Why I would reject E3D-V6 as a product prototype

Discussion in 'E3D-v6 and Lite6' started by RobBen, Jun 3, 2015.

  1. RobBen

    RobBen Member

    Joined:
    Jun 3, 2015
    Messages:
    3
    Likes Received:
    0
    This is a short introduction. I am a (professional) Industrial Designer. I know, it is kind of irrelevant, but it is important for my further explanations as I am well trained at analyzing problems and offering thoughts.

    For all Issues described further on, I will use a machine everyone knows as metaphorical comparison, just to emphasize the story.

    The coffee-machine.

    But now for the facts:

    I am running an Ultimaker Original with very few modifications, and recently chose to replace the original printhead with the acclaimed E3D-V6 Hotend that promised "no more clogs - more materials" which sounded awesome, as I am also selling prints for prototypes and presentation models and I could bring my UMO back into this business.

    In the original printhead there was a nylon-glassfibre-part behind the heatbreak that gets worn out every other month, so I chose an all metal hotend instead, with the promise of better reliability and less maintenance. So I placed my order for the bowden-set with fan, added a couple of nozzles and extras that quickly added up to about 150€.

    So i printed a couple of different mounts for the E3D before changing, to tryout which is good and to be on the safe side for printing happily away after implementation. I looked at the design when it arrived, and it was very reassuring, its heavy, its metal, its thoroughly machined. Looks good, feels good. Everything a consumer wants.

    After putting everything together on my saturday morning, I was eager to start printing. But, as always, calibration, readjusting the end stops, drilling holes for new screws etc. got in the way. Finally i could start a first small job - and oh joy, only 20 minutes in, a clogged hotend. First i could not figure out where, how and why. I ran the print with 210°C on PLA instead of the usual 195, as I figured the melting distance of the filament to be smaller, thus less contact with hot metal and higher temps for quicker melting.

    After disassembly I realized that the filament blobbed on top of the heat break within the heatsink - right between the bowden tube and the entry into the heat sink. How could that happen? How did the heat get there, when there was so much heatsink in between the heaterblock and the bowden? I restarted, and this time I got a clog somewhere in the heat break, which is a lot harder to get out. Great!

    So I started analyzing the whole thing with the question in mind: "Where is the heat, how does it move?", and realized some things.



    Problems and Solutions (a: for the user; b: for the manufacturer)


    1. The heat break screws into the heat sink very, very loosely. If you unscrew it a tiny bit, it wobbles.

    This is either a sign for lousy machining, or intentional gaps. This leads to a very bad heat transmission between the heat break and the heat sink. What you would want is a direct, ideally one part solution there. It is a bad idea to thermically disconnect the heat break from the heat sink, as the heat rather travels up the break to the top within the metal, than to cross an insulating gap of air into another metal. Furthermore you create a chimney effect within the heat sink: The heated air on the bottom moves up along the threading and ends up in the little chamber between bowden and heat sink.

    Solution 1 a): Use a generous amount of thermic paste - usually used to connect processors to heat sinks - in the threading. It is not ideal, but helps. You want to avoid enclosed air or any kind of air flow within the threading. Remove surplus paste on the bottom of the heat sink and the top of heat break within the heat sink by pushing some filament through until it comes out clean.

    Solution 1 b): Machine the heat break and the heat sink either in one part or as a push in conus where the heat break is low heat conductive while the sink is high conductive. That gives you a large surface for heat transmission between the break and the sink.


    2. The heat sink has a thicker metal centre on the bottom than on the top.

    This means that more heat is stored within the material than transported to the surface, where the air flow takes it away from the critical parts. What you would want is a large surface without a lot of mass. Mass stores heat. Guess why you have a heater block, and not a wire. The thicker bottom end just means that there is more heat inside to soften the filament way before its time in the heater block / nozzle.

    Solution 2 a): Forget it if you do not have a lathe. Lay down and cry out of frustration. Then look for a professional who has one. If you have one, mill out some of the metal between the disks until you have only a little left. Be careful!

    Solution 2 b): Do not create a heat storage on the bottom of the sink. Rather aim to distribute the heat as quickly as possible into the surface of the heat sink so the air flow will have a maximized effect. Look at CPU-coolers. The best ones have very thin copper lamella to have a high surface to airflow ratio.


