Dynamic load balancing \ force cancelling gantry

Just finished implementing this if anyone is interested [somei3deas.wordpress.com]

Video here if you don't want to read the waffle [www.youtube.com]

---

[somei3deas.wordpress.com]
[www.youtube.com]

Nice! I've been following your blog for a while now, I like the novel ideas, especially the max flow using a diamond hot end with simultaneous filament feed.

Why does it matter if the printer shakes around.
Just so the print surface is in sync with the extruder.

Movement from printer to floor only matters if it is shaking
your house or something next to it that needs to be still.

If you are setting diamonds or painting pin heads on the same
table as printer --- the movement induced to the table by printer is a problem.

Should you be concerned and measure bed to gantry movement and correct for that?

When the print head is close to the bed, and because the bed is rigidly constrained from any side to side or back to back movement by the linear guides, then for sure the amplitude of the oscillations will be the same for the bed as the print head, so in theory it shouldn't matter. However, as the bed moves further towards the floor, away from the print head, then the amplitude of the oscillations for the bed diminishes. At Z max, close to the floor, the bed will have almost zero movement but the print head will still have the same amount of movement. So tall printed objects would likely be tapered - wider at the top (maybe). As I explained in my blog, for decorative objects, that isn't a problem but for things that need to be dimensionally accurate, it is a problem.

Also, this is all assuming that the frame is perfectly rigid. In reality, when throwing in excess of 4Kgs around there is bound to be some flexing of the frame. Whether that is significant is hard to say. I'd need to install the printer on a concrete slab or some other inflexible base to be able to measure it, and for various reasons, that's impractical for me to do. But what is apparent is that the movement I measured isn't simply a deflection caused by the shift in mass, but more of a resonance being set up. So rather than acting like a pendulum, it's more like a guitar string being plucked or a tuning fork.

I plan to do some back to back high speed print tests both with and without the gantry active and see what effect, if any there is on print quality. Hopefully at high speeds, some sort of ringing type artefacts will appear, and also hopefully, these will disappear when the gantry is activated. If not, well it's been a somewhat costly experiment.......

---

[somei3deas.wordpress.com]
[www.youtube.com]

Rolls Royce also uses the same one pound coin technique to check for their engine lack of vibration. But a frame house in the UK ???

Edited 1 time(s). Last edit at 10/07/2018 03:06AM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
MKSA
Rolls Royce also uses the same one pound coin technique to check for their engine lack of vibration. But a frame house in the UK ???

Don't understand that - maybe something got lost in translation. The purpose of the dynamic gantry is to prevent the printer frame from shaking, not the house (which is mostly brick and concrete apart from the upper internal walls and floors - it's not a timber frame).

I seem to be labouring to get the point across but it's this:

If I do a high(ish) speed linear move and measure the frame and floor deflection, it isn't a simple deflection but an oscillation. Some sort of vibration occurs at the end of the move. The print head itself shouldn't oscillate. It starts from rest, accelerates up to speed, then decelerates to a standstill. There is nothing in the firmware or kinematics to make the stepper motors rotate back and forth at the end of the move. One could argue that belt stretch or elasticity is the cause of the oscillations but personally, I don't buy it but at the same time, I don't discount it. Maybe the floor itself acts like a spring, rather than a sponge. Alternatively, the frame itself could be resonating and this resonance gets transmitted back to the floor. I simply don't know. I could probably fit accelerometers to the various elements, or use high speed photography or some other technique to get a more accurate determination of the root cause, but I have limited resources available to me, at home, in my spare bedroom, with only pension income.

If it can be proven that these oscillations, vibrations, resonances (call them what you will) are faithfully transmitted back to both the print head and the print bed, then in theory, there should not be any adverse effects on print quality. But this brings me back to the limited resources for producing accurate, real time, measurements of the oscillations of each and every element of the printer. Personally, I doubt that this faithful reproduction is likely over the full range of Z movement where the 8Kg bed can be either close to the print head or as much as 760mm away from it.

So, that was the rationale behind this. By applying equal but opposite forces to the frame, the root cause is eliminated. Also of note is that if the cause was due to elasticity in the belts, then the oscillations should have still been present after the extra gantry was fitted. But they aren't, which to my my mind, rules it out as a cause.

I have in the past experimented with high speed printing by employing multiple melt chambers - see here [somei3deas.wordpress.com] and here [somei3deas.wordpress.com]. These were just feasibility studies but I plan to look into this again in more depth. I will run back to back tests with and without the dynamic force cancelling gantry to see what effect on print quality this has - if any.

I'll report the results on my blog. I rarely bother posting on these forums, and this thread has reminded me why I made that decision some time back.

---

[somei3deas.wordpress.com]
[www.youtube.com]

Impressive looking results in the video. Is that all official OpenBuilds hardware or chinese equivalent?

Quote
obelisk79
Impressive looking results in the video. Is that all official OpenBuilds hardware or chinese equivalent?

It's all official Open Builds hardware - I've read too many posts on various forums about peoples negative comments using "cloned" hardware which has failed. Most of the Open Builds stuff was bought from a company called "Ooznest" here in the UK but some of the first extrusions I bought were from a company called "RatRig" in Portugal.

The printer has "evolved" into what it is now, over a period of about 2 years but when I add up the cost it starts to get scary. For example, I hate having anything cantilevered, so all the wheels and idlers run on bolts that are anchored at both ends. That means for the wheels I use twice as many spacers, both plain and eccentric. So an average cost for a Delrin wheel, twin bearings, two spacers, nut, bolt and washer is about 10 GBP (the plain spacers are cheaper but the eccentrics are dearer). The real scary bit is when I add up how many I've used - last count (for the 7 axes) is 52 sets. So, I've spent around 520 GBP on bloody wheels!

---

[somei3deas.wordpress.com]
[www.youtube.com]

Coin test: [www.youtube.com]

Cantilever design has some advantage like a more open space but when not necessary should not be used as it requires in fact more material and more complexity to achieve the same rigidity. This is often forgotten by the would be designers of 3D printers.

Edited 1 time(s). Last edit at 10/08/2018 03:44AM by MKSA.

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
MKSA
Coin test: [www.youtube.com]

For info, way back in the 80s, in a previous life, I used to tune engines for a living. That's in the days of carburettors, mechanical distributors, points, condensers, adjustable tappets etc. Even in those days, you could do that coin test with just about any 8 cylinder or more engine (providing it was European and a road car) once it was correctly tuned. Jaguar V12s were particularly easy as were many Japanese engines. Sadly, most UK built 4 and 6 cylinder engines failed miserably. The trick with reciprocating engines is getting everything balanced. So pistons and con rods would be matched and bits shaved off the con-rod here and there to get the weight exactly right. Also, the crankshaft. Rolls Royce would of course do all that but they also used a very heavy flywheel. This is good for damping out any torsional vibrations but the heavy mass does not accelerate so well. That's fine for a Rolls Royce, but for a race car engine, it's very bad. Hence racing car engines have a very small, low mass flywheel which can be accelerated very quickly (but you can't stand a coin on them).

So the coin test is nothing new to me and the reason why I used it as a demo.

---

[somei3deas.wordpress.com]
[www.youtube.com]

i wonder if the pendulums they use in skyscrapers could be used in deltas. imma have to go get a nice size ball of lead...

thanks for the ideas!!!

Hey
a couple IMUs, an arduino , little code and a PC
would get you 3 axis measurement system
for your printer testing

three-axis gyroscope + triaxial accelerometer three-axis magnetic field
2 or more IMUs can be connected to bus on arduino

9 Axis IMU L3GD20 LSM303D Module 9DOF Compass Acceleration Gyroscope

[www.ebay.com]

Quote
cozmicray
Hey
a couple IMUs, an arduino , little code and a PC
would get you 3 axis measurement system
for your printer testing

three-axis gyroscope + triaxial accelerometer three-axis magnetic field
2 or more IMUs can be connected to bus on arduino

9 Axis IMU L3GD20 LSM303D Module 9DOF Compass Acceleration Gyroscope

[www.ebay.com]

One can use their smartphone. There are a few application for it. Look for accelerometer. Just that the mass of the phone is a lot more and not as easy to mount than just the sensor !

---

"A comical prototype doesn't mean a dumb idea is possible" (Thunderf00t)

Quote
warbunniex
i wonder if the pendulums they use in skyscrapers could be used in deltas. imma have to go get a nice size ball of lead...

thanks for the ideas!!!

Oh trust me, I looked into Tuned Mass Dampers (that's what they are called). Relative to the mass of the structure, the mass of the damper is small. The problem is that they only really work at a specific frequency - usually the resonant frequency of the building. The tuning comes from the mass of the "weight", the length of the pendulum, and the damping factor (you need some sort of motion damper between the pendulum weight and the frame). The problem with applying it to printers is that the frequency can vary over quite a large range. The printer will tend to oscillate quickly when doing short infill moves in corners, compared to long perimeter moves. But you might find something that would work for the worse case scenario. I have read a couple of posts on various forum from people who claim to have seen improvements simply by add a large weight to the top of their Delta frame. I have no idea how effective that might be as I don't have a Delta to play with.

---

[somei3deas.wordpress.com]
[www.youtube.com]

This technology would be better used to pull cashpoints out the wall.

Quote
MechaBits
This technology would be better used to pull cashpoints out the wall.

Thanks man. It's always nice to know that the hours I spend experimenting and developing printing techniques and hardware, then writing up my findings, publishing them on a blog which I maintain at my own cost, doing videos, also at my own cost, all in the hope that others might benefit, is being appreciated.

I get absolutely nothing out of this stuff. Just a bit of satisfaction that I might have done something which will benefit others.

As I said, above quote.............

"I rarely bother posting on these forums, and this thread has reminded me why I made that decision some time back."

---

[somei3deas.wordpress.com]
[www.youtube.com]

Lighten Up.... The Machine Looks Great...a real beast! If you make no money for all your work then my suggestion(joke) is a valid one, (if not totally unpractical I know you will never wheel it around the streets looking for a victim).

What would you rather I said?

By the way great write up on the blog, I did drop by to have a deeper look, it was well documented,
I'm sure I'd learn a lot from reading more on the blog.
(I really wish I could be arsed to do what you have done)

Too much tinkering makes for dull boys...Once again perhaps I have to add some context...

[www.youtube.com]

Your A bit of a comedian yourself eh...this line had me rolling on the floor for hours...

'So, I've spent around 520 GBP on bloody wheels!' I've been balking at the cost buildup with 4 wheels.

Which would last longer wheels or linear rails?

Edited 6 time(s). Last edit at 10/11/2018 06:21AM by MechaBits.

@ MechaBits. Apologies, I read your comment at a bad time. You would not believe some of the sh*t comments I get on my blog. Fortunately, one advantage of having my own blog is that I have full moderator rights and have it set up so that I get to review any comments from new visitors, before they get published. I'd just deleted comments and barred a particularly obnoxious a***hole before I read your post, so I kind of over reacted - sorry.

Anyway, ref those extortionately expensive Delrin wheels, I have no idea whether they would outlast linear rails or otherwise. All I can say is that I've been throwing around 4Kgs at highish speeds for a couple of years with no sign of wear on them, nor any need to re-adjust the pre-loading. Oh, and they are silent. The reason I went that route is that the frame is all OpenBuilds V slot, therefore any frame member can also be a linear guide. So in that respect, it seemed a bit daft to fit linear rails on top of what are already linear rails, if you get my meaning.

Edited 1 time(s). Last edit at 10/11/2018 09:35AM by deckingman.

---

[somei3deas.wordpress.com]
[www.youtube.com]

ok cool no hard feelings...re linear rails on vslot...I just did it, so you can have a little laugh there...
and on a cantilever, so extra giggle for you there too...
Now if I spent a £10 on wheels I could do away with the 2 linear rails and all the bits holding it in place,
but I wanted to try out something different, though it would be lower, less parts/cost and possibly more sturdy with wheels,
I was thinking about the harder ones, incase I want to use it on cnc

Edited 1 time(s). Last edit at 10/11/2018 12:00PM by MechaBits.

Are you using seperate motors for the load balancing, done in opposite direction, it's a pity weights couldnt be controlled by the backside of the belt, ie the bit that moves in the opposite direction, but that might interfere somwhere so obviously something that needs to be done from the gound up.

Yes, completely separate gantry with the motors mapped to the existing CoreXY axes but direction reversed.

