3D printer build log

[spuddevans]

So I haven't been so active on here for a while, but having started a new project I thought I'd post up a build log for you all to laugh at enjoy 

This will not be a quick build as workshop time is limited, I've already been pecking away at this for over a month to get to where I am today.

For a while now I have had the urge to build a 3d printer, reading around and doing a bit of research I decided on a "CoreXY" style. So the basics of this style is to create a box frame to which the linear bearings and everything else are attached. So a few weeks ago I started by deciding on the outside dimensions, 425mm x 425mm x 500mm high, with the plan that (hopefully) I will end up with a Print capacity of about a 300mm cube.

Over a year ago I had thought about building a 3d printer and had ordered some 400mm x 8mm rods and a bunch of 8mm linear bearings. In my research I found that many found the 8mm rods flex on larger printers, so that made me experiment with deflection and I found that I got about 0.4-0.8 mm deflection on a single 8mm rod with a highly specified weight in the centre (a box of coach bolts). So I got some 12mm rods and linear bearings for the Y axis. I plan on using 3 8mm rods for the X-axis in a "L" configuration, that should minimise deflection.

I have no fancy 3d renderings of what I am building, so that's another reason why this will be a longer build, I will be spending plenty of time thinking before hacking up metal.

So a little more on the design (such as it is). I had a few lengths of 25mm box section steel just lying around waiting to be used, so I cut 8x 375mm lengths for the stretchers top and bottom, and cut 4x 500mm lengths for the 4 uprights. I then set up a vice-stop on the mill and drilled & reamed a 8mm hole through, centered 50mm from one end in each upright. Then I drilled out one side of the 8mm hole to 16mm and then cleaned up the drilled hole with a boring bar to ensure a truly round hole (ended up being about 16.2mm) This was done so that when the whole box was assembled I would have a means of adjusting both the Y-axis bars to get them parallel and flat in relation to each other. I figured I would make bushes for these holes that could be have offset holes in them to carry the Y-axis bars.

It was at this point that I realised that I would need a way of inserting the 12mm Y-axis bars through the frame, so I enlarged 2 of the 8mm holes in the uprights to 12mm.

Then I started to weld up all the stretchers and uprights (after carefully marking each, and also drilling a few M5 holes for potential future use), this called for some thought as the process of welding introduces many stresses and it is remarkable how much it will pull and move parts. I ended up by tack welding only the very outside corners, making 2 sides and then carefully tacked on the stretchers to join the 2 sides into a cube. Then I followed a painstaking process of checking every corner with my 9" engineers square, finding the worst out-of-square culprit and placing a tack weld in the right place to pull it back into square. (it was really surprising how much the metal moves as it cools) Then I would wait until it had cooled before moving onto the next one. Lather, rinse and repeat....... This took quite a few sessions, but was really worth it to get a cubic frame that was less than 1mm out of square in any axis.

Then I made up 3 bushes that were concentric and one that was 0.35mm non-concentric (the error I measured in the holes I previously bored), fitted the arrived 12mm ground bars, and voila (bucket for scale)


Today I started thinking about making sure that the 2 bars were absolutely parallel and also flat with relation to each other. I previously was trying to get them running true in the same plane as the top of the frame, but I realised that the frame has a slight rock to it (about 0.7mm), so it struck me that the most important thing is to get both Y-axis bars true and parallel to each other, everything else can be made true in reference to the Y-axis bars.

But how do you measure any runout when the 2 bars are 400mm+ apart? Then I remembered "Rollie's dad's method" and wondered if it could help in this situation. As I recalled it, Rollies dad's method involved using 2 long bars resting on the ways which would then seemingly amplify any error and make it plainly visible at the 2 ends of the long bars. So I set the frame upside down (the only long bars I had wouldn't fit in the gap above the 12mm bars) and stuck 2 6ft lengths of 25mm box section on the bars and took a few steps back. (parts magnified for clarity) I set it on the seat to prevent me having to crawl on the ground to "sight" along the bars!!



Definitely can see that there is a slight twist to the bars. So I turned up another offset bush to replace a concentric one and tried again. A little better, but not enough, so a 3rd offset bush was made with this result. Much better!!