    3. The heat break does not have a very smooth (ideally polished) inside surface.

    That is in fact a huge problem. If the filament goes in, it suddenly faces a large drag on the heat break surface. This means that your extruder could fail or grind. Additionaly this means more surface to transmit heat to the filament, if the heat break is warm.

    Solution 3 a): Hard to tell. I heard using oil as lubricant. Some use a drill to widen the opening a little, some use fine sand paper to polish it. None sounds to safe, or easy to use. Oil can mess up a print, drilling can destroy the part. Sanding may deform the inside and leave uneven surfaces.

    Solution 3 b): Get your shit together. Its not so hard to machine a straight hole into an aluminium block. Have proper tools. Have quality control. Steal from other heat break designs! Look at the ultimaker original heat break. It has a large plate before the heat sink and gives off a lot of heat to the lowest lamella of the heat sink. Its just too short to go all the way up into the E3D-heat sink, otherwise i would use it.


    4. The heat break has a long threaded part in the heater block.

    This leads to more than necessary heat transmission into a part that should stay cool right outside the heater block. Here it is an advantage that it is wobbly, as there is not much surface to pick up heat. But it connects to the hot nozzle with a flat surface, so it picks up a lot of heat there. It is unavoidable i guess, but still a problem.

    Solution 4 a): I used a special heat stopping paste from the jewelry equipment shop. It reduces heat transmission up to 800°C, so we should be fine. Apply on the threading of the heat break, use a toothpick or something to rub it into the threading of the heater block close to the surface. Be careful not to get this stuff into the threading of the nozzle. There you want a maximum of heat transmission. I used the thermic paste for CPU cooling as well. It might be good to have a tiny layer of heat stop between nozzle and heat break.

    Solution 4 b): You could make the nozzle threading longer, and the heat break threading shorter. It should still tighten hard enough without risk of breaking. You could use a material to coat the heat breaks threading that has very low heat conductivity and resists heat up to 300°C. Would PEEK work?


    5. The heat break tends to have slightly different diameters on the inside.

    This can only be caused by bad material, bending while machining or a slightly eccentric drill or lathe, maybe a partially blunt drill.
    It leads to filament clogs when it gets warm because of all previously mentioned reasons, as the filament expands, fills the slightly larger diameters, creates more drag than the extruder can handle and then clogs.

    Solution 5 a): See 2 a). Get a lathe, carefully drill it open without destroying it.

    Solution 5 b): See 3 b). Get your shit together, drill straight holes. Have QC.



    Now imagine, you bought a complex coffee-machine for about 1500€, that you have to disassemble every other month just to get coffee out of it. And if you do not put another hour into calibrating it afterwards, all coffee tastes like a trolls armpit. If you calibrate it, it will only be every fifth coffee.

    Finally you find a solution for your trolls armpit problem, for only a tenth of the price, and order it. After installing the solution, you realize that the coffee does not taste like armpit anymore, but comes out as syrup - And half of the time nothing comes out, and you have to dissassemble the whole thing while it is running hot, burn your fingers while doing so, just to get some ground coffee out of the system. You have to do this once a day, if you want two coffees. Not really an improvement.

    After reading up on your strange problem you realize that its not uncommom, but rather everywhere. The solutions offered vary between going at your coffee machine with a power drill or doing strange shamanistic dances around it at the right time just after it started heating up. You wonder why you should pay this amount of money for a machine and an addon that is actually pretty simple, but makes it so complicated. How can it be that such a flawed system is being sold so often?


    To sum up with some of my thoughts about consumer 3D-Printing:

    I am amazed about something. There is no other industry that is able to sell really expensive consumer machines which need so much maintenance and effort just to make them do what they should once in a while. Thats why they are called consumer machines. You want to consume the possibility of manufacturing some stuff, even if its dollar store items. But you are not able to, unless you spend a whole lot of time with your toy.

    Imagine any, really ANY other product category having the same issues as consumer 3D-Printing, would you buy one? Would they even be on the market? Its a miracle. The fact that there is a whole sub-industry developing to compensate for the flaws of the main industry, and everybody accepts it, is sensational.
     
