looking to build new printer, what sliding mechanisms should i use?

i am looking for a 200x200x200 build platform. maybe 300x200x200.
i am trying to decide on a linear motion system. i like the use of spectra line.

1. hwin track system
   A very good system, the bearing units might be a bit heavy compared to others. very expensive.
2. 20x20 v-slot with their carriage system
   at $5 per wheel + spacers can get expensive fast. i have no idea how it performs
3. m8 hardened steel rods with lm8uu's
   lm8uu can be noisy and a bit sloppy. the hardened shafts are expensive.
4. m8 stainless steel rods with brass bushings or lm8uu's
   as above, but the stainless is much cheaper, lm8uu's will make the rods after a while of use.
5. m8 stainless steel rods with plastic bushings(hpde or other sliding materials)
6. 20x20 v-slot with bushings(hpde or other sliding materials)
7. 20x20 t-slot with bushings(hpde or other sliding materials)
8. 20x20 box with bushings(hpde or other sliding materials)

i have no idea how well the bushings work. they seem like a good idea. i haven't seen enough on them.
i am trying to figure which system would work best for the xy gantry.

i am planning to use the e3d chimera(dual nozzles) with bowden extruders. maybe have a nema 14 extruder for flex filaments. i want to have a enclosed build chamber with heating.
i am also looking for other designs i could use.

i have read through after hbot go xy and i like his designs but i want to try using spectra line first

Edited 3 time(s). Last edit at 09/15/2015 07:06AM by bradleyk.

So this is a real "Shot in the dark" build?

So using spectra line
--- there will be no known movement relationship between motor and extruder.
Do you intend to put on sensors / encoders to measure where things are positioned.
With timing belts --- motor moves X steps position changed by Y distance all the time!

Are 1 thru 8 choices?

You say use Hiwin linear slides
but
then talk about lu8uu bearings on rods

one or the other?

What are requirements -- resolution, accuracy, speed
they will show the best solution.

Bowden extruder --- good luck --- when all your hair is gone from frustration -- you will use a direct extruder

Quote
cozmicray
So this is a real "Shot in the dark" build?

i'm not sure i get your meaning. i have built a working delta printer but i made few mistakes with my first build, looking to fix them with a second build

Quote
cozmicray
So using spectra line
--- there will be no known movement relationship between motor and extruder.
Do you intend to put on sensors / encoders to measure where things are positioned.
With timing belts --- motor moves X steps position changed by Y distance all the time!

spectra line works the same as timing belts, assuming they don't slip. which will determine the max accel ratings

Quote
cozmicray
Are 1 thru 8 choices?

You say use Hiwin linear slides
but
then talk about lu8uu bearings on rods

one or the other?

What are requirements -- resolution, accuracy, speed
they will show the best solution.

yes they are choices, i want to be able to use this to create engineering prototypes not sure on exact requirements,

Quote
cozmicray
Bowden extruder --- good luck --- when all your hair is gone from frustration -- you will use a direct extruder

i am currently using a bowden, and i am having no issues with it.

cozmicray is right that as long as you do not have any specifictions for what you want to make or an actual budget, it is really hard to give good advice.

As to the essence of your question - I think few of us have tried all the options you list, I for one have not, so it is really hard to give strong input. With linear slides the quality, I believe, is proportional to the money invested. But if you do not invest in a stiff frame then you lose it all on that, and in the end Chines linear ball bearings or Hiwin slides makes no difference.

What really confuses me, is the spectra lines. I do understand and like the concept. But spectra lines is a cost reduction method. The idea, along with the idea of bushings directly on extruded alu, comes from the quest to lower printer post. Not something you do to get engineering prototypes.

Try to look up corexy on google and thingiverse and look at all the printer designs. Then think about which of the builds out there you would like your future corexy to look like. That could be a good place to start.

Thank you for your response.

Quote
LarsK
cozmicray is right that as long as you do not have any specifications for what you want to make or an actual budget, it is really hard to give good advice.

i don't have a budget as in i am trying to find what cost can be reduced and still keep a great printer.