I made up an extension for my digi-vern so I could check the distance between the 2 Y-axis bars and was very pleasantly surprised to find they were only 0.03mm out!! Here's the easy way I made the offset on my mini lathe, just inserted a shim between one jaw and the work.


Here's an offset bush next to a concentric one


As you can see I have marked the "low" spot on the offset bush, this makes it easy to put a matching mark on the frame so that when re-assembling I can be sure they go back in the right position.

Here's a couple of close-up shots of the insert/bush in place





Next up will be working on the Y-axis carriage's, but I gotta order some ali bar for making those.

Stay tuned for irregular updates.

Tim

[awemawson]

Very nice painstaking work Tim 

I'll be following with great interest - I've been meaning to make one now for several years but it's not risen to the top of the list yet 

I admit I'd have cheated and made the bar mounts adjustable, and tweaked them to plumb 

[spuddevans]

I admit I'd have cheated and made the bar mounts adjustable, and tweaked them to plumb 

I did consider it Andrew, but as the error was quite small it actually worked out quicker to knock out a couple of offset bushes.

Tim

[shipto]

Looks interesting so far, I will be following too. 
I did think until I read the text properly you was intending to make a monster 3d printer  then realised it was that long to check the bars.

[spuddevans]

A little more progress to report.

I've made up a couple of Stepper Motor mounts to be welded on to the frame, but I'm glad that I didn't weld them on as I've changed my mind on how I'm routing the belts ( I may change it again as the project progresses )

But I acquired a short length of 37mm square ali bar which I hacked off a couple of shorter lengths to make the Y-Axis ( I think ) carriages, and made a centre-pop mark for boring out to accept the 12mm linear bearings.




I then mounted it off-centre in the 4-jaw chuck in the lathe. I used a "wiggler" in the tailstock with the point in the centre-pop, then put a DTI on the shaft of the wiggler and adjusted until I had less than 0.1mm runout. Then it was just a case of drilling out to 18mm, then swapping to a boring bar to bore out to a nice tight-sliding fit for the bearings.

Then, putting the bearings into both carriages, and sliding a 12mm rod into them I mounted both end-to-end in the mill, supported by the 12mm bar, then skimmed each face in turn to make sure that each face is parallel to the bore.

Then I drilled and reamed the 8mm holes for the x-axis bars, which ended up being on 48mm centres, so 40mm between inner edges of the bars.

I couldn't help myself and assembled the bits together, it moves really easily with no racking.




There's still a lot to be done to these y-axis carriages, next up will be milling off some on the underside for mounting idler-pulleys for the belts, and also drilling and tapping some holes for screws to secure the bearings and also the 8mm x-axis bars.

Once all the work is done on the y-axis carriages I will see where more material can be milled off to lighten it further.

Thanks for watching,
Tim

[spuddevans]

A little more progress made today. I started working on the Y-Carriage blocks to whittle them down a bit for mounting the bearing-pulleys for the belts to run round. So I milled off a "T" section, drilled for tapping M3 for mounting the bearing-pulleys (the centre hole) and drilled two holes 15mm either side so that I can make a little bracket to fix the top side of the pulleys.




Another view with the bearing-pulleys put in place,




End-on view,




I reckon that, compared to the plain Ali' blanks, these are now at least half the weight, and I'm not done yet!! I still have a fair bit more I can mill off, but any major milling will wait until near completion as it gets hard to accurately hold in the vice when you don't have a couple of the original squared faces untouched.

Onwards and sideways,
Tim

[awemawson]

That's going to be one nice solid machine when you've finished Tim. Most of the 3D stuff out there on the web looks very flimsy so it's nice to see a proper build   

[spuddevans]

That's going to be one nice solid machine when you've finished Tim. Most of the 3D stuff out there on the web looks very flimsy so it's nice to see a proper build   

Thanks Andrew, solid is what I'm going for.

Like you say, some of the 3d printers being built have been quite "elastic", and seem to be quite lacking in basic engineering good-practice. There's some more being built now using Ali' extrusions, but many seem to make the joints out of plastic printed parts 

It's a slow build due to limited time in the workshop, but I'm really enjoying the challenge of figuring this out as I go.

Tim

[nrml]

In my experience (albeit limited), super rigidity isn't particularly necessary for a FDM printer. Vibration and resonance control are far more important and having a rigid all metal frame doesn't automatically eliminate this. In fact it might even make it more difficult to manage. I suspect (but can't prove) that having dis-similar materials in the frame might help break up resonance and might actually be advantageous. The forces and loads on a 3D printer are not really going to stress the frame and joints unless it is really poorly designed and / or put together.

I have no doubt that the care and effort you are putting into your project will result in an excellent printer, but I doubt that it will perform any better than a well designed and carefully built printer with plastic joints. I presume building it to your standards is more important to you than rather than just aiming to get a half decent working product in the quickest time possible.

I will be watching your build with interest as I have an ongoing corexy build (aluminium extrusions and aluminium corner brackets) which is currently on the back burner. My problems started with bearing issues. I found that igus drylins had far too much play for my satisfaction and oilite bushes were binding at the slightest hint of dust. I have reluctantly decided to scrap the smooth rods and use linear rails instead but this means a Z axis redesign, which is where I am stuck at. Your thread is a good motivation to pull my finger out and restart my project.

How is your Z axis implementation going to be? IMO it is the most challenging part of a corexy design.

[spuddevans]

In my experience (albeit limited), super rigidity isn't particularly necessary for a FDM printer. Vibration and resonance control are far more important and having a rigid all metal frame doesn't automatically eliminate this. In fact it might even make it more difficult to manage.

That's true, at the moment my frame does have a little "ring" to it, but I have a cunning plan to deal with that (mainly involving kiln-dried sand part-filling some of the hollow sections

I suspect (but can't prove) that having dis-similar materials in the frame might help break up resonance and might actually be advantageous. The forces and loads on a 3D printer are not really going to stress the frame and joints unless it is really poorly designed and / or put together.

You could well be right about the dis-similar metals, my concern would be the differing thermal expansion properties (I am intending to enclose my frame for ABS printing) I bow to your (and others) experience regarding the stresses experienced by a 3d printer, I was assuming that the moving mass of the print-head/extruder/stepper (I intend to use a direct extruder) would benefit from a more rigid/stiff frame to absorb the movements when the direction changes (unless printing really slow)

I have no doubt that the care and effort you are putting into your project will result in an excellent printer, but I doubt that it will perform any better than a well designed and carefully built printer with plastic joints. I presume building it to your standards is more important to you than rather than just aiming to get a half decent working product in the quickest time possible.

A lot of what I am doing is based on the materials I have access to and the techniques I am familiar with, rather than any thought of it being a better way of doing things. You're probably right about the end result being no better in performance than others, especially as even the most flexibly built printers can be run slow enough to not wobble when printing.

I will be watching your build with interest as I have an ongoing corexy build (aluminium extrusions and aluminium corner brackets) which is currently on the back burner. My problems started with bearing issues. I found that igus drylins had far too much play for my satisfaction and oilite bushes were binding at the slightest hint of dust. I have reluctantly decided to scrap the smooth rods and use linear rails instead but this means a Z axis redesign, which is where I am stuck at. Your thread is a good motivation to pull my finger out and restart my project.

That's interesting about your bearing issues with the smooth rods, I have seen that many others have moved over to linear bearing rails and blocks instead of smooth rods. I probably would have done that myself if I hadn't already bought the smooth rods and bearings.

I'd be interested to see your progress when you restart your project.

How is your Z axis implementation going to be? IMO it is the most challenging part of a corexy design.

I plan to use 3 8mm smooth rods (1 at each front corner, and the 3rd at the centre of the rear of the print bed) with linear bearings for keeping the printbed from moving in the X or Y axis', and lift will be provided by 3 10mm leadscrews located in roughly the same areas. The 3 leadscrews will be driven by a continuous GT2 belt with one stepper moter. The leadscrew pitch is 2mm, multiplied by 200 steps per rev of the stepper gives a vertical resolution of 0.01mm without resorting to microstepping or "gearing down" using different pulley sizes.

Hopefully all of that will give me a printbed that is very stable from movement in X and Y, and also that can be levelled relative to the print-nozzle, and won't sag or bow with weight from the print or from just the weight of the bed itself.

