This is my famous Voron Zero aka 247zero!
Speedbenchy world record and up to 1000 mm/s printing - yeah: It’s crazy
fast and it’s been a while. Now, it’s time to revisited and to review that beast
of a 3D printer. Why did I choose those upgrades? How fast is it really at quality printing? What
needs to be improved in order to go even faster at better quality? Ladies and Gentlemen, fasten
your seat belts and start your stepper engines, that’s gonna be an interesting
ride! Come on, do it! Sponsor for Voron Zero can be downloaded for
free at gumroad.com. It lately hit ten thousand clicks and one thousand downloads
– thanks a lot for your interest I have not expected that! If you appreciate my work: Consider
donating and rating when downloading. This helps a tiny bit to cover my expenses. A big thank you to
everyone supporting 247printing in which way ever! In early 2021, Stefan from CNC Kitchen showed
an adorable little printer going at quite insane speeds…at that time. For me, there was no doubt,
I needed to build that one and ordered exactly the same Formbot kit in China! After my better half
successfully built it (just kidding), a friend of mine asked: Have you tried printing a speedboat
according to the #speedboatrace rules with that? And that was when my YouTube channel was born,
because after two or three tries I uploaded an 08minute 48 seconds #speedboatrace print video.
That video was quite successful, as it was the fastest corexy printer in that challenge at that
time and because that hype about Voron printers was just about to take off. I loved the idea
of the #speedboatrace and I got a bit obsessed by its idea, namely: “Improve your printer and
your slicing in order to make it faster.” And that’s what I did: Step by step. On my Voron 0.0,
I tried to stay at default components as long as possible and I printed a second Speedboat still at
default hardware: five minutes and ten seconds and trust me this was tough on a Voron 0.0 with those
tiny Nema14 stepper motors and with stock Dragon Highflow hotend. The idea of using compressed
air in order to support the part cooling was not the greatest, but cooling is maybe the biggest
challenge to overcome at speedprinting and even more for a Voron Zero. The time had finally come
for upgrading the motion system, the cooling and the extrusion capabilities… in order to literally
print the world’s fastest 3Dbenchy in October 2021. Fun fact: The actual record holder at the
#speedboatrace also used my 247zero designs along with other upgrades. So, what’s the philosophy
and the ideas of my 247zero speed concept? Let’s dive in! After dedicating quite some time
for the speedboatrace, it was clear to me: Throwing stronger stepper motors, more stepper
motors or just more voltage at that printer only, couldn’t really be a clever solution for quality
at speedprinting: You’d only move the same masses quicker: On the same belts and on the same
supporting frame. That doesn’t help for accuracy and therefore doesn’t help for print quality.
Newton tells us one very simple rule: Force is mass multiplied by acceleration. In a simplified
manner, that’s mostly the only formula you’d need to think about in this concern. Stronger
motors just do one thing: More acceleration at same moved mass and therefore higher forces
hitting the construction of your printer, which causes displacements and therefore quality
issues on the prints. So, for the motion system I tried to cut off as much weight from the moved
masses as possible. That was when I had the best and the worst idea at the same time: I printed
the X-Beam, the toolhead and the joiners from PET Carbon Fiber. All in all I shaved off around
54 grams, which was an important enabler for that fastest benchy at the #speedboatrace. Though, it
had its downsides which we’ll discuss later …and 54grams doesn’t sound much, right? Let us talk
about relations. On a default Voron 0.0, which is a bowden concept, the all-in-all weight shoved
around is 268 grams…which already is ridiculously low. For perspective: The toolhead only of an X1
Carbon weighs 574g and is considered lightweight. For Voron 0, which is a bowden printer, shaving
off 54 grams of the moving masses is a weight reduction of over 20%. Looking at Newton’s
formula, we can profit from higher accelerations at the same forces on the construction. As a
result the resonances improved: The recommended maximum acceleration for quality printing is close
to 20 thousand mm/s² for the limiting y-axis. For the crazy three and a half minute
speedboat print I then was able to go at 150 thousand mm/s² acceleration… you might realize