Quote
LarsK
As to the essence of your question - I think few of us have tried all the options you list, I for one have not, so it is really hard to give strong input. With linear slides the quality, I believe, is proportional to the money invested. But if you do not invest in a stiff frame then you lose it all on that, and in the end Chines linear ball bearings or Hiwin slides makes no difference.

i plan to build the frame out of 20x20 extrusion in the hopes it will be rock solid. i have done some readings on people using linear slides and they prefer them over lm8uu's but there doesn't seem to be enough on them for me to come to a decision.

[forums.reprap.org]

Quote
LarsK
What really confuses me, is the spectra lines. I do understand and like the concept. But spectra lines is a cost reduction method. The idea, along with the idea of bushings directly on extruded alu, comes from the quest to lower printer post. Not something you do to get engineering prototypes.

i currently have spectra on my current printer and would like to reuse most of the parts, i am not opposed to buying belts just don't want to waste my current parts. if i need to upgrade my printer to belt's later that wouldn't much of an issue as i should be able to reuse most of the parts.
with my current delta the spectra is working great, i just didn't make it large enough or stiff enough.

I am using the maker slide on my corexy with spectra line. I am having no issues with the xy movement. If you look up my post below this you will see some pictures. If you want a video let me know and I will upload one or two.

In my corexy the x, y linear motion system is made with LM8UU and m8 hardened steel rods. The result is a noisy movement with some vibrations, but the print quality is still very good.
I have now ordered igus plastic bushings to replace the LM8UU.

Hi,

2020 are small extrusions,
the v-type can make good guides
but for structure 2040 should be better
if oriented in the proper direction.

If you want a good printer, don't be cheap.
Each dollar saved may complicate the building/calibration process,
and probably lower the print quality.
Use the right component at the right place.
The better component, the better prints.

I tested Hiwin, LM8UU, Igus, V-slots...
Definitely linear guides like Hiwin are the top.
V-slots work pretty well and silently with 625RS and derlin wheels.

Bronze and polymer bushings works great and silently too,
but 8mm smooth rods are not as stiff as you think.
You want to raise in diameter to get a nice motion at reasonable speed.
10 to 12mm should be enough for 350mm rods.
LM8UU are the worst just in front of tri standard radial bearing guides.

GT2 belts and pulleys are not expensive,
easy to use and works first time, any time.
Don't ever think to save more than $10 using fishing line.
That less than $10 will save you hours
and worth its weight of wasted plastic.

++JM

PS. I used both direct drive and bowden extruders.
I prefer bowdens because they allow to print faster
with at last the same quality than direct drives.
I manage to print flexible filament. There's no issue
if you print slowly and if your filament is well guided
from start to end. It's just about understanding bowden
and be patient during calibration time.
Once calibrated, it's nothing but fun.

Edited 1 time(s). Last edit at 09/16/2015 06:44AM by J-Max.

First, J-max as always says intelligent things and I agree with all - except the use of 2040. For your printer size (300x200x200) I am sure you can make it stiff enough with 2020.

Anyway - I wanted to put in a " not so technical " point. Simply - there is a lot of fun/challenge in building low cost printers .Trying to make bushings directly on the alu ( as you posted) and with spectralines is just as much a challenge as getting expensive linear rails to run 200 mm/s.

So, just remember that you need proportions in your build. Don't buy expensive linear slides if you won't spend the money for steel corners on your alu frame. Don't buy Duet electronics for cheap Chinese linear ball bearings - but do for hispeed printer builds.

If you say A you need to say B and even as it may be tempting to go for top of the line single product's, it is the combination that makes the printer.

What I want to say, if you go for spectra lines and reuse of old parts and parts printed with a not 100% printer then stick with linear rods and bushings or run directly on the alu (don't buy the cheapest Chinese linear ball bearings no matter what)

In the end corexy is very forgiving if you just reduce the speed. Quality impact of low qual parts are significant smaller then what I experience on my delta printer. On corexy reducing the speed almost always helps. Exception to the rule is the z axis which is vulnable to wobble no matter speed.