Of course, until it's actually built, all of that is subject to change 

Tim

[Joules]

Nice work Tim and good approach, I think nrml is pretty spot on with his assessment on resonance.  Quite a few comment on my print quality and the fact it only has 8mm rods and plastic carriage components.  What they fail to grasp is the fact my print bed is only 150mm square.  It's much less inclined to ring like a larger bed, the chassis on this machine is also a welded sheet steel box.  Yes you can press the rear like one of those clicker frogs.  However in normal operation nothing resonates.  If the rear had done I would have glued or riveted a spine across it.

Following your build with interest, just flush out the Chinese linear bearings and pack them with a good (LM2) grease every year or so, they will pick up dust and dirt.  Same reason I change the belts annually too.

[spuddevans]

Following your build with interest, just flush out the Chinese linear bearings and pack them with a good (LM2) grease every year or so, they will pick up dust and dirt.  Same reason I change the belts annually too.

Thanks Joules  and thanks for the tip on flushing out the linear bearings and repacking 


I got a little more done today (emphasis on "little"),

I tapped the 3 holes on each carriage M3. Then I made the pin that will screw in to hold the ball-raced pulleys. The plain pulley has a bore, or rather, the bearing in the plain pulley has a bore of 3mm, whereas the bearing in the toothed pulley has a bore of 5mm. So I turned up a stepped pin with a short section of M3 thread on each end, the idea being that the pin can be oriented either way up (as one carriage will have the plain pulley on the bottom, the other will have it on the top. ( I think ?!?!? ) )

Here's a pic.




I made it out of brass as it was easier, hopefully the forces will not be too much for brass, but I can re-make it out of steel if needed. I will be making a bracket that will fit on top to support the pin, so that will ease the forces applied to the pin.

Next up will be the aforementioned bracket.

Tim

[tom osselton]

Here's a idea I thought was good

[nrml]

Will you be using a water cooled hot end if you are going to have an enclosed build chamber?

[PekkaNF]

In my experience (albeit limited), super rigidity isn't particularly necessary for a FDM printer. Vibration and resonance control are far more important and having a rigid all metal frame doesn't automatically eliminate this. In fact it might even make it more difficult to manage. I suspect (but can't prove) that having dis-similar materials in the frame might help break up resonance and might actually be advantageous. The forces and loads on a 3D printer are not really going to stress the frame and joints unless it is really poorly designed and / or put together. ...

I have no doubt that this observation is true to some extent, but I think that it's pretty much the same thing. Bit like flipside of the coin. Good question here is what is good enough and where the error budget goes.

There are some little hobby projects I am working and then there are some pretty big machines we build at the work, they have linear rails from 100 mm to 10 metres and nominal sizes of linear rails from 25 to 65 mm and up. Some "axis" are electrohydraulic, some electronic servo system. I may have learnt a thing or two from them.

All (steel) frames are flexible, increasing pipe/beam diameter makes structures stiffer and you can make servo loops faster compared to increasing the beam skin thickness.

Filling structures with concrete without much attention to some very exotic details is a bad idea. Immovable structures filled with various substances can have very good damping, but it all really "depends" on scale. Sometimes for unseen high frequency dampening internal stringer/plate might do the job, or carefully placed heavy weight highloss mat might do - for some violent shaking (that is not only very obvious to see but it's so hard that rattles your teeth off) might be dampened by loose glass spheres or granules of metal. Sometimes it just best to kill the excitation if you can.

Some stuff you really need FEA and accelerometers/impact hammer and megabucks equipment/software.

BUT some stuff is pretty easy and straight forward. Like putting rails/actuator on same plane and CG of the moving part as close to this plane as possible. And avoid at all cost forces/gravity to change loading direction. Few years ago I could not spell "Abbe error", now I have a rough idea what it is and recognize it....basically it is when angular error manifestes itself in linear (unwanted) motion or error:
http://www.mdpi.com/applsci/applsci-06-00156/article_deploy/html/images/applsci-06-00156-g012-1024.png
That I see in very many designs.

What I'm trying to say here that I definitely see merit here in this design. X/Y axis are very close to each others, which is good. What might be worth of having a look is that when X/Y axis shakes this structure, it will move in relation of the table...worst case scenario is that X/Y axis will excite table and table might move random direction, more than doubling the error. If that comes to play, making frame uprights more rigid, or dampening the coupling into table OR alternatively making coupling into the table more rigid should overcome it. Anyways I think that top of that machine shaking is easier to fix than Abbe error inherent to many design.


