This post finally documents ideas and hints I had during the design of two hot ends I made last year.
At the time I already had ordered an E3D hot end, but the lead time was so big that I decided to give it try myself. Actually, since I attached it with this tiny E3D support on my Ultimaker, I almost never swapped my hot ends anymore because it is just so reliable... The only drawback is that the efficient heatsink could be made almost half the size for more Z space, less vibrations and less weight (hence my smaller, low cost --and low quality-- copycat by the way, see below). And it does not mean I stop thinking ;)
Update: the E3Dv6 just addresses these wishes. It is smaller and lighter. Sanjay does a really outstanding job, not only b/c he knows his stuff but also because he documents it thoroughly and beautifully. Let us hope Makerbot will not try to patent his designs -- And no, I got no free E3D to give such an opinion! ;-)
Back to the hot end: the cold filament enters a short hollow barrel which itself at some point goes through a "heater block" that heats enough to meld the filament. On the opposite side of the barrel is the nozzle. So the cold filament builds up pressure, which in turn pushes the molten plastic out through the nozzle.
Hot ends almost always embed a sensor to allow for a precise control of the temperature. The sensor is either a thermistor or a thermocouple (which works reliably with wider temperature ranges but need more electronics because of their very weak signal).
First of all, the nozzle of course! The output diameter defines how fat the produced plastic thread is, hence details vs. print time. Actually, nozzles can be made relatively easily out of brass with basic tools and a bit of caution, because the very small drills that are used are easily broken (0.2 to 1 mm, usually 0.4 or 0.5 mm).
The nozzle shape has complex impacts on the printing process. I tend to prefer those that are pointy, like that of Ultimakers, because smaller or flatter ones radiate much unwanted heat back to the printed parts when hovering above them.
When making a nozzle, a rule of thumb is to drill the "final" hole 3-4 time as long as is it wide, so that the flow is straight and not curved. Note that the slightest dirt or bump near the end of the nozzle also provokes such curved output (this is a clear signal that a cleanup is needed).
Dome-shaped nuts mostly do not have enough thickness at the end, so I favor regular 6 mm short brass threaded rods. With two shallow saw cuts and a specially made wrench, they can be changed easily (only with the "male" kind of nozzle, because they seal much better than female nozzles).
All metal hot end plugs and jams
The heating block: this is a simple block of aluminum just large enough for three holes: one horizontal hole is for the heating cartridge (a power resistor), which usually goes throughout the block. Another hole is for the temperature sensor. In my case I use glass-fiber insulated K-sensors that I bought on ebay after a few broken conductors in the stock ones, but thermistors will probably be my next choice for added robustness.
Finally the last 6 mm vertical through hole is for the barrel (on the top) and the nozzle (on the bottom). Both meet in the middle of the block, unless a female nozzle is used in which case the barrel goes throughout the heating block and it protrudes below by the few required millimeters.
To avoid leaks between the barrel and the nozzle, people often use a small piece of teflon tape, but it is not safe with high temperatures. Actually when the barrel and nozzles are well matched and tightened enough, leaks tend to seal themselves quickly, probably thanks to carbonized overheated "juice" from the filaments themselves (ABS at high temperatures is sometimes used to seal a hotend on its first use).
I also seldom used one-piece barrels that end shaped as a nozzle: no way a leak can occure but they are harder to make, obviously much harder to clean when dirt clogged the nozzle, and they cannot be changed as easily as a stand alone nozzle (better change the whole hot end then, which is faster in fact).
Active or passive cooling of the hot end?
Hot ends always feature a thermal insulation, with or without a heatsink. In both cases this is to reduce and separate the heated zone from the input and fixation points, that are often 3D printed and would melt themselves otherwise!
Personally, I do not like passively cooled hot ends, probably also because they usually rely on PEEK components to insulate the hot from the cold parts. PEEK is a hard and special plastic wear out in the long term because it is kept close to the maximum temperature they can stand, and which is exceeded for some plastic filaments (review here) or when the temperature control goes wrong which happens...
Unless itself actively cooled, the PEEK element is only cooled down by the movement of the head. Else it may become hot enough to defeat its own purpose: passively cooled hot ends tend to have a longer "thermal break", which sometimes extends so much that large parts of the barrel get filled with molten filament. This, in turns, is very annoying because the flow is impeded or stopped and a complete cleanup is then required.
Reciprocally "actively cooled" hot ends require a small fan (25 or 30 mm) that blows on a metal hot sink, which in turns drags the heat off the barrel just above the heater block. The bad thing is that active cooling requires extra power lines, and they add to the hot end weight and the overall noise (especially as smaller fans are quite noisier).