  2. 3Dmaker4U

    3Dmaker4U Member

    Joined:
    Nov 7, 2014
    Messages:
    17
    Likes Received:
    1
    Any, really ANY other product category had similar (well, could not bee really the same :) issues as 3D printing (in general, not consumer only). When? When they were in the same stage of development as 3D printing is now. And while one could argue a 3D printer is just a simple machine, I would completely disagree. It's true that the principles employed are fundamental. However, exploiting them with optimized performance and simplicity against cost it's anyhow but trivial.
    Anyway, we are at the beginning of the learning curve, and any experience is relevant.
    As for the coffee machine, well, I just use the traditional method of boiling water with coffee and sugar, and it works great. But there is no other simpler way to build certain things I need, to replace the "ordeal" of running a 3D printer ;)
     
  3. elmoret

    elmoret Administrator

    Joined:
    Mar 1, 2014
    Messages:
    586
    Likes Received:
    91
    Unfortunately a lot of your "suggestions" are based around misconceptions or misunderstandings. We'll get to those, but first lets address your jamming:

    1.) There should be no space at all for the polymer to "pool" between the end of the PTFE and the heatbreak. Leaving space here means incorrect assembly.

    2.) Jams in the heatbreak generally indicate that you are using excessive retraction. How much retraction are you using?

    As for your "design analysis":

    1.) Threaded parts must be, by definition, a clearance fit. You should not be operating the hotend with the heatbreak partially unscrewed. I have no idea why you would be. As for a "chimney effect" - this is silly, there's simply no air to speak of inside the heatbreak, it is filled with polymer!

    You cannot effectively machine the heatbreak and the heatsink as one piece because they require very difficult thermal properties! The heatbreak must have low thermal conductivity (and is therefore made of stainless steel), while the heatsink must have high thermal conductivity (and is therefore made of aluminum). I am surprised this was not apparent to an "industrial designer".

    2.) The heatsink has a thicker metal center on the bottom than the top also by necessity! The bottom of the heatsink has to be the width of the filament (obviously) plus the width of the PTFE (so a nice transition can be had to the heatbreak) plus the width of the heatbreak, plus the width of the heatbreak threads. This means it will have a larger outer diameter, but also a larger inner diameter, so there is no significant increase in thermal mass as you speak of. Take a look at the CAD drawing, you'll see the heatsink's wall thickness is near constant. Actually it is thinner on the bottom than the top! http://wiki.e3d-online.com/wiki/File:DR ... SINK-B.png

    3.) The heatbreak is absolutely polished internally. If yours is not, then I would recommend contacting E3D for an exchange as yours is defective.

    4.) The portion of the heatbreak that is in the heatblock is 4 threads long. Generally for strength you want the threaded length to be at least as long as the machine screw is wide. That would be 6 threads. They've already compromised down to 4, dropping to 2 or 3 threads of engagement would result in easy stripping in the aluminum block.

    5.) This is not a defect. I am not at liberty to share more (you can try contacting Sanjay at E3D directly), but this is definitely not a defect.
     
  4. RobBen

    RobBen Member

    Joined:
    Jun 3, 2015
    Messages:
    3
    Likes Received:
    0
    Thanks for your replys so far. But ... (theres always a but)

    1) There always will be a space with a bowden. Tiny space, or larger space. The bowden moves, a little, always. The clamp is designed to grip on to the tube as soon as it is retracted, this means there is no way to move the bowdentube in after pulling the clamp out without causing damage to the tube. Furthermore, the heat sink has a slightly sloped surface towards the hole for the filament, which is intentional to get the filament in, but as it is impossible to exactly cut this slope on a bowden tube without very special machines, there will be a gap.

    This should not even matter, as the heat should not reach this point. This area should stay cool enough to keep any filament below its glass transitioning temperature, even wax filaments.

    2) None. No retraction. I took it out of the problem hunt as it transports too much heat up the heat break. I usually use 1,5 to 2 mm.

    In regard to my design analysis.

    1) I know that threaded parts have to have a clearance. But wobbling threads that barely touch? Have you once cut a threading by hand? Mine usually do not wobble, especially not if they are more than 10mm long. I did not say i operate it unscrewed, it was just an observation about the accuracy of the threading. It should not wobble when its screwed in for >95% of the threading. The chimney effect does not take place inside of the heat break where the filament is, but between the heat break and the heat sink. It transports heated air, if only a little bit of it, up through the threading to the tiny gap at the end of the heat break, where it can cause trouble.