Pekka

[spuddevans]

Will you be using a water cooled hot end if you are going to have an enclosed build chamber?

I haven't decided whether to watercool yet, I like the idea, I'm just not sure how the water-filled tubes will effect the free movement of the print-head, perhaps if I used very flexible (perhaps silicone) tubing?


Thanks for that info Pekka, That's the error (abbe) I'm trying to avoid. If I'm understanding it correctly, I did have this error on my mill for a while, the head was trammed to the table fine, when I used a short spotting drill to start a hole and then raised the head up to fit the much longer drill bit in I found the tip of the drill-bit was slightly off from the spotted hole, despite the head reading ok when I trammed it!! (I ended up having to tear the head down completely to shim the castings to get everything correctly set up)

I got the brackets made today, just a couple of 8mm x 10mm x 43mm scraps of ali, with a slot milled in it to accommodate the upper bearing.

Not much, but some forward progress.




Tim

[nrml]

There are some very neat and easy to implement ideas for water cooled hot ends on this thread https://forum.e3d-online.com/index.php?threads/e3d-water-cooled-mod.53/

Something like that connected to a little peltier water cooler unit like  the ones sold on aliexpress.com and a computer CPU water cooling pump/reservoir would make a really neat water cooled hot end.

[spuddevans]

There are some very neat and easy to implement ideas for water cooled hot ends on this thread https://forum.e3d-online.com/index.php?threads/e3d-water-cooled-mod.53/

 
Thanks for that link, that's given me a lot to think about. Hmm 

I am definitely leaning towards water cooling, and can think of some ideas for water-cooling the steppers as well. 


Tim

[hanermo]

Great build !
I like the att. to detail and the high qty of mechanical components You did.

[spuddevans]

Thanks hanermo 

I got a little time in the workshop and started by drilling some oblong holes. (now I know why square drill bits are so expensive, but they would have saved a lot of sweat and effort with drills & files)




This might make it a bit more obvious what they are for




ok, one more




From another angle




The square holes are pretty rough looking, I will tidy them up a bit later when the belt path is more precisely known. I could have got away with making both of the holes the same size (the smaller size), but hey ho, it is what it is.

Next up will be drilling and tapping the carriage blocks for grub screws to grip the 8mm x-axis bars, and shortening the same 8mm bars to their proper length. This will allow me to measure precisely the distance between centres of the pulleys of both carriages, then I can make up a bracket to hold the pulleys at the front of the printer frame (for the rest of the belt path). I will probably also drill and tap a couple of holes to stop the 12mm linear bearings from sliding out.

Then will be the making of the mounting hardware for the X & Y steppers. I've been thinking about the X & Y resolution, using 200 step stepper motors, and even using 16 tooth pulleys on the steppers, that will give me a full-step resolution of 0.16mm (32 mm of belt movement per rev, divided by 200 steps per rev = ) Using 20 tooth pulleys on the steppers will give me 0.2mm full-step resolution.

However, I have had a cunning plan. I have a couple of 36 tooth pulleys, which I could pair with either 20 or 16 tooth pulleys, and make a belt-drive reduction (I also have some short loops of GT2 belt) of either 2.25:1 or 1.8:1. As I have no experience with 3d printers, has anyone any experience with adding belt-reduction "gearing" to increase the resolution?

I know about micro-stepping, and I will definitely be using that (primarily to quieten the steppers), but I do know that the individual micro-steps are not exactly evenly spaced between full steps, hence the thought that micro-stepping would not be the most accurate way of increasing resolution.

On the other hand, adding a belt-driven reduction may also increase "springy-ness" ( actual scientific term there ) and resonance when abrupt direction changes are made. It certainly will reduce the travel speed by the same factor of reduction, but will also increase the torque.

If anyone has any thoughts/opinions, speak now ( or later, I won't be working on this part for a couple of weeks or more )

Until next time....

Tim

[Joules]

You will be 3D printing covers to tart up those holes. 

[spuddevans]

You will be 3D printing covers to tart up those holes. 

Probably. I would weld them up a bit, but I'm afraid that the welding would cause too much distortion in the frame. Anyways, they will be almost hidden when the whole project is complete and enclosed.