Quote
bradleyk

Quote
LarsK
What really confuses me, is the spectra lines. I do understand and like the concept. But spectra lines is a cost reduction method. The idea, along with the idea of bushings directly on extruded alu, comes from the quest to lower printer post. Not something you do to get engineering prototypes.

i currently have spectra on my current printer and would like to reuse most of the parts, i am not opposed to buying belts just don't want to waste my current parts. if i need to upgrade my printer to belt's later that wouldn't much of an issue as i should be able to reuse most of the parts.
with my current delta the spectra is working great, i just didn't make it large enough or stiff enough.

Why are you taking apart your working printer ? How will you print new parts if you find that you need something after assembly? Why not just bite the bullet and spend those extra 100 USD to get all new parts and start from scratch?

Hi,

Maybe I accidentaly say things more intelligent than I am.

I bought for the first time 2020 profiles for my CoreXY project.
The bed is 250x220mm, I know this is not standard,
but dimensions comes from parts laying around in my shop.
Honnestly, I thought that profiles was better.
And the hardware corners not as squared as expected too.
I had too much vibrations and profiles bended too easily IMO.
I replaced them and wish to have bought stronger profiles at the start.

You're right, some may have different ways to enjoy their build.
I respect those who enjoy most the experimental way.
I prefer spend some time using the tool than working for it.
But I appreciate everybody's probably not like me.
As 2020 was not satisfactory to my eyes,
I guess 2040 should be correct, especialy for the rails,
adding the expected stiffness to the whole structure.
Maybe 2020 are enough for the stiles.
If you don't expect as much as I do,
maybe 2020 are enough for the whole structure.

++JM

The F306 from Fusion3 uses spectra line, and is considered one of the best CoreXY printers available, so I wouldn't knock spectra.

They're also using carriages on extrusions.

Another printer worth looking at is the C-Bot by Carl Feniak-- uses belts and a cantilever bed, but also uses carriage-on-extrusion.

Unless you're going over 500mm build area, I wouldn't think you'd need over 2020 extrusions. Carl used 2040, but he also routed the belts through the slots in the extrusions. With spectra, you could run the spectra through the slot of the 2020 and do a crossover without any issue.

Please post links to Spectra line printers
and
How they do the accuracy and precision.

How one powers the line (pulley, capstan)
--- it must be a Pull - Pull system, and it must go around a motor shaft.

How is positional accuracy maintained?

Tnx

Quote
J-Max

GT2 belts and pulleys are not expensive,
easy to use and works first time, any time.
Don't ever think to save more than $10 using fishing line.
That less than $10 will save you hours
and worth its weight of wasted plastic.

++JM

For this build money isn't really an issue it will just take me longer to finish the build, i just prefer to not over engineer, i like to look and idea of spectra line as well as it will keep a cleaner build.
If i need to reprint later to upgrade and use belts just don't say i told you so

Quote
J-Max
I manage to print flexible filament. There's no issue
if you print slowly and if your filament is well guided
from start to end. It's just about understanding bowden
and be patient during calibration time.
Once calibrated, it's nothing but fun.

That is great i will have to order some flex filament on my next order and give it a test. everywhere i have read is don't do it.

Quote
LarsK
First, J-max as always says intelligent things and I agree with all - except the use of 2040. For your printer size (300x200x200) I am sure you can make it stiff enough with 2020.

things like the alu extrusion i can't easily and cheaply upgrade, i will follow J-max and make the top supports 20x40, it doesn't cost much to upgrade to it from 20x20 and will help to make the printer squarer.

Quote
LarsK
What I want to say, if you go for spectra lines and reuse of old parts and parts printed with a not 100% printer then stick with linear rods and bushings or run directly on the alu (don't buy the cheapest Chinese linear ball bearings no matter what)

Why are you taking apart your working printer ? How will you print new parts if you find that you need something after assembly? Why not just bite the bullet and spend those extra 100 USD to get all new parts and start from scratch?

