QU-BD Revolution XL Speed Test 9000mm/s Accel @ 1000mm/s

This is a quick video of our machine aluminum Revolution XL printer doing some speed tests. The production version will have cable management in our Nanolinks cable chain and have bead blasted anodized parts instead of just raw machined parts as in the video. We are running Azteeg electronics from Panucatt with Texas Instruments based stepper driver (which we like A LOT better than the 4983/8s). All the electronics and power supply are tucked up underneath the machine.

Revolution XL Speed Test ( Youtube )

Chelsea - QU-BD

I took the freedom to put a ruler onto this video. Assuming the motors are NEMA17 (that's about the only size reference there) your build envelope is 220 x 220 mm (XL = extra large?). The X carriage takes 11 frames for moving these 220 mm, the video is 24 fps, so the movement takes 0.46 seconds.

220 mm / 0.46 s = 478 mm/s. Half of what you claim and much closer to what's achievable elsewhere. Now, which extruder can keep up with these 478 mm/s?

BTW., where's the extruder in this video? Extruder weight reduces acceleration.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Traumflug Wrote:
-------------------------------------------------------
> I took the freedom to put a ruler onto this video.
> Assuming the motors are NEMA17 (that's about the
> only size reference there) your build envelope is
> 220 x 220 mm (XL = extra large?). The X carriage
> takes 11 frames for moving these 220 mm, the video
> is 24 fps, so the movement takes 0.46 seconds.
>
> 220 mm / 0.46 s = 478 mm/s. Half of what you claim
> and much closer to what's achievable elsewhere.
> Now, which extruder can keep up with these 478
> mm/s?
>
> BTW., where's the extruder in this video? Extruder
> weight reduces acceleration.


I would tend to agree on the speed, i don't think many of the mainstream hotends would keep up to even half of that, i've got a hotend here destined for use with delta bots but i doubt even that would keep up, unless you only use the high speed for travel and do the actual printing at a much lower speed,

---

-=( blog )=- -=( thingiverse )=- -=( 3Dindustries )=- -=( Aluhotend - mostly metal hotend)=--=( Facebook )=-

The issue is even at 9000 mm/s2 you'll never hit 1000 mm/s in a move inside the bed volume.
Still pretty quick though.
The accelleration is more interesting than the speed setting IMO.

Sorry I wasn't a bit more clear, the extruder is mounted in the video, its just an all new design that is unlike anything currently on the market. It is geared, weighs less than 6oz including the motor, its all metal and it can keep up

The X-Axis is 232mm in length.

Also...and I'm going to check with Nathan on this, since he would know for sure, but if you are calculating speed by timing the movement over a given distance, wouldn't that be the average speed vs. the top speed it achieved during that move? When I calculate the average speed I get the same result as you (basically except for the 232mm instead of 220mm).

Chelsea - QU-BD

Edited 1 time(s). Last edit at 02/03/2013 12:37PM by Chelsea - QU-BD.

Chelsea - QU-BD Wrote:
-------------------------------------------------------
> Also...and I'm going to check with Nathan on this,
> since he would know for sure, but if you are
> calculating speed by timing the movement over a
> given distance, wouldn't that be the average speed
> vs. the top speed it achieved during that move?
> When I calculate the average speed I get the same
> result as you (basically except for the 232mm
> instead of 220mm).

Yup, exactly. His method calculates average speed, not peak speed.

From the equation s = u . t + a . t^2/2 , if accel is 9000mm/sec then speed of 1000mm/sec is reached after approx 55mm starting from rest, according to my calculation.

Therefore if nominal speed was achieved, I would expect to see a trapezoidal profile, of about 0.11 sec accel, 0.12 sec at 1000mm/sec then decel to zero of 0.11 sec. That adds up to about 0.34 sec, which is a little quicker than observed.

To measure accurately you would need to put an encoder or scope on the steppers and look at the actual rate.

bobc Wrote:
-------------------------------------------------------
> Therefore if nominal speed was achieved, I would
> expect to see a trapezoidal profile, of about 0.11
> sec accel, 0.12 sec at 1000mm/sec then decel to
> zero of 0.11 sec. That adds up to about 0.34 sec,
> which is a little quicker than observed.

There's an error of +/-.09s, since we do not know when between frames the motor stopped. I'd say they're telling the truth, or at the very least, very close to it.

Of course, without a extruder than can do 1000mm/s, it is a moot point. I can't get mine over ~200mm/s.

Traumflug Wrote:
-------------------------------------------------------
> Now, which extruder can keep up with these 478
> mm/s?

That's the question. Interesting achievement though if it "can keep up" as claimed, given the performance of the current QU-BD extruder design.

A video with the extruder actually extruding wouldn't harm the extruder's reputation I believe.

---

--------------------------------------------------------
3D extruder performance tests
dzach's ORDbot Hadron build

Chelsea - QU-BD Wrote:
-------------------------------------------------------
> Sorry I wasn't a bit more clear, the extruder is
> mounted in the video, its just an all new design
> that is unlike anything currently on the market.

In what way is it unlike anything? Seems pretty much like every other extruder anyone's got right now. The only potential innovation is the claim at "all metal", but I know there have been several others who also made all metal extruders for their machines, to no great increase in quality/throughput.

It doesn't matter if the extruder is made of adamantium the real sticking point is in the nozzle and how efficiently it conducts heat into the plastic at high speeds.

You are pushing around three stepper motors on that axis. I can't help but think if you just set up an H-bot belt layout you can get much better speed for the amount of motors you have. Just mount your four axis motors (two Y, two X) at all four points of the H, leaving only four pulleys moving along with the extruder. With no increase in parts won't that simple upgrade greatly increase your speed? And because your system is made of metal it is ideally suited for the H-bot, as a metal frame gives it minimal desire to flex under diagonal motion. Why stick with the motor-lugging design you have now?

In agreement with Destroyer, and in-line with that never ending argument between easy-to-build and high quality machine.

It seems many peoples new designs are not taking advantage of the ideas that seem to be very proven-to-work.

1. and 2. The lack of Bowdens on most of our machines for one, again in agreement with Destroyer, and lugging extra mass around by moving motors doesn't seem like the soundest machine design.
People have their reasons against bowdens, but take a look at the machines that have the highest quality prints and tell me what they all have in common?

3. Bushings instead of bearings. Less weight, less slop, with a proper fit/proper materials the friction is minimal.

4. Belts. Tight belts restrained from flopping are extremely accurate. I've seen machines my whole life using timing belts for some very precise movement. (Personally I've been strongly considering modifying my prusa to use belts on Z)

The advantage of a Bowden cable designed extruder is the low mass which allows fast and controlled acceleration and deceleration. The disadvantage is stringing and oozing. If you could have a direct or geared extruder that was as light as a Bowden cable setup or close to it...that's what we did. At less than 6oz of total weight which includes the motor we can get the same acceleration figures (if not significantly faster) as an Ultimaker.

I will agree with you on the bushings, we were actually planning on switching to them at some point.

In regards to the motor lugging design, it works. Its fast, it accelerates extremely quickly. Could we get higher speeds with an H-belt, maybe, but the same could be said if we mounted the 22kg*cm NEMA 23 motors that we are using on our RPM and moved the belts around. Overall footprint was a design constraint as is cost in order to target a certain price point; the Revolution is only about 10"x10" on the exterior. We wanted to create a matched system, the printer is only as fast as the slowest component so there wasn't any need to go faster than 1M/s @ 9000mm/s accel.

Chelsea - QU-BD

Chelsea, you can defend your design until you're blue in the face and people will still find things to nitpick. The best way to shut 'em up is to post prints from it. Output quality (and print time) are really all that matter.

Quote

3. Bushings instead of bearings. Less weight, less slop, ...

I find exactly the opposite. Bushings needs some play and it rapidly gets worse. Bearings have pre-loading so stay tight until the point they wear out.

---

[www.hydraraptor.blogspot.com]

Chelsea,

putting the crazy high speed aside, and focusing more on the extruder...
- is it still using that filament drive gear as on the MBE extruders ?
- is there a spring loaded idler mechanism yet ?
- is there a "flat" on the tips of the nozzles?

---

-=( blog )=- -=( thingiverse )=- -=( 3Dindustries )=- -=( Aluhotend - mostly metal hotend)=--=( Facebook )=-

In regards to the 'X-Truder'...
-No the cold end is completely different and totally not compatible with the current MBE extruder as the motor size is different.
-No, there is no need due to the way the filament is fed, it cannot slip, period. We will be carrying ProtoParadigm filament in our store that we will be recommending, although other manufacturers high quality filament will work just fine. We will be posting a list of all the filament manufacturers that we have verified are 100% compatible.
-There aren't any changes to the nozzle at this time, it works well in its current state, although we have been experimenting with a stainless steel version that can tolerate a little more abuse.
-It is significantly smaller at roughly half the size of the MBE extruder
-The only downside to the Revolution's/RPM's extruder is that it costs significantly more to manufacture, when sold a la carte it will be significantly more costly than the MBE
-The design is open source

Chelsea - QU-BD

let's see it print.

Quote

Bushings needs some play and it rapidly gets worse. Bearings have pre-loading so stay tight until the point they wear out.

Maybe this is a bit off topic here: Zero play doesn't guarantee most accurate movements. In fact, zero-play bearings can, depending on the situation, have more inaccuracies than bushings with some play when forces are applied.

The unfortunate thing about bearings is, forces are taken by a few small spheres only, so they sink into the rod when forces are applied. With bushings, the contact surface between bushing and rod is much larger, so no sink-in happens. We talk about micrometer or even smaller ranges here.

Also, bushings need an adjustment mechanism to re-tightem them as wear grows.

This is more food for thought than a clear decision help. As always in life, bushing vs. bearing is a trade-off.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I think that most of the people who participate in the designs of 3d printers can make a machine that whips around like that.

What is missing though, is a video of it actually printing something, and a close up of the quality of the final product (along with some verifiable proof of accuracy).

Were you afraid to demonstrate those things?

I find it funny how there isn't a single positive comment in this thread. The closest is that the machine is "still pretty quick though".

Instead, people purport that since it isn't printing anything, it's useless.

Obviously the lesson here is that one should never post works in progress, as people only care about the end result.

Quote

I find it funny how there isn't a single positive comment in this thread.

We're used to announcements which praise the best invention since the introduction of sliced bread If somebody claims to have substantial better results than others there has to be a reason, because others aren't stupid either. And the video fails to explain that reason, exaggerates a bit ... the usual stuff

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Maybe I'm misreading but are you looking for a reason that it is moving quickly or a reason that we posted the video?

We were happy to acheive that speed and acceleration while still carrying a full extruder with motor around. I have seen other printers move that quickly (OrdBot / Ultimaker) but not while carrying an extruder and in the Ultimakers case, not without a huge amount of vibration. We achieved this by using custom premium linear bearings, custom premium chromed and ground to 9RMS case hardened smooth rods and custom dual motors on each axis, which when coupled to the Texas Instruments based stepper drivers are smooth as silk. There is no exaggeration in the video, the math is sound.

Chelsea - QU-BD

Oh wait! That speed would be great when doing a 200mm bridge

Yet you still avoid showing it printing, or a print it has done, or even mention it's printing abilities. Strange.

Quote

Maybe I'm misreading but are you looking for a reason that it is moving quickly or a reason that we posted the video?

More for a reason on how your claims/advantages are achieved. How do you manage 500+ mm/s with an AVR based electronics. Other AVR based electronics (same CPU!) max out at some 200 mm/s unless they reduce microstepping. In other places I found the Revolution XL model described as having approx. 350 mm * 350 mm build size, the Revolution XL in your video has about 30% less. It's this type of inconsistencies which rings bells here.

You explained some of them, so let me say thank you for that.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Please send me to where you saw the 350mm build size, that it an error and needs to be corrected.

There might be a little confusion since we have three different printers being sold simultaneously. The RPM does have a 332mm long axis?

Revolution (small belt driven)
[store.qu-bd.com]

Revolution XL (large belt driven)
[store.qu-bd.com]

R.P.M. (Revolution Prototyping Mill) (HUGE and ball screw driven)
[store.qu-bd.com]

Chelsea - QU-BD

Chelsea,

I've always wondered: how do you pronounce your company name?
(I tend to say Q-Bid)

Dave

'Cubed'

Chelsea - QU-BD

Quote

'Cubed'

Chelsea - QU-BD

[i.imgur.com]

Quote

Please send me to where you saw the 350mm build size, that it an error and needs to be corrected.

Can't find it any longer, sorry. It was a quick google only.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY