I've published a commentary on 3ders.org  today. It's a personal view on developments surrounding "smart

I’ve published a commentary on @3ders.org today. It’s a personal view on developments surrounding “smart hotends”, featuring several of @Kai_Parthy 's recent inventions, what the community did so far and what @Stratasys_Incorporat patented.

I would love to get the community’s thoughts on all of this - especially from guys like @airtripper , @Whosa_whatsis , @Erik_de_Bruijn , @Daid_Braam , @Brook_Drumm , @Sanjay_Mortimer , @Thomas_Sanladerer , @Nils_Hitze , @Roland_Littwin , @Jeremie_Francois and all the other 3DP brains out there :wink:

What’s your take on it? Are hotends currently smart enough? Do we need one specialised hotends per material? Or is there something like the one and only smart uber hotend?

Make them ubersmart. There was someone who did experiments with a load cell to determine nozzle pressure, this allowed them to fine-tune retraction and speed settings. I assume it would also help detect clogs and temperature settings (to a small degree). It can also stop oozing and stringing.
load cell (or nozzle pressure) and filament diameter sensing can greatly increase the reliability of the machine.

@Stephanie_A this is @airtripper (aka Mark Heywood). He did some great work there on the sensor side, Kai is more focused on the hotend design itself. I think multiple heat zones (like in a standard extruder) could be helpful.

After tons and tons of testing, I don’t think there will be 1 hotend fits all solution.

I don’t think the proposed smart hotend will be cost efficient and 100% working for PLA. And what will work perfectly for PLA won’t work for PVA.

I don’t know how broad a range of materials one tool head will eventually manage, but I don’t think it’ll matter a ton in the end.

I think we’re going to end up with switchable tool heads eventually. At that point, the question of material flexibility for a single tool head becomes much more of a trade offs question than a fundamental capabilities one.

@Florian_Horsch_flouS The downsizing at Makerbot had little to do with the poor quality of the smart extruder . The company was scaled by people that weren’t used to dealing with scale and growth of that magnitude. That, combined with wildly optimistic sales and continued growth forecasts the company wasn’t structured in an efficient manner. They still sell all they produce but didn’t get the hockey stick pattern of growth they expected. There are also economies of scale at play in that there were several duplicated areas when Stratasys bought Makerbot. Much of the downsizing reflected those duplications.

lovin it - "Just because the current generation of hotends somehow works, we shouldn’t hold back to take the next major step towards a much more reliable and versatile design. " True

Well, I sure hope that this idea turns out to be a dead-end, because if it turns out that this system is exactly what we need, the market is going to stagnate for another two decades waiting for this patent to expire. When will people learn that every new patent issued extends the dark ages by 20 years?

I’m pretty dubious of the claims, though. Yes, some filaments will benefit from a little pre-heating, but there are just as many that would probably be better-off being cooled to below ambient before entering the melt chamber, which this is not capable of doing. If you could add multi-point dynamic filament channel diameter to this system, you’d probably have something, but I don’t know of any technology that would make that possible under these conditions, certainly not one that would fit the size and thermal constraints of this application.

@Ross_Hendrickson , @Camerin_hahn and I have done a lot of brainstorming on how to build sensors to figure out these types of filament properties so that standardized data can be included in a UFID (https://plus.google.com/communities/107859862288161234107) tag. We want each of these systems to test one property at a time and test it well, not try to discern everything at once from a single signal with dozens of factors and other unknown noise factors (for example, someone just gave me samples of some PLA filament with a rough surface that feels like sandpaper, which would surely affect the measurements on this sensor more than it would affect the melt properties).

I think hot ends need to be more modular, with some simple, inexpensive components that are easily replaced and some more expensive durable parts that (theoretically) never need replacing. The current philosophy, which this design doubles-down on, seems to be that all parts are durable, which isn’t true and will be proven even less so as filled materials become more popular.

Very good article @Florian_Horsch_flouS - it was great to finally meet you and @Kai_Parthy in Berlin. I’m also of the view that we will need quite different hot-ends for different materials. Having a more complicated (multi-heat-zoned) ‘programmable’ hot-end will not totally fix the issues for many-material, print speeds v quality and a whole host of other considerations, many we still don’t really understand. FDM is still such a crude and basic way to make finished parts, although it does produce strong and accurate prints when setup well. I’m not convinced that adding multi-staged heating complexity into a technology this basic really adds much to the output or even makes it easier for ‘everyone’ to 3Dprint. I would always encourage people to be very aware what materials they are putting into their printer, and exactly what temperatures they are being melted at. - Switching to a good quality thermocouple makes for a significant step forward for most users, at least temperatures being quoted should be closer to reality. Even if that makes it easier to support people and improve print quality, that’s what most people need.

It would be nice to see how this system can detect material type, I’m not convinced yet, due to the massive range of properties ABS and PET show by different manufacturers. Even PLA can act very differently depending on the batch, colour additive and impact modifiers.

By far the easiest way to get good settings is to have the material data read from the material spool into the printer, that way the hot-end does not need to try and work it out.

I’m all for a little more intelligence in extruders and hot-ends, but going to this level of complexity does not seem worth the hassle, benefit or print reward, it may be a step too far for FDM.

Another really tricky aspect is when you start going faster, and that’s a significant factor for most people, printing fast with good quality – then ‘melt’ temperatures are so transitional that the energy transfer seen by the material is the bigger issue. Start to increase the melt-zone or move to a multi-zone warming-heating-melting system can get you more speed, but other issues start to crop up. With retraction, oozing and physical speed of travel moves still being way too slow for the process of spitting out molten plastic from a tiny nozzle. Sensing pressure here will help, if you can get a closed-loop control heater system going fast enough…

I don’t want to be or sound negative. I’m sure we will crack many of these problems to keep things open and advancing for the greater good. More open discussion on them is a fantastic thing.

What does everyone else think? Please challenge my views too, they are only one opinion.

That was a lot longer than expected !

At the Makerbot booth at CES they were talking about specific extruder heads for specific materials. They had a hammer (and several other objects) that had a wooden type handle and some sort of metalic infused plastic as a head. He told me it came off a prototype machine that uses different heads for different material.

I don’t know of a process used in manufacturing that has the same toolhead for all applications or materials. I don’t see the future being fused deposition but rather resin or power based. I’m with Rich in that this level of complexity is a lot of effort for the benefit. In terms of what the industry considers a precision tolerance FDM likely isn’t going to be able to acheive that and if it does, another process like SLA may be able to more cost effectively exploit the technology making any advance in FDM a moot point.

Honestly, I need to make one of these simply as a hot end test rig for quality control.

The second heater is interesting. I think it’s unnecessary/overkill though. The highest acceptable cold-end inlet temp is roughly the same as the optimal chamber temp for warping prevention. In other words, merely running your filament feed through your temp-controlled build chamber will adequately pre-heat the filament to the highest temp that can still be pushed through a feed path without deformation. (Filaments that don’t benefit from elevated chamber temps, like PLA and PVA, have such low glass points that preheating them will cause jams.)

Material identification via specific heat capacity sensing is a good idea. Personally, I was looking at heat capacity sensing as a means of melt-power compensating hot blocks to allow a wider and faster dynamic range of melt flow rates without PID loop instability. But that’s only really applicable to high-speed printing whereas material detection has some nice broad applications. In any case, there are fairly simple ways to accomplish this with a standard hot end containing no extra sensors.

The article is vague on what materials don’t do well with a current hot end. I don’t think there needs to be a different hot end for different classes of plastics. It might just be a matter of degree, when you’re talking the upper reaches of performance. I wouldn’t mind a system that’s designed for quick change of hot ends though. That is a matter of developing a holder.

Having four thermal sensors sounds like an invitation for trouble.

@Jeff_DeMaagd PLA, high temperature (300C+) and water soludable materials in a single hotend proves to be very tricky to impossible.

@dstevens_lv I’m partly with you. Definitely it was based on the gap between their estimates and the reality in sales figures. At least in the EU they could deliver quite quickly, but poor word of mouth about the performance of their extruders led to a situation where at least informed customers switched to other vendors. Adafruit’s CEO seemed to be on the same track: http://www.3ders.org//articles/20150622-interview-adafruit-puts-ceo-of-makerbot-in-the-hot-seat.html

@Whosa_whatsis true that! Regarding the patents I feel the same way like you. I don’t personally get my head around them. Like you, I don’t see that the original purpose (incentive to invest in inventions) is still achieved, even worse: There are plenty of indications that the opposite is done (especially for small ventures).

On the other hand: I fully understand Kai, that he wants to protect his work - he’s a one man show. And he has plenty of experience with being first on some bigger things (LAYWOOD…), but then seeing others ($_insert_filament_vendor_here) making the money out of it. Funnily enough we could now all ask ourselves if Kai is even able to defend his patent, given he hasn’t a large legal department working for him. This leads (again) to the grotesque realization, that patents are not working correctly. Argh!

@Richard_Horne : Missed to link you to the article (as many others I guess)! Our chat at #3DPSBerlin still resonates with me :slight_smile: Still, I thought that it might help Kai to summarise some of his ideas, compared to what others are doing. Even if I’m not the “OMG, this definitely will solve all our problems” guy.

I think only a working prototype can challenge your, my very own and the doubts of others. @Whosa_whatsis , isn’t it funny that you can file a patent without a working prototype or at least a mock-up of what it should look like in reality? Those sketches and text are so vague, that you don’t even know what the patent is all about.

@Jeff_DeMaagd , it’s not only about though combinations like @Daid_Braam mentioned, but also about more ease-of-use. What is the ideal temperature for a certain batch of a certain color at a certain print speed? All of this should be answered by a truly smart hotend (without user input!). But then again @Whosa_whatsis … wouldn’t be the algorithm more patent-worthy than the actual hardware design? Funnily enough software patents in the EU are a big mess (http://blogs.fsfe.org/ciaran/?p=115). Effectively there are no enforceable software patents in Europe, which leads to more innovation in the eyes of many. I wonder if we’ll live in times where this thinking is applied to normal patents.

@Florian_Horsch_flouS but temperature is not just about extrusion, it’s also about adhesion.

A recommendation I’ve heard from various people (law profs and the like) re: patents for EnablingTheFuture.org is to file provisional applications then allow them to expire. It costs about $100, and serves as incontrovertible prior art on the idea from the day you file. You don’t get any rights, but you also make it near impossible for others to get (valid) patents on the same ideas.

It’s a cheaper, less centralized route than a defensive patent pool for OSHW.

Just overnighted a 5kg sensor. I was asking about how to measure filament pressure some time ago simply for quality control of hot end production, but it never occurred to me all of the other crazy things you could do.

@ThantiK another option without a load cell is to see how much weight your extruder can lift during a load sequence with the nozzle removed, and then see how much it can lift with the nozzle attached and extruding. The difference is the filament force due to back-pressure. Doesn’t really tell you anything for live prints, but it’s an easy way to characterize a flow/pressure relationship without any special gear.

@Florian_Horsch_flouS Regarding legal action for a patent, plaintif or defendant CNET has a piece a few years back from a study by the American Intellectual Property Law Association. http://www.cnet.com/news/how-much-is-that-patent-lawsuit-going-to-cost-you/

Without deep pockets a small operator is at a significant disadvantage. Even with the protection the large operations are going to able to design around may if not most patents which will further complicate the ligitigation.