a big gap between insane speedprinting and fast printing at quality! Can we ever close that
gap? Now that we can use higher acceleration and therefore also reach higher speeds, we also
need to extrude the filament at a higher rate. Esun’s ABS+ at temperatures at or
above 280°C quickly became the go to filament for the participants of the
speedboatrace-challenge. What we still needed were ultra-high flow hotends that
were able to melt the plastic fast enough. On my Voron 0.0 I started with a Dragon
High Flow hotend, which is good for around 30 mm³/s continuous extrusion. Sticking to
the speedboatrace rules you print at 0.25mm layer height and you typically use
the max. allowed line width of 0.5mm, as it simply means: Less movement necessary
and therefore lower print times. Theoretic continuous extrusion at 30mm³/s means:
You could theoretically print at 240mm/s. Now, because a 3Dbenchy is not continuously
extruded, as there are a lot retracts and small moves, you can set the requested top speed quite
a bit higher, but at some point, you run into underextrusion. CHT was not a topic yet at that
time. The Slice Engineering Magnum+ is capable, but heavy and way too expensive and bulky to be
mounted on a Voron Zero. So, longer meltzones had to do the trick. And that’s why I am happy to have
met Nitram with his crazy Delta printer. He is one of the early contenders at the #speedboatrace
and he experimented with longer meltzones in order to overcome the extrusion topic. He came
up with a Volcano style heater block for Mosquito style heatsinks and heatbreaks and organized group
buys for his block. All in all, a fully equipped aluminium Volcomosq is even a bit lighter than
my Dragon High flow. It’s good for around 40mm³/s extrusion rate, which I successfully tested with
printing at 1000mm/s – for science of course. I still use an E3D Hemera at bowden config,
which does its job, but there are quite some drawbacks – I’ll talk about that later.
Now that we leveled up motion and extrusion: What about the cooling?! Cooling was a tough
topic for my 247zero and it still is for most printers. The bowden toolhead of the Voron Zero
is defined for tiny 3010 blower fans, which are the same for the direct drive MiniAfterburner of
the Voron0.1. Part cooling always was the greatest draw back of the Voron Zero, so I experimented:
With bigger 4010 fans, with over-volted 4010 fans and with externally frame-mounted part
cooling inspired by MirageC’s HevACS. I was convinced very early that this was the perfect
solution for a small printer like the Voron Zero. I also liked the idea of removing additional
weight by getting rid of the toolhead mounted fans. I still wanted to point at least some
parts of the air stream towards the nozzle, so I quickly designed and added these fancy air
ducts. Four crazy 40mm by 28mm Delta Industries server fans are responsible for huge amounts
of air thrown at the build space. All in all, it’s an… let’s say… effective concept:
Quickly designed -definitely not perfect- with powerful fans and probably more air flow
than you’d ever need for regular printing. The rest of the story is history: Yes, I
printed a fast but ugly three and a half minute 3DBenchy (BSO), I did my super-scientific
1000mm/s experiments and I printed that 9minutes 3DBenchy at let’s say… near to acceptable quality.
The question many of you had and also a question no one with one of those crazy printers really
answered yet is: What about real-life or let’s say “casual” quality printing? Or in other words:
How fast is this 247zero concept really? Let’s compare it to a wide field of 3D printers I
have tested and which you have already seen in my last videos. We have a Prusa MK3S+, a BIQU
Hurakan with Klipper Firmware out of the box, the Bambu Lab X1 Carbon, the LDO Voron 2.4, the
LDO Voron 0.1 and the Flsun V400 delta printer with Klipper out of the box. Where possible, I use
the default slicer settings as well as the default Firmware settings. Concerning the models to be
printed for the print time comparison, I had to make a complete change in comparison to my last
videos, because of the tiny build space of the Voron Zero. We have Black Beard by Wekster, Minas
Tirith scanned and published by Sir_Ren, a “before Wednesday” Version of the Addam’s thing
by JS-Studio and of course the 3D Benchy designed by Daniel Norée at settings sticking to the
speedboatrace rules. Talking about settings: For the Voron 2.4 and Voron 0.1 I used my
speed/quality optimized settings in Prusa-Slicer which worked well for dozens of print hours.
