Building a Kossel: Need advice on final parts

+1 for everything Doug said.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Alright I can scratch that off the list then! Thanks! I think I'll see what I'll get. It'll indeed be either the option you've mentioned, or maybe the BI V2.5 frame someone's selling on the deltabot google group.

Do you know how much of a difference the Haydn's Magnetic rods are with and without the Delrin ends? There is quite a price difference between them and I'm not sure if it's worth it right now to opt for the Delrin rod ends.

I've got the delrin ends and believe it or not my machine became less precise with Haydns magnets, not by much and I'm not criticising them as they are great and beautifully/precisely made, but I had metal carriages and effector before and had to go back to printed ones to use them. The big difference is they do not detach even printing at extreme speeds whereas the k800 ones I had before did.

I figured since it was a no expense spared build I'd have the delrin ones. You could always use a little grease between the balls and the arms.

Hopefully dc42's new effector system (Haydn compatible) will redress the balance.

All metal is really a big improvement, I was sceptical at first, but unless you have a cnc it's not easily or cheaply adapted or upgraded until the big name start producing the parts and the Chinese copy them.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

It's great to hear the Haydn's rods don't detach at extreme speeds though! I wouldn't want to wake up one day with a mess all over my printer cause one of the arms detached.

And hmm... I realised that an all metal build would be an improvement, but I didn't expect it to be that big of a difference really. Maybe I could CNC the parts at my school (if they'll allow me to). Guess I know what to do.

Thanks!

There always many way to do thing, you can use magnetic arm without being force to use plastic carrier or effector... just build your own, you only need to be able to screw the ball at 45 degree angle.. a simple .125 " plate bend can do the trick easy and very lightweight compare to a full cnc effector/carrier. Same for the metal corner you can also build your own if you dont care about using some elbows grease... and are good with your hands go for it, they dont need to be fancy and need a cnc to work. I made some very solid one using .250" L angle cut at 60 degree each end and some drilling and L braket to secure the V-slot rails. They beat the robotdigg one by a large amount. They are larger, have adjustment everywhere and are so big they make a hexagon hybrid delta kind off ( 6 side but 3 rails) so the whole thing act as level effect for the rails.


BTW dont put grease as DjdemonD recommend its a bad advice, grease will create uneven arm lenght. You very light silicon oil instead, ( I use RC motor bearing oil, its almost like water) I lets it sit on the nylon/plastics end for a while then I wipe it off lightly. I probably lube more often than with grease but its not a big problem to lube them once a month if needed.

That is true but that does take more work and you have to make sure that everythings exactly the same. Errors in the carriages/effector will probably show in the print quality or accuracy of the printer. I'll probably just keep it simple and either use the Robotdigg ones, CNC the parts or just use 3D printed parts to start with.

Thanks for the advice! I'll see how the parts run without any lube and decide later on what to do with it I guess.

That is true but that does take more work and you have to make sure that everythings exactly the same. Errors in the carriages/effector will probably show in the print quality or accuracy of the printer. I'll probably just keep it simple and either use the Robotdigg ones, CNC the parts or just use 3D printed parts to start with.

Thanks for the advice! I'll see how the parts run without any lube and decide later on what to do with that I guess.

that where you are kind of wrong You do not need to have all part exactly the same if you build in some clearance for hole, you adjust it when you first build it and put everything square ( you kinda need to do it anyway with pre build part, a 3d printer is not something you put A on B and expect it to work with the highest quality). Best build have adjustment all around and carefully measure everything before tightening,. something you CANT do if the part do not have some sort of adjustment and are screw in instead of Bolt in.

My diy built all done by me so far rival the quality of any other printer with prebuild stuff. but I took my time to really adjust it well and spend quite some time measuring and setting everything perfectly square. That really the key of everything specially for delta, the quality of the building reflect the quality of the finish printed parts.

I ran my magnetic dry for almost 4 month before I lube them, there were no wear or anything but they more really more freely now that they are lube a little, less friction.

I'd second that you can build it all by hand if you wish, and as long as everything measures up true, you will get the same results, and I imagine a great deal of satisfaction that you built it with your hands.

However if you are not so handy, then you can print parts, but accept that for some areas of the machine, frame corners, carriages, effector etc.. there will be some flexibility and therefore play in the mechanics as a result. Or you can buy parts ready made and here you are limited by whats on sale, how much it costs and whether its made correctly.

GroupB - if you had access to a high quality CNC machine for free would you make an effector to replace your hand made one?

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I know I would like to machine my own effector. I have a nice alu design that I would love to make. Will have to modify it a great deal, knowing what I know now, but loved the idea of a billet alu effector.