    It is also apparent that composing this element in two materials is an advantage if not a necessity, but i feel the advantages are lost by the execution in this case. I was quite frustrated so I drew up a simple file and tested it in a simulation. You are right! If you have a full contact surface between the heat break and the sink, like you have in the technical drawings from E3Ds website, its nearly perfect! The heat moves up the heat break and is instantly dissipated to the sides into the better material. But it does not work if you apply a loose threading with very little transition surface. Suddenly the heat moves up far more than anticipated from the technical drawings and causes all the observed problems.

    2) The heatsink is thicker at the bottom because of the additional mass of the heat break that is screwed in. It sums up to a higher total of storage capacity for heat, in an area where it should be dissipated as quickly as possible. In this case it is not about the wall thickness, but about total, added thickness of all metal parts.

    3) Done. But still: How do I read about this problem in nearly every thread concerning clogs? If this one part of the whole assembly is so crucial, why not put special focus on it? Nevermind. Frustration i guess.

    4) I know this is problematic. Thats also a point I do not see easily fixable, and a lot of people are putting thought into this. This point was more analysis than complaint. It is still an issue.

    5) Interesting insider knowledge on radar? If it is intentional, but causes problems, why do it? Would it be even worse otherwise? Damn.

    I know this is supposed to be one of the best Open Source Hotends on the market - So its even more annoying when it just does not work. And please, regard some of my analysis as suggestions to making it better. I would buy V7. If it costs more, or if you make a deluxe version that has some of the luxury features that would make it perfect, my wallet is open. Still i am a tinkerer, and I feel challenged. Otherwise I would have sent it back already. I am thinking of throwing some steel and copper on the lathe sometime soon and improve on the current design.

    To just quickly answer 3dmakers reply: This whole technology is old. Older than most of us here, i guess. Time to get mature. Older than some of the technology used that enables us to comment on each others comment, right now. At least for this I don't have to do shamanistic dances around the computer - well, not that often, anyways ;)
     
  5. elmoret

    elmoret Administrator

    Joined:
    Mar 1, 2014
    Messages:
    586
    Likes Received:
    91
    To bottom out the PTFE and remove the gap, simply pull up on the clip after inserting the tubing fully, then while holding the clip up be sure the PTFE is bottomed out.

    This is crazy talk! You can test this by trying to blow through the heatsink with the heatbreak tightly threaded in and the end of the heatbreak plugged by your finger. If your mouth cannot move air, then I am certain that the orders of magnitude less) force from hot air rising cannot either. Furthermore, even if it were possible for a small amount of air to travel between the two parts, air is an insulator, with very low specific heat. It cannot carry appreciable thermal energy when compared to aluminum or even stainless steel.

    In addition to the fact that you want to have two different materials for their different thermal properties, there's the issue that a one piece design means that the heatbreak and heatsink both have to be replaced if the heatbreak is broken, like on a print head crash.

    The wall thickness *is* what matters! Thermal conductivity is defined as W/(m*K), or watts divided by meters times degrees kelvin. There is no mass term, there is no volume term. Only length, as in the length from the cold part to the hot part. Thermal mass absolutely, 100% does not change the operating temperature of a device at steady state, only in transients. For reference: the definition of specific heat is dQ/m*dt, where dQ is the heat supplied (joules), m is the mass, dt is the temperature change.

    As it is, the wall thickness of the heatsink is roughly 3mm, about 0.7mm of which is taken up by the threads for the heatbreak. That leaves about 2.3mm to serve as a base for the fins. Lets see how much temperature drop we're seeing there, noting that the heatsink dissipates roughly 8 watts and the thermal conductivity is 167 W/m*K.

    The equation for heat through a cylinder is:
    q = 2 π k (ti - to) / ln(ro / ri)

    Restating:
    dT = q*ln(ro / ri) / 2 π k

    dT=307*ln(6/3.7)/(2π*167)

    where:
    q = heat transferred per unit time per unit length of cylinder or pipe (W/m, Btu/hr ft), ~8/0.026 in this case, or 307W/m
    k = thermal conductivity of the material (W/m.K or W/m oC, Btu/(hr oF ft2/ft))
    to = temperature outside pipe or cylinder (K or oC, oF)
    ti = temperature inside pipe or cylinder (K or oC, oF)
    ln = the natural logarithm
    ro = cylinder or pipe outside radius (m, ft)
    ri = cylinder or pipe inside radius(m, ft)

    That's 0.14 degrees Celsius from the inside of the heatsink to the outside. Even if you managed to drop the wall thickness in half (sacrificing a lot of strength!) you'd be reducing the temperature by a tenth of a degree. I hope we can put this point to rest now.

