@Ryan_Carlyle . Now where is the fun in that!?
I completely see your point, but developing something that the open source community can build themselves, improve upon, and is cheap to make… THAT is the goal.
This new printer I’m working on, while crude, is a little like my original Printrbot… No one really tried it or gave this approach serious consideration b/c it’s not what an engineer would build. Well, I’m not an engineer, so there ya go.
I am hoping it will entice a few people to try their hand at very large prints since it will be cheap to build. The barrier will indeed be the cost of plastic. Pellets are a lot cheaper, so looking forward to that.
Yes we should leave everything in the hands of Industry, like the 3d printers that haven’t been even known about for 30-40 some odd years.
3d printers as a side project were unreasonable to build and design, now every person that knows how to turn an m3 screw has N designs of his own.
The engineering required to get good results in any hardware project will not match the complexity of many open source software projects… they go on just fine (and end up better than other solutions)… Apache, GNU Tools, Linux Kernel, R, Python, C, GCC?
I mean those people are thinking about a lot more than heat dynamics and the flow rate of material through a cylinder.
Why not promote trying in words? Words can promote just about anything, why ruin them by only showing a functioning real world product tied to a monster of a word like OEM?
Big data has changed a lot of this ideology @Ryan_Carlyle and it did it a long time ago. Hardware developers are just finally respecting software developers and software’s complexity enough to even talk. That’s called having been and still being a snob.
@Andrew_Hodel There’s stuff that makes sense to DIY, and there’s stuff that doesn’t. We make PCBs and nozzles and gantries, we don’t make motors or microchips or bearings. (Not good ones, anyway.) Some things require legitimate expertise to do well, others require scale production to make sense.
I have complete faith that @Brook_Drumm can make a pretty good pellet extruder, because he has access to the required tooling (like CNC lathes), resources to build many iterations, and lots of community good-will to tap for outside assistance if needed. But it’s foolish to ignore decades-old industry techniques and existing body of knowledge here. We don’t need to reinvent the wheel. If you want an open-source pellet extruder, start by looking at what the industry does, and see how you can make it cheaper/easier. (That’s how Dr Bowyer started the RepRap project.) It’s a million times easier to reverse-engineer than to start over. Brook should start by buying an existing product, doing some benchmark tests, and taking it apart.
I say this because I see WAY too many people sticking a wood auger screw in a heated barrel and calling it a pellet extruder. That’s naive and ineffective. Do some background research on the physics and established practice first so you don’t run into obvious issues people figured out 50 years ago.
Personally, I HAVE done a ton of research on pellet extruders, and don’t think the “standard approaches” are viable for precision 3d printing aside from spiral-vase prints… unless you can figure out the really fancy high-end control techniques ORNL is using in the BAAM to compensate for flow lag effects. That’s viable, but when you consider how much trouble the open source community has had getting even something “simple” like working extruder advance algorithms into any firmware other than Sailfish, I just don’t see it happening. Pellet extruder volume control algorithms will be like JKN on steroids. We’re talking multi-second flow lags with non-linear strain-history response due to complex viscoelastic behaviors. It’s utterly nuts how hard the math is to extrude pellets at anything but steady state conditions.
@Ryan_Carlyle I really like your viewpoint and will research those links. I agree that we should not ignore any current, well-developed tech!
Our work with Tinyg gives me hope that they will bring significant improvement to extrusion by applying significant research and math to the problem. They are calculating volumetric flow and adding acceleration to the process of extruding. Admittedly, they are doing not for solid 1.75 filament, but no reason why this approach might put pellet extrusion in teach to get some benefit over the simple methods currently employed. Their goal is to eliminate retraction and ooze. They are calculating the volume needed for extrusion for the specific angles of corners that are pathed which should eliminate the corner build up problem.
I can imagine that much more control and measurement of flow will be needed to compensate for the pockets of air that exist in between pellets. No real ideas there yet, or even data to know whether or not that this will be a problem.
One crazy idea I had in passing, that might end up being useless is: what if it were a two step process…
Use a Lyman extruder to make thick filament- say 6 mm diameter… The final output would be measured in length and recorded. You would need to have a cooling loop that “stores” a bit of extra filament in reserve, like a serpentine path with some sort of slack at the final output
The extruder would essentially be a large 6mm extruder that works very much like today’s extruders- managing its flow in firmware normally.
The two systems would need to talk to each other. The pellet extruder would need to be able to keep up with the highest volume needs that the printing speed requires, of course. But If the extruder slows due to lower speeds on perimeters, bridges, or travel moves… Then it should signal the pellet extruding machine to slow and stay within limits as to not over-produce and spit out more than needed to avoid making too much. The slack would stay within tolerance… Let’s say a you have a tolerance of 6-10 feet of slack that allows enough time, at max speed/extrusion, to allow the pellet extruder to ramp up and make more until the optimum slack (extra material) can be supplied to have the two systems reach equilibrium.
It’s not super elegant, but it would split the problem into two parts and could be done relatively easily with current open source tech.
Just an idea, but it would be one wants bring the two camps together- those with expertise in pellet extrusion and those with 3D printer expertise.
I do think that a filament diameter measurement system is needed to give extra control to the final extrusion if out of spec filament comes along down the line. Current tolerances in factory-made filament is largely “good enough” for most desktop machines, but making filament from pellets can currently lead to larger tolerances. This final check would allow the extruder to compensate for perfect volumetric flow even with out of spec filament-- either out of round, or too small of a diameter. You can’t do much with filament that is too large and doesn’t fit the final hole it needs to travel through.
Oh, one other benefit from this approach: the weight on the final extruder head can be quite light, since the large motor required for pellet extruders can be kept off to the side and just feed a conventional extruder.
One more: the approach mentioned above could also be a candidate for multi-head/material machines. Just add another pellet extruder (or four) and have them feed multiple direct drive (conventional) extruders.
We are working on a dual head bot that parks the extruder not in use at the side. This isn’t an original idea, I’ve seen it several times in existing open source designs/products. It does add an extra X carriage and linear rail(s), but manages ooze with an ooze bucket and a wiper to minimize mess.
@Brook_Drumm Extruding filament on the fly into a reasonably lengthy buffer, and then measuring actual diameter to compensate for variation, should be a perfectly viable approach. It’s just a lot of moving parts in one printer
Do we even have “affordable and effective” filament extruders yet? They all seem finicky to me. All the Kickstarters have been failing due to technical shortcomings. Seems like we need to really nail down open source filament extrusion before we try to scale it up 50x and add rate-change complexity. But maybe I’m just not dreaming big enough
A thought on diameter compensate… it doesn’t respond instantly, at least not in the systems I’ve seen. Tracking the distance between the sensor and extruder, and then adjusting flow rates possibly mid-segment, is a lot to ask. Given how much trouble we have getting open source pellet-extruders to work reliably and output +/-3% diameter filament AT STEADY STATE, it’s kind of hard to imagine constantly changing the flow rate to loosely track extruder flow and getting good results. If you do major screw speed / drawing speed changes, you’re going to have big diameter problems. Maybe so bad that diameter compensation doesn’t work well enough, or the extruder jams. So off the top of my head, I’m thinking you’d need a pretty long filament buffer and some kind of clever feedback algorithm that very gradually adjusts the screw speed as the buffer changes length.
At a certain point, you have to say “this is getting silly” and change the approach. For example, instead of making filament in order to get the “near-positive-displacement pumping” performance of a filament-pusher extruder, just put an actual positive-displacement pump between the screw and the nozzle. I talk about that a bit in the google groups thread I linked earlier. That gives you precise volume control without having to actually make filament. It’s just not a cheap piece of kit, primarily because of the high temps we’re dealing with. There’s some legit engineering there in the seals, “molten bowden” flex lines you might need, and so on.
Filabot has been around a while and gets good results. The Lyman extruder, which I haven’t revisited in over a year had a pretty sophisticated system to manage tolerances. I built one that worked but never built the new measurement system. There are at least two others I can think of that boast industry standard tolerances.
The extruder could simply be used to make the spools of filament for the big machine I guess. It would simplify trying to control the feeds. Existing machines are designed for 1.75 and 3mm. I’m not sure if there is a larger standard size we could set up a pellet extruder for. I guess making a large direct drive extruder to extrude regular filament will be the fastest approach. It begs the question- if a larger standard size doesn’t exist, what size should we make?
Crazy thought: what about an extruder that accepts three (or more) 3mm filament strands? That should be pretty easy and doable with some clever machining in the hot end. As long as the final tip doesn’t build up too much pressure, it should be manageable.
Ok, so back to the original notion, a large format 3D printer to make body and even frame parts for an EV @Brook_Drumm , how about a large Delta bot? I’m also curios about your EV project and, of course, am going to want one of your TinyG based boards as soon as they come out.
@Brook_Drumm I like the tri-feed idea, we could try it with existing mixing extruders like the diamond hot end. Should give more surface area for heat transfer to melt filament faster than a single feed at triple feedrate.
I would propose 4.7-4.8mm as the next standard filament size. This is 2.6x larger than 2.85mn filament, which is 2.6x larger than 1.75mn filament. And it maintains the same clearance to a standard PTFE tube ID. And it’s a standard plastic welding rod size in the US (3/16"). Lots of benefits.
@Samer_Najia The issue with a delta is the build volume arrangement compared to a car body shape. You’d need an exceptionally tall delta to get the build diameter large enough for the car’s length. (Unless you print the car body vertical with loads of supports, which would reduce the diameter requirement but would put the weaker layer/layer bonding on a bad orientation for chassis strength.)
For extremely big volumes, cartesian bots are better than deltas, because the printer only needs to be slightly larger than what you’re printing. Almost all of the giant “house printing” projects around the world have used big bridge gantries.
It’s merely a dream but the open nature of Local motors design is a possibility. Also a printable version of https://www.osvehicle.com would be cool… Leveraging their good work!
We are interested in a stretched Evo, so we can put in a conventional body on the 4 seater. While I would love to make a sleek sportscar like thing, we are more likely to make most panels flat a-la ‘Cube’ or Honda Element…or maybe like a mini H3.