Also, there are a few very good looking ones floating around. Including one that was printed in metal, that was a sweet design.

Lykle

Yeah I've read a lot already that with Delta printers, you have to make sure you build it well and take your time to get everything right. It does makes sense in a way that you could create the parts on your own, but it would probably be easier and faster for me to CNC the parts if I can use the one at my school. I'll see what I end up doing with the parts I guess and in the meantime look for the designs that are floating around according to Lykle. Thanks!

Quote
GroupB
You do not need to have all part exactly the same if you build in some clearance for hole, you adjust it when you first build it and put everything square ( you kinda need to do it anyway with pre build part, a 3d printer is not something you put A on B and expect it to work with the highest quality). Best build have adjustment all around and carefully measure everything before tightening,. something you CANT do if the part do not have some sort of adjustment and are screw in instead of Bolt in.

+1 for this. No part is ever perfect, and you will need to adjust pretty much everything once it's built and you have some errors in your calibration. If you've got some experience under your belt then you can shortcut the process by being meticulous at build time, but even then you'll still need to go back around adjusting things when you upgrade/change something later on.

There's nothing wrong with printed corners if you have other means of adjusting the squareness of the build (ie diagonal bracing of some sort, preferably adjustable with screw threads). However printed effectors are out, in my book. They are just too flimsy, heat intolerant, and not exactly at the right angles.

Quote
DjDemonD
GroupB - if you had access to a high quality CNC machine for free would you make an effector to replace your hand made one?

To replace probably not since its working fine and it weight almost nothing and I move on to other projects that take all my time, a new printer maybe but it would be very similar design to the one I have right now but the 45 degree tab will be mill instead of bend or a brand new design like the e3d style heatsink milled with tab for ball rather than a plate where you attach the head, why not have the head with the joint on it,I dont think the heat will go to the joint ( mine is always cold from the middle to the top and just little bit warm below that).
So far the cnc effector I saw floating around look heavy as hell like 1 inch thick to allow the 45 degree and no extra aluminum remove in the center or at the tab because they use screw in balls instead of bolt in ball ( adjustment again)... lets say im not impress with the design of CNC effector I saw so far, they dont think about the weight so much they just copy the plastic design and they done.

Some part like an effector or the carrier dont really need CNC to be effective , off course I bend mine with my hand and a jig that can cause error ( I have to throw one away it was off a little) but if you have access to a nice commercial bender or even a small one I think its far more effective way of doing it than CNC one, specially if you have to pay for CNC and material. Sometime there no need to over complicate a design.

The only upgrade I wanted to make was to use mgn rail instead of vslot but im not sure anymore, I use the official one with they carrier plate and they work nicely and they are precise and silent and probably weight less. I plan to test the mgn on a new cartesian/core hybrid but I need the rails for support the X not something you really need on delta, then I will see if they really that much better than v-slot.

And yes its always great to hear family and friends when you show them the printer and they ask where you got it and how much it cost and you tell them you made all the parts yourself and its cost way less than commercial printer and way better and faster. its priceless to see they "whoa" reaction when you do a little Delta demo, I never get tired of it, but I guess everyone here who did a build know what im talking about

Quote
Riggi
Okay thanks for the warning on the idlers nebbian!

Just an update on this.

Even expensive toothed idlers from reputable companies have this issue:

[www.youtube.com]

I think the only way to get zero runout idlers is to use flanged bearings, unless you have a lathe and can true up the idlers yourself.

You can also use pulley instead of idler and put the shaft on bearing on each side so the shaft turn with the pulley, that what I did on my delta, when I did it the toothed idler were rare, since its not a core XY you dont need to have 2 idler per shaft so pulley on shaft with a flat is an option, you probably were thinking about support on each side anyway.

That's a good point GroupB, and I was considering doing that. Packaging is a problem though, as the whole assembly has to mate with the existing printed vertices at the top of my delta. I've ended up using 623ZZ flanged bearings with an extra printed flange, I just did a calibration and I'm back to 0.03mm deviation.
[www.thingiverse.com]

There are so many things that can throw delta calibration off, poor idler bearings were the last thing I would have ever considered. It's a pity that the ebay and robotdigg listings don't say "Warning, lots of runout, don't consider these if you want accurate prints".

Sorry to sidetrack the original post, back to scheduled programming now

so where we can buy good quality idler?

Just used flanged bearings seems to be the solution.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

No I don't mind your small sidetrack at all Nebbian! It's an interesting read, one of which I didn't know off yet. Thanks for informing!