I do plan to make a solidworks model of every part first so i don't run into that issue. i understand where you are coming from, if i need to i can order those parts later, i am thinking of upgrading to a new control board and i will be getting a 32bit board. i will have to make a post to see which one would be best to get.
At this stage i will be salvaging the electronics and hotend.

so far from this thread i have decided to use v-slot with their carriage system
i like designing in solidworks, ordering , then redesigning when i have the parts in my hand

that F306 look's nice. that is kinda what i was going for. but i was hoping to design it with the motors inside the frame to keep it cleaner.

Quote
cozmicray
Please post links to Spectra line printers
and
How they do the accuracy and precision.

How one powers the line (pulley, capstan)
--- it must be a Pull - Pull system, and it must go around a motor shaft.

How is positional accuracy maintained?

Tnx

have a look at richrap's delta this is one design to keep the positional accuracy
but just having tension on a spool with a few wraps will not slip.

Hi,

Quote
bradleyk
so far from this thread i have decided to use v-slot with their carriage system
i like designing in solidworks, ordering , then redesigning when i have the parts in my hand

Save time and download solidworks parts @ [openbuilds.com]

++JM

Thanks, that is a nice find.
I will have a look at making a quick design tomorrow.

i have done a quick mock up, i haven't started the z axis. i am thinking of using a cantilevered design using spectra to keep it level



here is the design in step

Fine first draft.

I think the way you use the 2040 now does not give you much stiffness. The problem is how the vertical profiles are secured to the top and buttom structure.

Using 2040 in the lower rectangel will give you two more points to secure the vertcal profiles.



Basically that is the 3 base designs (and obviously combinations of these). Since you are already going for the 2040 on top I think you should do the same below.

How do you intend to fabricate your parts? For example for the carriage? Are those all printed parts?

Hi,

I agree with LarsK, and would prefer the A solution.

++JM

yeah i didn't like the way the profile are attached at the top, but i wanted to run the spectra line inside the joins.
thank you, i have updates the design now. using the 2040 below do you think i will need to use the corner brackets in the bottom? it would look nicer without them. and make a door / heated chamber easier to install.

yes i plan to use printed parts for the carriages



Step

Edited 1 time(s). Last edit at 09/18/2015 06:41PM by bradleyk.

z axis


i plan to use spectra line to hold the platform level. i will be using it on both sides.
the z axis rails and ball screw are a bit long as they are leftovers from my cnc upgrade
they are 20mm rails with a 16mm ballscrew

i just need to decide if i want to make my printer 450mm tall to have a total z height of 300mm
i think that will be a yes.

reinforced frame


Edited 5 time(s). Last edit at 09/19/2015 09:26AM by bradleyk.

Hi,

An idea you might consider : [www.thingiverse.com]


++JM

Quote
J-Max
Hi,

An idea you might consider : [www.thingiverse.com]

bed2_preview_featured.JPG

++JM

I'm doing this with kevlar line (a bit less give than spectra) and it's working well so far. I put the motor on a length of extrusion in the middle of the uprights to minimise the angle change from winding the line around the drum.

Hi,

Two things:

First for the Z axis I think you can either stick to the more "traditional" bed design. Again, your printer is not super big. Or I think you should go all the way with the spectra wire.

Imagine if you put a stepper on that spectra wire and then you can just remove the ball screw completely. Never tried anything like it but would be really interesting. Ball screws are really what kicks the printer costs up.

Whatever design you go for, you need to think about the bed wobble. The thing is you only put one bearing on each rod.



Look at the left side first. I try to sketch a design like yours moving downwards in a situation where the friction of one bearing is higher then the other. Then you risk that the bed skews like shown on the stippled lines.

Now look at the right illustration. This is what I call the "traditional" bed design. The good thing about that, is that there is a static load on each bearing. The static load is the reaction forces from the moment about the rod caused by the weight of the bed. This means that one side of the bearing will be in full contact all the time - unless bed excited by a side-verse or upwards force greater then the bed weight but that should not happen. So you have a very defined load case with good contact points. The downside of this, is that the construction needs to be pretty stiff so that it does not flex.