    E3D themselves says the heatsink should be cool to the touch:

    http://wiki.e3d-online.com/wiki/E3D-v6_ ... nt_Jamming

    and that you can use thermal paste if you like to improve the conduction between the break and heatsink. E3D does recommend this. You could make the argument that they should be including it (fair, and I'd be inclined to agree with you), but except for certain edge cases most folks simply don't need it.

    Last I heard, the reported jamming rate for E3Dv6 hotends was less than 0.2%. Considering that the defect rate on Apple's iPhone defect rate is 1%, I'd say that E3D's not doing too bad...

    It does not cause problems.

    All of the early achievements are locked in patents and trade secrets by companies like 3DSystems and Stratasys. The Reprap folks E3D, et al) have had to reinvent the wheel without enjoying access to the details of those first few decades. Though 3D printing has been around for a few decades, the RepRap movement is 7 years old. Nowadays, you can buy a printer for $1000 that rivals or bests a $20k Stratasys. I'd say that's some quite impressive progress for 7 years!

    As to your point about this technology being older than most of us, I doubt we have any 6 years olds on here, but you never know I guess!
     
  6. Sanjay

    Sanjay Administrator
    Staff Member

    Joined:
    Feb 27, 2014
    Messages:
    125
    Likes Received:
    118
    Before I go into the points where I disagree with your analysis of the design I want to state that I have no record of any contact from you on any of our support systems. If you’re having problems please please let us know, we have a comprehensive system in place for troubleshooting, support, correcting errors, and replacing parts if they are defective. As it stands I don’t know why you’re seeing these issues because I don’t know what extruder, filament, drive gear, print settings, speeds, temperatures, retractions, thermistor settings, or cooling setup you’re using.

    From the scant information I can glean from this thread, it looks like your cooling setup (fan/airflow) may be part of the problem. That or perhaps issues at the extruder.

    What mount/fan duct are you using? Is the fan connected to the power supply to be always on?

    If you contact us through support we’ll figure out what’s wrong, and if needs be we’ll replace stuff or work with you to get the working product you’ve paid for.

    We’ve shipped tens of thousands of these hotends, to large OEM printer manufacturers, resellers, NASA, MIT, the list goes on. We’ve been reviewed by some of the foremost people in the sphere of 3D printing, and reviews have been universally positive.

    As Tim stated our reported defect rate is below 0.5%, (where ‘defect’ is people coming to us and just saying “I have a problem”) with almost all cases being resolved with simple support to the user.

    If you’re having problems we will support you, we will get you up and running, and we will replace parts where needed. But immediately launching into a tirade on a forum without even reporting an issue seems improper. I can understand your frustration, but please let us help you.

    If the fan correctly positioned with air flowing over the heatsink, and with the fan connected to the power supply to run at all times then there will never be temperatures over 30-40C at the top of the heatbreak. We know this from extensive testing and temperature probing, not conjecture or reliance on simulation.

    What is happening here is most likely high retraction distances pulling the semi molten tip of the filament right up to where the bowden tubing meets the heatbreak.

    As Tim correctly states above, threads necessarily require clearance. When the heatbreak is tightened the bottom surfaces of the heatbreak threads are forced into contact with the upper surfaces of the heatbreak. The heatbreak is absolutely in thermal contact with the heatsink, otherwise it would be hot - which it isn’t.

    The assertion that there is some sort of “chimney effect” is completely absurd and I am not even going to entertain it with an explanation, Tim has already explained this.

    Thermal paste can be used here, but it is only ever needed in extremely marginal conditions such as heated chambers or people using cooling systems that do not move as much air as the recommended setup.

    It is completely thermodynamically infeasible to create a one piece machined item that has the conduction and dissipation requirements. People have tried this, and have failed. I do not fully understand your proposal regarding a “push in conus”, but I think I understand what you are getting at in principle. It is worth noting that short of an interference or shrink fit (which would make replacement of the heatbreak or disassembly impossible) there will always be clearance between two items, even when using a cone. In fact threads are a particularly good geometry to create thermal contact between two items.