Another crazy idea I had was some sort of rod attached to the X carriage, sticking up vertically but pivoting about a point central to the bed. Kind of like a plate fixed to the top of the printer or some way above, with a hole in the middle, through which the rod passes. Then a big weight on top of the rod, equal to the weight of the carriage. Essentially, the weight would waggle about in the opposite direction to the carriage. But then sanity took a hold and I realised that because of the XY distances, the pivot point would need to be somewhere in the attic with the weight outside above the roof. That would have been OK but I couldn't figure out how to stop birds from using it as perch which would upset the fine tuning.

---

[somei3deas.wordpress.com]
[www.youtube.com]

You know the other config you where talking about, with the long ofset couter lever thing,
I'd been thinking of a similar thing but different maybe you know of an exaple where it;s been used?
It sort of cross between what you said & an ultimaker XY,
where head has 2 rails on X & Y so more than ultimaker, but the head sits in the middle, and is not controlled by the belts
but by the rails position and the head floats/constrained between both sets of rails.
Seen anything similar?(obviously this doesnt do what you need here)

Edited 3 time(s). Last edit at 11/02/2018 12:26PM by MechaBits.

Haven't seen anything like that. As it turns out, this whole thing has largely been a waste of time because even when the printer is doing it's "magnitude 8 earthquake thing", the prints comes out just fine. So although it does a very good job of stabilising the printer, there is nothing to gain by doing so. My own fault - I fell into the trap of finding a solution to a problem that doesn't exist and I should know better.

---

[somei3deas.wordpress.com]
[www.youtube.com]

well you should be able to make another printer pretty quick from the donor parts.

So what, ideas need to be turned into reality to gain new knowledge, who cares if reality ignores them
If i would count the number of after all useless ideas i persued i would go mad.

---

[www.bonkers.de]
[merlin-hotend.de]
[www.hackerspace-ffm.de]

Agreed. The reason for posting on these forums was that others might benefit. So the fact that the idea didn't give any improvement to print quality is still useful knowledge. Although it cost me some time and money, at least it might save someone else some time and money if they stumble across this thread at some time in the future. All knowledge is useful.

---

[somei3deas.wordpress.com]
[www.youtube.com]

I'm verry happy with your idea regarding this force cancelling idea. It could be the answer for me stopping my printer from walking off the table

Quote
Mbasecnc1975
I'm verry happy with your idea regarding this force cancelling idea. It could be the answer for me stopping my printer from walking off the table

Someone I know suggested maybe putting casters on the printer, then designing a part to print that would make the printer do a little dance. That'd make a good YouTube video - "The worlds first dancing 3D printer".

---

[somei3deas.wordpress.com]
[www.youtube.com]

Quote
deckingman

Someone I know suggested maybe putting casters on the printer, then designing a part to print that would make the printer do a little dance. That'd make a good YouTube video - "The worlds first dancing 3D printer".

Sorry to say that, but I've beaten you regarding the dancing printer, because I tried caster wheels under my Prusa i3 a few years back. They wouldn't exactly make it dance but you will find out soon, if the floor is level or not...

BTW: The video is great, but IMHO the text scrolls too fast for foreign readers and there is a typo ( qulity )

HTH
Olaf

Edited 1 time(s). Last edit at 12/09/2018 12:33PM by o_lampe.

Quote
o_lampe

Quote
deckingman

Someone I know suggested maybe putting casters on the printer, then designing a part to print that would make the printer do a little dance. That'd make a good YouTube video - "The worlds first dancing 3D printer".

Sorry to say that, but I've beaten you regarding the dancing printer, because I tried caster wheels under my Prusa i3 a few years back. They wouldn't exactly make it dance but you will find out soon, if the floor is level or not...

BTW: The video is great, but IMHO the text scrolls too fast for foreign readers and there is a typo ( qulity )

HTH
Olaf

Thanks for the feedback - I'll bear that in mind for future videos. I could narrate the videos instead of using scrolling text but I hate the sound of my own voice.

Also, yes I noticed the typo after I'd rendered and uploaded it (my video editing package doesn't do spell checking). I was just too lazy to correct the mistake then re-render and re-upload it all over again.

---

[somei3deas.wordpress.com]
[www.youtube.com]