It's rather disappointing to hear that the CNC effector that's flying around is a rather poor design. Weight seems to be very important for printers just to have as little inertia as possible on the hot end. Maybe I'll make my own design or alter the existing one if I decide to CNC parts or not. Also I'm curious now though, how long did it take you, on average, to build your own 3D printer GroupB?

About a year into learning the in and out of delta ( at the time the info on delta were prettry rare), and it was my first build and first 3d printer ( yah I know ppl told me not to do it but I did it anyway), then about 4 month design in CAD, I change my design a couple time, first a friend offer me CNC work then decide to step back so I redesign to be able to make all part myself with simple tool ( drill press and table saw ) with less material, that why I end up with L shape aluminum cut to 60, easy enough to make and super rigid.

There were no robodigg corner when I start my design and build only plastic corner and I wanted a full metal build...

When I receive the aluminum I start to cut and drill and sand/file for about 2 week, Then I build the frame and look at it every morning for a year waiting for my custom part to be done...friend of a friend who were suppose to be done a month after I finish my frame but keep giving me excuse so it took 1 year for 6 hour job and must of it was butcher because he was tired of me asking when the parts will be done. ( never again Ill make that mistake, next time I check local shop and pay more). Once I receive my missing part and mod them to fit, it took about 1 week to finish the build and another week to make the firmware work and fix little extraction issue I had ( bad thermo sensor reading, it was way offset).

Edited 1 time(s). Last edit at 02/01/2017 10:28PM by GroupB.

Oh wow I was expecting it to take a while, but not about 2 years. You must've quite the perseverence to be able to stay interested in your project for so long. How did you experience calibrating your machine? Was it a pain in the ass? A friend of mine isn't convinced of my choice to build a Delta printer because he believes calibrating the printer will be a nightmare and you'll have to do it frequently. I know calibrating a Delta printer is more difficult compared to others (say a cartesian) but I don't think it's that bad as long as you build it well and have a sturdy frame.

Delta calibration before autocalibration came along, was achievable but quite a steep learning curve. With my first delta I tried it and did not persevere and used Rich Cattel's Marlin with autocalibration (and FSR's under the bed as z-probe), and got good results. David's newer least squares method is much more efficient, and faster. So if you use Duet, then calibration will take minutes, (assuming good mechanical build etc..) all you need to input is approximate z height, printable radius, rod length, the algorithm will take it from there.

My mini delta I designed is too small to easily fit a z probe, so I calibrated it manually but using the least squares system [escher3d.com] its fairly self explanatory. It takes a few iterations to derive a reasonable solution. It works well on the mini delta as the bed printable radius is only 100mm.

You will find that the larger the bed, the more difficult it becomes, so the use of grid levelling after calibration compensates for uneveness in the bed (which might throw autocalibration off somewhat) or not quite enough mechanical or sensor precision to derive a perfect solution.

All of this is much more difficult than with a cartesian where you just make sure the 3 axes are orthogonal to one another, and work out how many steps/mm you need to enter into the firmware, and determine the travel limits of your axes. So if you pick things up quickly and are willing to learn and persevere go for a delta, if you don't yet know your stepper from your endstop a cartesian might be an easier route to take.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

It sounds smoother to calibrate a delta than I thought it is. It's interesting to hear that the larger the bed, the more difficult it becomes to auto calibrate a delta printer. Maybe I should try to go for a slightly smaller printer so I have less issues with calibrating the printer. But all in all, I do know a stepper from an endstop so I assume it's safe to go for a delta (which I was planning to do anyways)!

Another question my friend had is if a delta is truly worth it for the extra speed it offers. Apparently there are cartesian printers already that can go as fast as a delta printer (I have no sources on this), which made him ask the question why you would go for a delta over a cartesian as a cartesian is easier anyways.

It's just a question of geometry, let's say your calibration leaves 1 degree of error in your nozzle tracking across the bed in this simplified example. Your first layer might be 0.3mm thick. So across a small bed you might still get an acceptable first layer at any point on the bed more squashed in some places but stuck down everywhere. But if your bed is 300mm your nozzle will be flying high at one side and digging into the bed on the other. With larger size there is less tolerance of error, in addition to the machine being larger which might end up more flexible unless it's heavier in build to the third power.

As for why build a delta, there is a lot going for them mechanically, stationary bed, stationary motors, big z height and small footprint with neat packaging for electronics etc.. No need to stack motors like a cartesian machine. Corexy addresses this for x and y but still has a moving bed for z. The real reason I suspect we have deltas is they're seen as harder to build and calibrate so there's a touch of elitism, and they are beautiful and beguiling machines to watch, more so for the uninitiated.