The best thing with actively cooled hot ends is that they allow for a very short melting zone. Whatever the drawbacks, I will probably never use passive PEEK-based hot ends again, now that I enjoyed the very good reliability. Still, it is a bit unclear to me whether a shorter melting zone is always better, because of the increased risk of plugs when retraction occur and if the barrel is not perfectly smooth. It sounds to me like the "better grip is always better" for the filament driver, which proved to be false in my opinion.
Thermal insulation of the heating block?
The idea is to insulate the heating block itself from the rest of the world, and let only the tip of the nozzle protrude below. This also improves the regulation because of the main fan which cools down the part being printed shall not impact the heater block itself (as it tries on its side to maintain the temperature).
Some people wrap the heater block around ceramic ribbons. I usually use wrinkled kapton layers, trying to embed some air between the layers for better insulation. This works well, but the best so far for me was to enclose the whole block in a hard stainless steel sheet that I perforated for the cables, barrel and nozzle.
With a small piece of kapton for the side, you can see that it does a really good job at keeping the hot end hot, and only where needed (I went up to 260 and still could keep my thumb on the heater housing). Obviously my finger is NOT on the nozzle itself that protrudes slightly from the insulation chamber below.
Finally it is easy to check that you did it right by printing an object that requires heavy retraction, such as those voronoi objects with fractal holes everywhere. By the way it also stresses a lot the filament driving system and the hobbed bolt (both home made below).
 |
| My extremely small "cheesy hot end" (all-metal and actively-cooled), here with an Ultimaker printed support. This hot end was very easy to make (only used basic tools) and also because the barrel is quite short. |
Update: the E3Dv6 just addresses these wishes. It is smaller and lighter. Sanjay does a really outstanding job, not only b/c he knows his stuff but also because he documents it thoroughly and beautifully. Let us hope Makerbot will not try to patent his designs -- And no, I got no free E3D to give such an opinion! ;-)
Oops, for newcomers: first of all, what is a hot end?
It is the critical part of the printer in which the feeder pushes the filament, either directly ("direct drive") or through a slippery tube (a "bowden tube", usually made of low friction teflon of PTFE). Direct drives are close to the hot ends, so they make them bulkier and heavier and the movement is slower unless they are moved by bigger motors and support rods. Reciprocally, bowden tubes move the driving motors father to the frame of the printer, for a much lighter and smaller moving hot end. So they allow faster speeds and arguably better quality, but they add annoying elasticity in between.Back to the hot end: the cold filament enters a short hollow barrel which itself at some point goes through a "heater block" that heats enough to meld the filament. On the opposite side of the barrel is the nozzle. So the cold filament builds up pressure, which in turn pushes the molten plastic out through the nozzle.
Hot ends almost always embed a sensor to allow for a precise control of the temperature. The sensor is either a thermistor or a thermocouple (which works reliably with wider temperature ranges but need more electronics because of their very weak signal).
 |
| Some hot ends: Ultimaker (V2), the shiny E3D (v5), and a home made hot end of mine (on the right) |
OK now, what makes a hot end? And some hints to DIY!
Printer nozzle
 |
| A 1-mm nozzle made out of a 6 mm brass threaded rod. |
When making a nozzle, a rule of thumb is to drill the "final" hole 3-4 time as long as is it wide, so that the flow is straight and not curved. Note that the slightest dirt or bump near the end of the nozzle also provokes such curved output (this is a clear signal that a cleanup is needed).
Dome-shaped nuts mostly do not have enough thickness at the end, so I favor regular 6 mm short brass threaded rods. With two shallow saw cuts and a specially made wrench, they can be changed easily (only with the "male" kind of nozzle, because they seal much better than female nozzles).
The brass or stainless steel barrel
The barrel is the essential metallic tube in which the filament enters cold, and to which the nozzle is linked, through which the molten filament exits. It must be sealed. It must be metallic to cope with the temperature, and stainless steel is probably the best choice apart from the difficulty to shape it: it is both very hard, smooth and does not conduct heat well, which in turns helps keeping the melting zone only where we want it.