For the 247zero I did a ton of testing silently some time ago in order to find a good balance
for quality at high-speed printing with PLA. The input shaping maximum recommended acceleration
for the essential y-axis, is stated close to 20,000 mm/s² for the actual setup. Acceleration
ranges from 18,000 mm/s² for perimeters up to 30k mm/s² for non-quality effecting features like
infill or travel moves. Speeds are set between 230 mm/s for outer perimeters and 800 mm/s for travel
moves. For the infill I go full bore at 600 mm/s and I combine the infill for every two layers,
because: I can. This demands a very high flow rate at that speed. Meanwhile, I upgraded to a CHT
0.4mm nozzle along with the CNC Kitchen Volcano adaptor, which I tested a while ago. For PLA
printing at manufacturers’ specified temperatures, the setup is good for 45 to 50 mm³/s on
a Volcomosq and that’s where I capped the max. volumetric speed in Prusa Slicer. But now…
let’s finally have a look at the print times: The big picture excluding the 247zero is quite
the same as on my last comparisons. With a BIQU Hurakan you already save quite a lot of time in
comparison to the Prusa MK3. The X1 Carbon print time of Minas Tirith is quite noticeable though.
I suspect the combination of low acceleration for those many top solid layers and drastic slow-downs
for small features as well as overhangs as the cause. I am not 100% sure though, why it takes so
much longer. Please leave a comment if you have an idea. Comparing the 247zero to the Prusa MK3, the
differences are quite remarkable. For prototyping it can make a huge difference if you have to wait
minutes instead of multiple hours – at comparable quality. Looking at the relative print times,
we can group the printers into four categories: The last generation, represented by the Prusa
MK3S+, the lower end current gen represented by the resonance tuned BIQU Hurakan running
on Klipper, the enthusiast current gen with the Vorons, the V400 and the X1 Carbon and we
have the 247zero representing “the Next Gen”, which will most probably be – and I say this
with humble confidence – more than two times faster than the current gen. Is that mind
blowing? Yes, definitely, but there’s still a lot to do… that’s why I said “will be”…at least
concerning my 247zero concept. Let’s discuss. At first glance the print quality
of the 247zero is quite comparable. Looking closer we still see slightly wavy
patterns on the surfaces of the prints. The cantilever print bed and the
frame are too wonky in order to support the accelerations and speeds
for that last bit of surface quality. In addition to that: My PET Carbon Fiber 3D
printed X-beam lacks torsional stiffness. This altogether induces tiny displacements while
printing – even at these kind of reduced speeds and accelerations. The good news: There are quite
easy to do fixes for these issues. You’ll see in the next video! I’ll also show some not seen
before solutions for less weight on the moved masses on XY – make sure to ring the bell!
You won’t miss that. Concerning extrusion: At the moment I am evaluating alternative
hotends, nozzles and extruders, because more flow is always more better, right? Also, with
the Hemera slipping at its limits it gave me quite an annoying time. We need more torque
and also higher speed capabilities, because the necessary higher Pressure Advance-values
quickly showed the limits of that bowden setup. You can listen to the step losses here…
The cooling performance of the 247ESPC is absolute overkill in various aspects: The air
flow and cooling power is absolutely sufficient, but the necessary auto-cooling is very, very hard
to dial-in without facing warping – even on PLA. The energy consumption of the four 4028 fans
at full bore is tremendous and worst of all: The noise level is killing me. Being next to
the 247zero and 100%-part cooling at over 90dB forces to ear protection and can be heard
two floors above my printer room. Further downsides of my actual concept and designs:
No enclosure possible and you need a capable external power supply. Let’s fix this! I set
a secret due date for the next video, so let’s see how far I can get concerning the rework of
the 247zero! We will see, because it’s obvious: FDM still didn’t reach its limits, yet…because
DIY hasn’t reached its limits (“SPEED LIMIT”), yet. I’ll also give my best to set my humble marks
in order to close the gap between a three and half minute Benchy Shaped Object and ultra-fast
printing at high quality. With that being said: Thanks a lot for watching, Servus und auf
Wiederschaun! I am not going to tell you that subscribing liking, sharing and commenting
will help and support 247printing. Thank you! Further… further… further…fuuuuurtheeeeer…
And you typically use the max allowed line width. Des hamma doch scho
ghabt? *Childish giggling*