I am not sure how your spectra lines will play into all this. I don't have a feel for how elastic it is going to be and with all those pulleys. But I am pretty confident that you, no matter what, will need two bearings on each rod.


The second thing, about your current structure with the 2020 and 2040.

For the lower rectangle I think you have it very well supported now. For connecting the vertical to the horizontal 2040 you can use plates like these:



You can also print something like that but with angles implemented. I am not so sure you need the 45 deg bracing if you do that, but it sure wont hurt and you can always remove (or implement) it later.

What I am concerned about is the upper rectangle. In particular what I am trying to sketch here:



I know in this sketch I have sketched it as a angle but in fact in your current revision you are using a steel plate - The thing is, the steel plate is even worse. Single bolts cannot take up moments of rotation. It will work like a hinge.

In general you really need to get more stiffness downwards from the upper rectangle . You need some kind of bigger plate. That plate needs to be really firmly bolted on to that 2040 horizontal and then go and connects with at least two bolts on the 2020 vertical. Or you need 45deg bracing like you did on the lower part.

Quote
FatFreddie

Quote
J-Max
Hi,

An idea you might consider : [www.thingiverse.com]

bed2_preview_featured.JPG

++JM

I'm doing this with kevlar line (a bit less give than spectra) and it's working well so far. I put the motor on a length of extrusion in the middle of the uprights to minimise the angle change from winding the line around the drum.

i did think about using that or a similar design, but thought the ballscrew would give better results.
i might need to have a look at building a test one using that design.
RichRap has done that aswell. but he used a geared stepper. here

the z axis is the one i hadn't found a design that i really like, that design looks interesting, but what stops the front back wobble across the access?

Quote
LarsK
Hi,

Two things:

First for the Z axis I think you can either stick to the more "traditional" bed design. Again, your printer is not super big. Or I think you should go all the way with the spectra wire.

Imagine if you put a stepper on that spectra wire and then you can just remove the ball screw completely. Never tried anything like it but would be really interesting. Ball screws are really what kicks the printer costs up.

do you think the ballscrew increase accuracy? i have the ballscrew from a previous build.
i plan to do some experimenting with the designs. i might also design it without the ballscrew so others could build one.

Quote
LarsK

Whatever design you go for, you need to think about the bed wobble. The thing is you only put one bearing on each rod.

IMG_1700.JPG

Look at the left side first. I try to sketch a design like yours moving downwards in a situation where the friction of one bearing is higher then the other. Then you risk that the bed skews like shown on the stippled lines.

Now look at the right illustration. This is what I call the "traditional" bed design. The good thing about that, is that there is a static load on each bearing. The static load is the reaction forces from the moment about the rod caused by the weight of the bed. This means that one side of the bearing will be in full contact all the time - unless bed excited by a side-verse or upwards force greater then the bed weight but that should not happen. So you have a very defined load case with good contact points. The downside of this, is that the construction needs to be pretty stiff so that it does not flex.

I am not sure how your spectra lines will play into all this. I don't have a feel for how elastic it is going to be and with all those pulleys. But I am pretty confident that you, no matter what, will need two bearings on each rod.

thank you, that is very useful. i might be able to preload the platform with the spectra line but i have no idea how well it will work. if i stay with my current design the 20mm linear bearings are 40mm long and don't have much play but i might look at printing the z axis now and testing it before i get the printer assembled.

Quote
LarsK

The second thing, about your current structure with the 2020 and 2040.

For the lower rectangle I think you have it very well supported now. For connecting the vertical to the horizontal 2040 you can use plates like these:

2015_05_04_13.30.46.jpg

You can also print something like that but with angles implemented. I am not so sure you need the 45 deg bracing if you do that, but it sure wont hurt and you can always remove (or implement) it later.

those 45deg bracing also server a second purpose of giving me somewhere to attach a translucent 4mm acrylic or perspex sheet to create a heated chamber.

Quote
LarsK

What I am concerned about is the upper rectangle. In particular what I am trying to sketch here:

IMG_1701.JPG

I know in this sketch I have sketched it as a angle but in fact in your current revision you are using a steel plate - The thing is, the steel plate is even worse. Single bolts cannot take up moments of rotation. It will work like a hinge.