I managed to grab a few mins in the workshop today and drilled and tapped for M6 grub screws to grip the 8mm ground rods. Then I drilled and tapped some M3 holes to hold the 12mm linear bearing in the carriages. (I drilled/tapped 1 hole at each end, carefully positioned so that the M3 washer is just protruding over the edge of the bearing)




View of the M6 holes (I counter-bored 6.2mm to a depth of 20mm, leaving approx' 9mm to tap M6, much easier to tap)



Onwards and sideways

Tim

[PekkaNF]

This is very interesting thread.

.....
Onwards and sideways

I was told " Wet fly - down and across"

Pekka

[spuddevans]

Well, it's amazing how life can get in the way of building a 3D printer!!!

So after quite a looooong break from this project, I've been getting back into this again. There's not a huge amount to show from the last few weeks, but I've been getting a few small items ticked off the "to-do" list.

Here's a shot of where I'm up to at the moment, I've been working on the Z-axis which is based on 3 leadscrews coupled with some 8mm smooth rods. This is a poor semi-overhead view


I made a frame consisting of an upper and lower sections of 25mm box steel, well actually I made 2 frames, one for the front and one frame for the back, and these are where the Z-axis guide rods and leadscrews will be held in.

The leadscrews have been turned down to 8mm at their ends where they go into the 25mm box-section steel, and for each leadscrew there are 2 flange-bearings (countersunk one each on the top and bottom sides of the box-section) which keeps them running smoothly, and the 40t pulleys are also 8mm bore.

I very precisely drilled and bored the 25mm box-section frames for each set of holes (for each front carriage there is one guide rod exactly 25mm centres from the leadscrew, for the rear carriage there was 2 guide rods, both exactly 25mm centres from a central leadscrew). There are 2 lengths of 25mm box section in each frame, one set for the rear, and one set for the front. Each set were drilled and bored together to ensure alignment was kept accurate.
The 25mm box-section frames were carefully milled to length to be a precise, snug fit into the main printer frame, and thus they only need a M4 screw at each end to hold them in place. (they actually hold themselves in place without the screws, but the screws make sure that when everything is moving and warmed up, nothing will move)

The Z-axis carriages are milled out of aluminium bar, here's a shot of one of the front 2 carriages taken from below (the other 2 holes are drilled and tapped for mounting the leadscrew-nut, but as it's in-progress it's not worth putting them in as they'll be dis-assembled and re-assembled a good few times before the whole thing is done)


and from above;




And the rear central Z carriage with 2 guide-rods (2 are not strictly needed, but they turned out to be useful for holding bearings for guiding the belt) from below, showing the 2 Igus bearings (held in a blind pocket) and the leadscrew nut.


and from above;


You can see on that pic the "T" piece that links all 3 Z-axis carriages, and also you can see the present fixing arrangement (a M4 screw attaching it to the carriages thru a slot in the "T" piece). This is going to be changed as I have an idea to incorporate the bed levelling into this mounting.


This is a shot of the routing of the Z-axis belt around the rear central carriage leadscrew and guide-rods. I wanted to make sure that the belt had >90 degrees of contact on each of the pulleys driving the leadscrews, so you can see that the 2 guide-rods were very handy to pop a couple of bearings on to help "wrap" and guide the belt around the 40t pulley


A wider view showing the belt-tensioning device



And a close-up view of the Z stepper mount;




Here's a closer shot of how the belt routes around one of the front leadscrews and guide-rod;



The leadscrews are 2mm pitch, and they are driven by 40t pulleys, the Z stepper has a 20t pulley on it, which will give a full-step resolution of 0.005mm (more than I need, I may put a 40t pulley on the stepper)

One of the 3 leadscrews has virtually no runout/wobble, but the other 2 need some "adjustment" to get them closer to running true. It doesn't seem to wobble the whole carriage, but there is a slight, but noticeable, increase of resistance to the belt movement. So the plan is to take them out and then find out exactly where the bend is and attempt to straighten it out.


So next up is straightening out the 2 wobbly leadscrews, and then sorting out the bed-levelling arrangements and the mountings for the bed itself.

Stay tuned for the next update (but maybe don't hold your breath, LOL!!)

[WeldingRod]

Important tip: take your leadscrews out and slide your carriage up and down by hand.  If it doesnt run free over the whole span, solve that first!!!

Sent from my SM-G892A using Tapatalk