Now, without a lathe, the barrel is the hardest part to make because it is usually require to drill a 3.2 or 3.5 mm hole all along the center of a 6 mm threaded rod, without openings on the thin sides that are left.
 |
| Yet another failed barrel ! But it is somehow OK for me if I succeed at least once, which I did ;) |
Using a good drill press can give you some results but it is hard to drill perfectly straight after 30 or 50 mm of length even with hardened stainless steel brand new drills (do not use old ones as they will not drill straight). Also I like to drill with riskier 3.5 mm inner diameters more than, say, 3.2 mm, because it gives less friction with the 3 mm filaments I use.
Beside the difficulty to make them, I did not find any counter effects to have a "wide" inner diameter given that small inner diameters blocked my Nylon trimmer line and wood filaments anyway a lot. Even though I fail more often than I succeed, this is all right to produce a few nice barrels with basic tools, and it does not necessarily mean you cannot achieve a high quality item, as the more you do, the bigger the chance you will make a nice one (see the idea behind?)!
Beside the difficulty to make them, I did not find any counter effects to have a "wide" inner diameter given that small inner diameters blocked my Nylon trimmer line and wood filaments anyway a lot. Even though I fail more often than I succeed, this is all right to produce a few nice barrels with basic tools, and it does not necessarily mean you cannot achieve a high quality item, as the more you do, the bigger the chance you will make a nice one (see the idea behind?)!
 |
| A few experimental barrels and hot ends. I made the small wrench to help me change the threaded rod brass nozzles. Dome nuts are not worth in my opinion because they are too shallow in the end. Brass is fine for a barrel, but stainless steel is better as an insulator. I found no real benefit to remove the threads just above the heater block for a better thermal break (like on a machined E3D barrel), especially as I broke a few of them trying ;) |
Now, I soon realized that some of my barrels did cause a lot of jams, especially during retraction (actually it happened to me more than once with the E3D also). The only solution I found at the time was to reduce retraction to low values like 1-2 millimeters, which really is too small for good printing quality when the printer has a bowden tube. I read that some recommend dipping a piece of filament in vegetable oil to "treat" the metal surface initially and once for all (I never used any lubricant on my feeding system yet, so I cannot tell whether it really works). Some others (e.g. Rob at well-named 3dprinterhell) even do not recommend all-metal hot ends with PLA (which sticks to steel more than ABS does).
As for me I got a lot of failed prints first, because the barrel would become obstructed one way or another. I knew my settings were all rights, and that my extruder was not the culprit. And strangely, it happened sometime with two barrels that had same size and diameter!
As for me I got a lot of failed prints first, because the barrel would become obstructed one way or another. I knew my settings were all rights, and that my extruder was not the culprit. And strangely, it happened sometime with two barrels that had same size and diameter!
The thing is that even though you manage to drill correctly the barrel, it also must be very smooth or it will impede filament flow and provoke a lot of grinding or jams. The slightest coil or metal "hair" left in has a major impact, especially with retraction, i.e. when the driver goes backward to pull the filament a bit and reduce pressure while the head is moving without printing. Retraction is a must to prevent oozing and leaving bits and strings of molten plastic all around.
All metal hot end plugs and jams
But when the filament is retracted, molten or half molten material moves up in the barrel and may reach a zone cold enough to "freezes" instantly, especially with all-metal actively cooled hot ends and very short thermal transition zones. The filament will then bond to the tiniest scratch or dirt left in the barrel and sometimes it will be impossible to resume, because of the beautiful perfect cold plug it makes.
Hint : never shut down your printer as soon as you are done printing if you have an actively cooled hot end! Otherwise and because of its thermal inertia, the heater will silently heat the head, sometimes up to the point where filament is molten in a much larger zone than the usual short expected thermal transition. This may results on the same kind of "long plug" in the head, which will prevent movement on your next print.
Hint : never shut down your printer as soon as you are done printing if you have an actively cooled hot end! Otherwise and because of its thermal inertia, the heater will silently heat the head, sometimes up to the point where filament is molten in a much larger zone than the usual short expected thermal transition. This may results on the same kind of "long plug" in the head, which will prevent movement on your next print.
When you get such plugs/jams, the only way out is to remove the hot end fan and let the whole hot end heat again until it is possible to push the filament manually again (but your print is ruined). Now your barrel may also have some sticky dirt left, which will again increase the likeliness of subsequent jams (it drove me crazy at some time, even with the early E3D).
So you need to take the whole hot end apart and clean and smooth the barrel thoroughly. It is easy to purge the head (see this old post for example). But it is much harder to make it really smooth and remove all oxidized dirt.
 |
| One cheap and easy way to polish the interior of a hot end barrel: use sand and a concrete drill with a smaller diameter than the inner tube. |
Usually, smoothing a barrel very finely requires costly equipment (namely specific reamers). Now, I did it with very basic tools: use a concrete drill bit that is 0.5 to 1 mm smaller than the inner diameter and which protrudes from the other side of the barrel. Plunge the tip of the barrel and the drill bit into a pot of sand (the finer the better, but it will get fine anyway when it gets ground). Using high speed rotation, the grains of sand are fragmented and dragged up and into the barrel, polishing nicely the inside. It obviously takes longer with stainless steel than with brass, but it did the job pretty well for me (a video is here by the way).
Update: the super active user "A2" on the reprap forum pinpointed here a very smart way to do it more professionally. I quote him here, all about the so-called Needle eye lap polishing technique that still can be home made, using a brass rod which is undersized of the bore.
Update: the super active user "A2" on the reprap forum pinpointed here a very smart way to do it more professionally. I quote him here, all about the so-called Needle eye lap polishing technique that still can be home made, using a brass rod which is undersized of the bore.
Hammer one end of the brass rod flat, until it is a tight fit to the bore.
Using a sharp chisel, split the flat section in the middle, don't cut through the end.
The split down the middle performs the function of a spring, and the grove helps retain the abrasive.
Use clover lapping compound, beginning with a coarse medium, and working down to a fine medium.
Don't dwell in on spot for any amount of time, or you will create a barrel shaped profile.
Heating blocks
 |
| A heating block with a stainless steel housing, a conic nozzle and barrel. The last hole, for the temperature sensor, is on the other side here. See the chapter below for why a housing is extremely efficient. |
To avoid leaks between the barrel and the nozzle, people often use a small piece of teflon tape, but it is not safe with high temperatures. Actually when the barrel and nozzles are well matched and tightened enough, leaks tend to seal themselves quickly, probably thanks to carbonized overheated "juice" from the filaments themselves (ABS at high temperatures is sometimes used to seal a hotend on its first use).
I also seldom used one-piece barrels that end shaped as a nozzle: no way a leak can occure but they are harder to make, obviously much harder to clean when dirt clogged the nozzle, and they cannot be changed as easily as a stand alone nozzle (better change the whole hot end then, which is faster in fact).
Active or passive cooling of the hot end?
Hot ends always feature a thermal insulation, with or without a heatsink. In both cases this is to reduce and separate the heated zone from the input and fixation points, that are often 3D printed and would melt themselves otherwise!
 |
| Molten PEEK: I tend to avoid this insulation material for many reasons. |
 |
| Improved ultimaker hot end, with an additional passive heat sink that starts just below the PEEK insulation, and wrinkled kapton tape around the heating block. |
Reciprocally "actively cooled" hot ends require a small fan (25 or 30 mm) that blows on a metal hot sink, which in turns drags the heat off the barrel just above the heater block. The bad thing is that active cooling requires extra power lines, and they add to the hot end weight and the overall noise (especially as smaller fans are quite noisier).
The best thing with actively cooled hot ends is that they allow for a very short melting zone. Whatever the drawbacks, I will probably never use passive PEEK-based hot ends again, now that I enjoyed the very good reliability. Still, it is a bit unclear to me whether a shorter melting zone is always better, because of the increased risk of plugs when retraction occur and if the barrel is not perfectly smooth. It sounds to me like the "better grip is always better" for the filament driver, which proved to be false in my opinion.
Thermal insulation of the heating block?
The idea is to insulate the heating block itself from the rest of the world, and let only the tip of the nozzle protrude below. This also improves the regulation because of the main fan which cools down the part being printed shall not impact the heater block itself (as it tries on its side to maintain the temperature).
Some people wrap the heater block around ceramic ribbons. I usually use wrinkled kapton layers, trying to embed some air between the layers for better insulation. This works well, but the best so far for me was to enclose the whole block in a hard stainless steel sheet that I perforated for the cables, barrel and nozzle.
 |
| Now this is what I call a good insulation of the heater block (home made all metal hot end with stainless steel housing). Less upward heat convection makes a shorter thermal break, and radiated heat towards the print is also reduced. |
Finally it is easy to check that you did it right by printing an object that requires heavy retraction, such as those voronoi objects with fractal holes everywhere. By the way it also stresses a lot the filament driving system and the hobbed bolt (both home made below).
A video of the printing process pictured below,
the many retractions are a good test to check the
filament flow reliability.
 |
| Probably the best kind of object to test retraction (design by richrap) Still, my print is quite dirty, but at least I got no single plug, and my home made filament driver is very reliable too :) |