Throw away your short belts - direct drive.

[polymorph 1]

For speed, low pitch is better. A two-start 5tpi would be better than a 12tpi single-start, or worse allthread at 18 or 24tpi. Right now, a single revolution of the stepper moves it, what, a few inches? I don't have my Ultimaker in front of me so I can't measure diameter/circumference. So already a 5tpi ballscrew would require the stepper to move 5 to 10 times as fast for the same speed. Might have to upgrade the steppers, drivers, and power supply for that.

Ballscrews are spendy... $35 is pretty cheap for that length of ballscrew and nut.

[IRobertI 521]

I probably got my terminology mixed up, English is my second language and mechanics my... tenth What I meant to say was that they all seem to have a short distance between threads as in lots of turns per inch.

[polymorph 1]

Ah! Yes, that is precisely the problem. You may have been looking at Acme thread, also, not ballscrews. Usually just the nut for a ballscrew is $35 or more.

[IRobertI 521]

This was the one I took a look at yesterday:

http://www.ebay.com/itm/SFU1605-Ball-Sc ... 1e705fc70f

 

[polymorph 1]

That is a ballscrew and nut, all right!

[antiklesys 3]

Wouldn't it be too messy to add the ballscrew and nuts while we can add 2 motors for direct drive to remove the long belts?

[snowygrouch 1]

Hi,

I spend quite a bit of time looking at alternative drive systems, I was thinking to eventually

go to ballscrews. The issue is that belts are really light, and have really low friction and are super cheap.

Its very hard to beat that combination, some lead screws and a nut are not expensive but the

friction is higher, and also it will not be long at all before they wear enough to produce significant backlash.

This means that ballscrews are the best option left, and good quality ballscrews are not cheap at all, then

you need to double up the steppers.

I think the main issue with the belts is that there is no sensible method in the standard design to tension

them. Yes you "can do" it, but its involving some little plastic bits that are either a massive pain to

install or will start bending after a short time.

Its hard to make a balanced comparison between a proper system with long belts, PROPER tensioners and

good sprockets...and ballscrews. I think that with some high quality idler wheel tensioners, the performance

difference between ballscrews and belts might not be a great deal.

I certainly dont remember seeing any commercial 2D printers using ballscrews....and the tiny

UMaker printhead is hardly in the weight range to start getting worried enough about the loads to

say that industrial standard linear guide rails and so on are required.

I will probably upgrade the remaining long belts with some really good tensioners, view the results and then

make a call on ballscrews. Im not going to go and spend 400 euros just for fun before I see the results from

a properly done belt system. However the short belts are definetly in the bin for ever, for me.

C.

[ianeagland 0]

Hi

I started with a ballscrew repstrap. One of the biggest troubles I had was ball screw inertia. I could not get the acceleration or speed you need for 3D printing. It is the low inertia required to move the Ultimaker extruder head that makes it a much better machine than a mendel. I have built a Repstrap, Mendel & finally purchased an Ultimaker. It is streets ahead in speed and therefore resolution. (At low speeds high resolution is like watching paint dry!)

Regards

Ian

[polymorph 1]

I know I am not willing to give up the speed. I'm still in the process of tuning mine up and making improvements. I have Bertho's ball bearing pressure plate upgrade and have turned my speed up to 100mm/sec. It flies! Had some of OlyMEGA over last weekend, they are all familiar with the Prusa Mendel machines and were very impressed with the Ultimaker's speed.

At some point I am going to make this mod, too, and get rid of the short belts. I can hear a ticking sound in the rear left top corner, I think it might be the short belt riding up on the gear pulley teeth and snapping back down.

[alaris2 1]

aside from getting rid of the short belts which I hope to be able to do too at some point in the very near future

the next thing is to deal with the long belts. I've measured considerable backlash in prints caused by the belts (right now I can't be sure how much is accounted for by the short and how much by the long) even when the belts are so tight they're at breaking point. there comes a point where additional tightening of the belts actually makes things worse - there really is a sweet spot.

staying in that sweet spot is difficult - I find constant adjustments are necessary if you really care about surface quality.

but two interesting alternatives have presented themselves and might be worthy of discussion (other than the ballscrews)

1) synchromesh cable. it looks a bit like fishing wire (used by the Tantallus) but with a grip which stops it slipping. there are various printers trialling this and it seems to be fast, light and small.

one question I can't find an unbiased answer to is what the long term properties are like (does it end up stretching?)

2) GT2 belts instead of the MXL we use. these are designed for accurate motion, not just as timing belts and are being tested on repraps. the word is they are good and the difference in quality is noticeable.

again, can't find an unbiased answer because 'noticeable quality improvement' on a reprap isn't hard.

the price would also be high - we need 8 pulleys and 4 belts whereas their system uses half that.

[polymorph 1]

Yeah, there are definitely better belts out there. Not my field of expertise. Anyone have a link to a good website or webinar or something similar where I can learn more about the differences?

[snowygrouch 1]

 

http://www.gates.com/catalogs/index.cfm ... d=5339#ddm

Gates make very good belts, the only issue is finding the right catalogue !

FENNER being the more standard choice, but still a good make.

So what you are looking for is called Syncronous belt. So called because it has teeth and so has syncronisation

between drive and driven parts.

The standard ones seem to be "Inch Trapezoidal" the standard for which is "ANSI IP-24"

So american sizes, the question is if this standard of toothform is what we really need. Problem being that

if its NOT. Its also new sprockets time.

[owen 19]

Yeah, there are definitely better belts out there. Not my field of expertise. Anyone have a link to a good website or webinar or something similar where I can learn more about the differences?

Check this Post https://groups.google.com/forum/?pli=1#!topic/ultimaker/nZXP6dU7dE0for a place to buy, and Chuck Ernst who using some of the better ones.

[alaris2 1]

I made a note that for GT2, the following are known to work (but you can see from the price why I haven't tried it yet)

http://www.royalsupply.com/store/pc/Gat ... tm#details

http://www.royalsupply.com/store/pc/Gat ... p17193.htm

the synchromesh cable is here - much cheaper alternative

https://sdp-si.com/eStore/PartDetail.as ... roupID=768

I'm fairly sure that new pulleys are required in both cases - our current ones won't match the tooth profile.

[ianeagland 0]

Hi

Will the Synchromesh Cable really work? What stops the core twisting under load and effecting position.

You might like to try Misumi for belts. Their pricing is very competitive in the UK.

[yzorg 9]

For monting the Steppers on the back of the UM and rotating them downwards by 90degree...

I found these online..

https://sdp-si.com/eStore/Direct.asp?GroupID=1124

can these work for us or are they not precise enough?

[snowygrouch 1]

I dont know it depends on the quality of manufacture, but for me I want to put as few things as possible

between the stepper and the shaft. So I will stick with my ugly solution.

Bevel gears are very hard to set up to have no backlash, because you have 2 axis to align accurately instead

of just one. They are also straight cut gears, which will not transmit reversing directions very well (depends on the

contact ratio), but spiral gears would be better. Then the cost starts going way....up.

I think a better solution is to fit pancake stepper motors which are bigger diameter but much shorter.

http://www.haydonkerk.com/LinearActuato ... fault.aspx

 

[alaris2 1]

ian> good question, and I don't know the answer, but it apparently works. there are several new printers coming out using this method. I found it while investigating the fishing wire trick used by tantillus. how long it works is another matter!

yzorg> SG is correct, using bevels is not recommended, actually no gearing really is since it will just introduce additional points of failure or backlash. if the mod is considered too 'ugly' (personally I think it's just fine), pancake motors are the correct solution. let's face it - you have a big extruder sticking out the back of the UM already..

GT2 looks the best, but most expensive, overall solution - but I'm not ready to spend all that money just yet. I favor the direct drive as a 'does most of what you want' solution instead.

[codemaven 0]

I have a few meters of synchromesh cable that I bought to use on a CoreXY platform... if I ever get around to building it. I can't comment on how it works yet, but the stuff feels more sturdy than belts and I think it will work great.

Cheers,

Troy.

[bob 0]

So I went and bought everything that I needed for this... (for those of you in germany, you can get custom sized schafts for not too much at Maschinen-Werkzeuge.com) and then I was browseing thingyverse and found among others this:

http://www.thingiverse.com/thing:44078

anyone tried it with printed couplings? unfortunately I allready have the china couplings and they are failry good, but it would still interest me.

[snowygrouch 1]

I wouldnt,

Whoever drew those had absolutely no idea about how a flexible coupling works.

The whole point, is that they should offer small resistance to bending, but high resistance to rotation.

If you put those ones on, you will get massive angular misalignment every time you have a big acceleration

movement. Stay well away.

I also have no idea at all how you will get a good clamp to the shaft. These things go back to grub screws

basically, which is the worst possible solution. So, no - dont use them.

C.

[bob 0]

indeed a good point

 

[cora111 0]

I moved my Ultimaker to this modification and everything works very well. It also improved the print quality a bit. There are also some other advantages over the old belt drive system. The new direct drive system is a bit quieter. Also much (!) less force is needed to drive the rods and the print head. This results in colder stepper motors which improves their life span. Another advantage is that the motors are outside of the Ultimaker which makes it really easy to access all parts now: rods, belts, end stoppers, the motors of course, and it's easier to encase the build chamber with plexi glass panels. So this mod is great in combination with a heated build chamber which I will make in the next time.

Instead of using the original motor positions I swapped them, so the x-axis motor is on the left (the Ultimaker logo facing to yourself), the y-axis motor is on the back with the feeder and the filament spool. This setup looks much nicer. Please keep in mind that you have to swap the electric connections of the motors as well! Or you go with a custom firmware where X and Y is swapped.

Instead of these China couplers I used the 5-to-8 mm couplers from German RepRap. They are a bit larger (and more expensive), so the motor bracket needs to be adjusted. Just use a rough file and you are fine. I wasn't patient enough to wait for the China couplers so I went this way.

If you have a newer revision of the Ultimaker you have some black round spacers instead of the old wood spacers. The plastic spacers are great as they have the perfect length and also providing some sort of heat insulation between the motor and the bracket.

If you want me to post some photos of my setup just ask.

 

[snowygrouch 1]

Glad it worked well, providing you are very careful with alignement when tightening the bracket and motor bolts

there can be a big reduction in the force on the bearings, which will give a lot less friction.

Why dont you post a couple of pictures here anyway, might be interesting even if its not "required".

C

 

[IRobertI 521]

Don't listen to snowygrouch, pictures are always required

(no, I still haven't done the mod myself, all the parts are just lying there next to the machine, mocking me. I'm a pansy relying too heavily on the "if it aint broke" policy )