in the E3d nozzles, it would give us a reason why some of them are very much worse than others.
But in retrospect a particularly high extrusion pressure would account for what I observe, KissSlicer tool paths are particularly bad.

[Eaglezsoar]

I am interested in the long term results of your experiment. Will it fix the PLA problems permanently, only time will tell.
It is fascinating as we wait in anticipation of your experiment, it feels like we are part of an audience watching Thomas Edison invent the light bulb.
Your experiment is especially fascinating because it has ramifications for all that are using the E3D.