In general you really need to get more stiffness downwards from the upper rectangle . You need some kind of bigger plate. That plate needs to be really firmly bolted on to that 2040 horizontal and then go and connects with at least two bolts on the 2020 vertical. Or you need 45deg bracing like you did on the lower part.

do you think if i used these in all those corners it will fix the issue. i would love to add 45deg bracing but they will get in the way of the linear carriage.

[www.ebay.com]
also there are printed corners on the inside that weren't in the last pic.


Edited 3 time(s). Last edit at 09/19/2015 07:09PM by bradleyk.

Quote
LarsK
Imagine if you put a stepper on that spectra wire and then you can just remove the ball screw completely. Never tried anything like it but would be really interesting. Ball screws are really what kicks the printer costs up.

I use some cheap spectra line for the Z axis of my smartrap: no wobble, almost zero cost, Z resolution of about 50 microns per full step. That is one problem that I solved completely, among thousands (that will be solved by a new printer).

cool, i hadn't found many people using spectra line for z axis. i will design for spectra tonight.

I also do not use a ball screw. I use belts. As seen below I have have a 2:1 system. When I says 2:1 I mean that the system is geared. So 2 mm turn on the motor gives 1 mm change in bed height.



My experience is that when I just put on the z-axis 1:1 it did often have problems raising and lowering the bed - And that was just my temporary 200x200 bed with 6 mm glass plate. There are many people out there who are doing it 1:1 so it is probably just me and my parts with much friction or something. On the other hand, some people (such as RichRap apparently) put geared steppers - So it might not just be me....

I am not saying that direct spectra (1:1) wont work. It worked for Cristian - But I would probably make some good tests, as you already mentioned you want to. Remember that many of the used stepper drivers can only deliver ~1A power so you reach a point quickly where you can't just go and get a bigger stepper - although where that point is, is still something I need to learn as it is not THAT simple as just saying 1A stepper and done.

The consequence of this all is that as long as you stay with std. electronics you have just 3 options; 1) Buy stepper with gearing already 2) Make you own gearing (like mine) 3) weight reduction and reduce friction in system so the stepper can overcome.

- Regarding using that steel angle for reenforcing the structure - I think it looks like a great idea. That will add a lot of strength. If you later have problems you can reinforce further but for now I think that will be enough.

Quote
LarsK
I am not saying that direct spectra (1:1) wont work. It worked for Cristian - But I would probably make some good tests, as you already mentioned you want to.

In my case, it is (2:1) calculated on the motor shaft (2 motor steps lift or lower Z by the length of one step, where a full circle is motor_shaft_diameter * pi), otherwise the carriage would fall down when the motor is turned off. For my printer spectra line is the simplest solution thanks to the force of gravity and the short length of Z. For other reprap designs, it may work so so or not work at all.

Edited 3 time(s). Last edit at 09/19/2015 10:30PM by cristian.

Christian is also running a 2:1. i will either be using a 2:1 pulley system, or a gear system on the motor. I do have a nema 23 stepper and i am planing to order some dvr8825, which should give me 1.5-2a if i need it.

Spectra line without pulleys is 0.02mm per step, which it too much for the z axis aiming for 0.08mm-0.2mm layer heights. I might have a look at getting some 0.9* steppers for x and y, or just using the 32 microstepping, but then the precision goes down at that level. I might need 32 bit for thoose.

Hmm, honestly had not payed attention that the DRV8825 can deliver so much current. Thought the only real advantage was the 32 microstep instead of the 16.

Still, gearing might not be a bad idea as it is also my experience that it helps you from not having bed crashes on power off / resets.

I was looking at this project:
[www.kickstarter.com]
and came to think of yours. If you see the bed suspension. They use a screw at each point to control the bed. You could do the same with the spectra line. Sort of much like you already sketched in the first one. Looks to work fine for them.

I think something is off if you start needing 0.9 deg steppers. With 16x microstep you should be just fine for resolution.