Maximum theoretical print speed

I was hoping someone had already done some hard thinking on this. What is the maximum theoretical print speed of FDM using a material like PLA or ABS? What are the primary factors effecting speed or is it as simple as speed of plane movement and speed of extrusion?

Take something like the stratasys FDM machines, how fast are they in comparison with something like a properly tuned mendel?

As I stated in another thread you may want to search for, the hard limits are thermal- each layer must freeze before the next one is put on top. Thinner extrudate allows faster cooling, but also necessitates far more machine work per layer, and more layers per object so you asymptotically approach some ideal.

The thermal conduction of the plastic can't be changed unless you change to a different type of plastic. You could make the build chamber colder, but then you get warping issues. Ideally, you would use a cooler that rapidly cools to some intermediate temperature like 60 celsius then leaves the object there- fan forced perhaps? This is where a heated chamber would be handy.

All in all, it's an asymptotic process- each improvement in build speed takes exponentially more work than the previous one, and is exponentially smaller than the previous improvement.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

I did give the search a go but I really couldn't find anything in all the noise. Any advice on search terms or possible date range?

Or maybe you just recall the conclusion, how far is mendel from the theoretical max today? Is it 10% of max, 50%, etc?

I think Mendel will go as fast as you want. Certainly faster than commercial machines as it it much lighter. I have an extruder that will do 0.5mm filament at 128mm/s. My Darwin can also move that fast, just never got to marrying the two together. It might shake itself to pieces though.

As TH says, cooling will be a problem and I have no idea whether quality will suffer due to filament dynamics. Should know in the next couple of months though. I have the resources, but I just don't have the time to try all these things.

---

[www.hydraraptor.blogspot.com]

For now excluding filament/heat dynamics... I think heated bed or heated airspace can address some of those issues.

So if I have my math right you can push out 2.5 cubic centimeters per second or is .5 the circumference of the extruded material? It doesn't seem to me that belt driven planes can move that quick or at least there will potentially be inertia issues in direction changes. Do you have any thoughts on if rack/pinion planes could move faster?

0.5mm is the diameter, so 0.5/2 ^ 2 * pi * 128 * (conversion factors) = 1.5cm^3 per minute.

With the acceleration feature of the FiveD firmware, the belt driven stuff should be able to move that quickly without problems. Steppers' maximum speeds are far beyond what most people imagine because they can't instantly start moving at anywhere near that speed, and nor can the carriages they connect to.

---

-----------------------------------------------
Wooden Mendel
Teacup Firmware

Even without acceleration my Darwin could do 120mm/s from a standing start with NEMA23s running at 1/3 power. I think Mendel would probably match that with high torque NEMA17s as it has less moving mass.

The problem of cooling goes away if the layers are big. I currently slow down when the layers get tiny, so I could (on Mendel or Darwin) speed up when the layers get big.

---

[www.hydraraptor.blogspot.com]

Thanks, yea, those mm2 to cc just don't work well in my american brain. So that makes the theoretical maximum speed to build a set of mendel parts almost 15 hours, approximately 1300cc I think. At least I didn't cause a satellite to crash a la NASA. Realistically that might be more like 25-35 hrs including all the various setup/teardown/babysitting. That really seems like too long to me...

One thing we're going to have to take into account is inertial forces. As the extruder goes around the turns doing the raster scan or sharp corners it's accelerating a fair amount. At 128 mm/s the extruder's experiencing 6 g of acceleration as it goes around the raster scan turns(print speed doesn't change throughout the print?).
0.001(m/mm)*((128 mm/s)^2)/(0.5*0.5mm(width of filament))=65.5 m/s^2= 6.6 g

Centripetal acceleration=(velocity^2)/(radius(which in our case is equal to 1 half the filament width))

So inertial forces increase as a square of print speed. For example, if we increased the print speed to 200 mm/s, the extruder would be accelerating at 16 g as it goes around the corners of a raster scan pattern. This is more than the maximum acceleration of a fighter jet.

At higher print speeds things might become inaccurate due to the stretching of the belt and elastic deformation of the axes. We might be able to go to higher print speeds by not doing a raster scan pattern and not printing things with sharp corners at such high speeds. Is this feature already implemented though?

I was thinking that the build could be in two phases, outer shell which you do in .5mm fiber and inside sparse fill with a second extruder at 1.5mm.

Thoughts?

goinreverse,
I think 1300cc is the volume of the parts, but I make them all with 25% infill so the predicted build time is less than that. No idea if it is practical though till I try it.

---

[www.hydraraptor.blogspot.com]

freds,

Having two filament sizes is tricky unless you keep the same layer hight, in which case all you gain is stronger infill.

Consider what happens with 0.5mm outline and 1.5mm infill. You would do three outline layers and then a layer of infill. What happens if the side of the shape are not vertical but sloping outwards. You would have to do extra outlines with the 0.5mm extruder to fill in gaps that stepping out in 1.5mm steps would leave.

The same with small features in the outline. The infill would not be able to make it into all the corners of the outline so you would have to build them inwards with the small extruder first.

Certainly possible in theory but adds to the software complexity greatly. Its not something I would bother with now but it would be the only practical way to make very big things with fine detail.

Another way to make parts very fast it to make them completely hollow leaving a small hole in the top to pour in polyurethane afterwards and another to let the air out. These could be like normal risers on traditional moulds and the sprue cut off afterwards with a saw.

Don't make the mistake I made and try injecting PU under pressure. The pressure makes it set instantly and block the nozzle.

---

[www.hydraraptor.blogspot.com]

Gene Hacker,

Skeinforge does have a setting called Fillet to deal with the acceleration around corners by rounding them slightly: [www.bitsfrombytes.com]

It is on by default (at least in the version I am using) and I haven't played with any of the settings, so I have no idea how much it actually helps. But I haven't tried printing at higher speeds either.

nophead,

What if the larger extruder squashed the filament, so you would end up with (for example) a 1mm wide 0.3mm thick "stripe"? In fact, doesn't skeinforge already allow you to set the extrusion width separately for perimeter vs infill? I wonder what would happen if you set the infill ratio to something like 3x or 4x.

Yes you could make the infill wider but that does not speed up the build. It just makes the parts stronger, but they are already very strong with 25% fill.

If you make the fill even sparser then you have problems bridging it when making solid layers on top.

---

[www.hydraraptor.blogspot.com]