    You are completely misunderstanding the thermodynamics of the system. The assertion that “mass stores heat” is not true, and demonstrates a lack of understanding of the principles at play. The job of the heatsink is to maintain a controlled thermal gradient across the small heatbreak junction. The additional material at the bottom of the hotend serves as a heat-pipe to pull heat away from the bottom sections where it can be dissipated evenly by all the fins. If you make the cuts between the fins deeper at the base of the hotend you will trap heat there and actually increase your heatbreak temperatures and ruin the thermal gradient control. I know this, because we have tried it at length and spent a lot of time with that particular geometry in order to get the most effective removal of heat.

    Making comparisons to computer heatsinks here is unhelpful. A CPU heatsink is designed to maximise absolute heat removal, our heatsinks are designed to maintain a particular thermal gradient at particular points in the system.

    Yes, it does actually. We have a very controlled process for surface finishing the inside of the heatbreaks. There is no need to go attacking your heatbreak with abrasives, drill bits etc.

    We have our shit together. Go and drill some deep holes in small 316SS parts and tell me how easy it is before you go making statements like that. We have QC, we have some of the best tooling available on planet earth. The ultimaker “heat break” is a terrible example to use here and barely performs the function of a heatbreak.

    Once again, this is not how thermodynamics works. The amount of heat transferred up through the thin section of the heatbreak is dictated by fouriers law. You could make the hot side heat break threads 10x as long as they are now, and it would make negligible difference. Except in cases with extremely high extrusion rates the hot side of the heat break is isothermal with the heater block.

    Introducing PEEK or heat stop paste can only serve to fill the gaps between the threads of the break and the block that were formerly occupied by air, increasing transfer of heat to the break. Additionally neither of those materials are rated for continuous use at 300C.

    This is not a manufacturing error at all, and simply an effect of the surface finishing process used inside the heatbreaks. It’s accounted for and managed, it is not at all an indication that anything is wrong. It is not the cause of problems.

    I am not entirely sure I follow your coffee pot analogy entirely, but I did find it quite funny.

    I do not make the above statements to provoke you, but I feel it is necessary to rebut derogatory statements that are not based on a firm understanding of the principles involved.

    If you’re having issues, please let us help you, please contact support, and we’ll do our utmost to support you.
     
  7. mr_clark1000

    mr_clark1000 Member

    Joined:
    Jun 10, 2015
    Messages:
    4
    Likes Received:
    0
    I have to say, I agree with all the responses to this thread so far. The E3D system is very well machined (I machine metals as well so can understand good machining) and a lot of thought has gone into the design.

    I would say that the original poster has managed to get caught up in a complete misunderstanding of thermal properties of the materials that make up the E3D hot end. Claims such as these should only ever be backed up with hard evidence, i.e. an accurate thermocouple reading from various locations around the hot end and inside the heat break. Doing so would show that the temperature within this area is in fact very low and well below the glass transition temp of PLA.

    To me, it seems this is an emotional rant based on frustration more than anything else.
     
  8. RobBen

    RobBen Member

    Joined:
    Jun 3, 2015
    Messages:
    3
    Likes Received:
    0
    Sorry, busy week, not much time for forums.

    I did last week, and did again this monday after quickly glancing over the replies - still no response. Filled out the survey etc ...

    Without thoroughly reading through, just a 5 minute comment: On one hand it is supposed to be good to dissipate heat away from the break as quickly as possible, on the other it is claimed to be good to spread heat evenly across the whole part. I do not get it. I had a look at some other all metal hot ends like the micron3d, which avoids any gaps in the setup from top to bottom, uses tightly threaded heat breaks and a massive first layer in the heat sink instead of a thicker bottom center which seems to do the trick to not having to oil my filament.

    If the heat would dissipate quickly enough from the heat break into the heat sink, there would be no melting problem between the heat sink and the bowden setup, would there?

    In the end i chose the E3D because it was supposedly easy to implement into my UMO, as the heater blocks are compatible (according to the e3d wiki). And of course I wrote out of frustration. Not because I was so happy :) Handling frustrated customers is what customer service is about, the happy ones are no problem. :?
     

Share This Page