As for speed its all about having as little moving mass as possible. In a delta the moving mass is the effector and everything on it, the rods, the carriages and the belts. In a corexy there is an x axis (steel rods/linear guides) moving back and forth, plus a print head and y-carriages which is much more likely to be heavier, in a cartesian there is an x axis and a y axis moving about, with motors moving on them (200-400g a piece). The more moving mass the more inertia and so the less accurately you can accelerate and probably more importantly decelerate that mass, and change direction so that your corners are sharp etc... To be fair its all limited by the speed at which you can melt and extrude the plastic. See if you can find videos of high speed cartesian printing, there will be some, but there will be more corexy's and deltas. There was a slightly childish speed contest posted on the general forum maybe check it out, (no cartesian's entered when I last looked).

Edited 2 time(s). Last edit at 02/02/2017 08:59AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Oh okay thanks for your explanation. I suppose it would be better to manually calibrate a bigger delta printer then?

And thanks for the summary about a delta printer and it's speed compared to others. It still does seem to me like the more attractive option vs the others. Maybe this summary will be able to convince my friend and he will see the light after all!

In the end you are not going to build or buy one definitive 3D printer. You are going to start somewhere, with something, then either give up (unlikely having read your posts) or enjoy it immensely/be frustrated by it (mainly enjoy but its ups and downs for sure) and want to move on to the next one - bigger/smaller/faster/more innovative, it all moves so fast.

I started 3D printing around 2 years ago and I've fixed up a secondhand I3, bought a ready built kossel and improved on it, built a corexy kit (smartrapcore) then enlarged, enclosed it and modified it to unrecognisable, self sourced a Kossel XL, and designed a microdelta. I'd quite like an out of the box printer like a cubicon or ultimaker 3 just to see what this type of machine is like, but would probably not enjoy it so much, as there's less to tinker with. Would also like to try resin, I'm a backer of ONO smartphone printer but I'm 50/50 about whether they'll deliver.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Yeah exactly. Maybe I should go with something smaller just to start with and just modify it bit by bit, instead of going for a bigger one with all the good stuff. I'm pretty sure I won't give up on it, considering I also will be using it for school related projects and probably also some personal projects if I have spare time for those. I already have experience with both the enjoyment and frustrations with school projects, having to design products and stuff so I'm sure it'll be very similar to owning a 3D printer. I'll see where it takes me I guess!

SLA printers are quite interesting as well and I've just checked that ONO smartphone printer. It does look nice but the question is always with kickstarters wether or not they'll deliver or if they'll reach expectations. I did look at SLA printers as well, but I think it's too early for consumer SLA printers cause the resin itself is still quite expensive as far as I know. Maybe the ONO printer is the beginning, maybe not. Only time will tell I suppose.

Quote
Riggi
Oh okay thanks for your explanation. I suppose it would be better to manually calibrate a bigger delta printer then?

Auto-calibrating a large delta isn't difficult if you have a good Z probe. Either one that uses the nozzle itself as the probe (such as the piezo sensor that there are a few threads about), or one that is small enough to mount close to the nozzle (such as my IR sensor) and you have taken care to minimise effector tilt. There is more information on auto calibrating a delta at [duet3d.com].

Edited 1 time(s). Last edit at 02/03/2017 02:37AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Ah okay. I was probably going to go for your IR sensor so it's good to hear that auto calibrating works fine on a larger delta printer with it. I'll bookmark this link for when it's time to build my printer (which may still take a while sadly)

Quote
dc42

Quote
Riggi
Oh okay thanks for your explanation. I suppose it would be better to manually calibrate a bigger delta printer then?

Auto-calibrating a large delta isn't difficult if you have a good Z probe. Either one that uses the nozzle itself as the probe (such as the pizeo sensor that there are a few threads about), or one that is small enough to mount close to the nozzle (such as my IR sensor) and you have taken care to minimise effector tilt. There is more information on auto calibrating a delta at [duet3d.com].

No I wouldn't manually calibrate unless you wanted to verify your auto-calibration because you suspect your probing method was not accurate or unless probing using a sensor is not possible, such as on my micro-delta where there isn't room for a probe on the effector.

David I agree with you about larger deltas not necessarily being harder to calibrate, but this presupposes that despite the larger size the bed is still as flat and the mechanics are just as precise and rigid, as the smaller machine. This is what I am getting at, its easier/cheaper to make a micro-delta rigid than a very large delta. My micro delta has 10cm aluminium threaded rod for arms but it doesn't bend much as its short. If we scaled machines-up to be as rigid as their smaller versions by the cube of the increase in size, this would hold true. It seems common to double the size but not increase the frame thickness by 8.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions