SD3 with "Z" Axis Ball Screw. (Page 1 of 2)

Any test prints (with pics) yet?

I went down this path but had to give up. I machined the 8mm end myself and rushed and botched it. This introduced a whole bunch of wobble into the prints which I couldn't dial out.

I might post up my pics in terms of what not to do later on.

Right now I'm running the Thingiverse torture test that MAKE: magazine did (http://www.thingiverse.com/thing:33902).

I was having some print problems before the screw installation which I still need to dial out.

I was having a problem with circles not being "circular", so I ran some tests using cubes.  It turns out my "Y" was short, so I changed the Y calibration from Y88 to Y90.50.  That got the squares square and the circles, nearly circular, but all of the sudden, I started having problems with the infill going all the way to the perimeters on small circles (3-4mm dia. like screw holes) and sharp corners.

For this test I increased the flow in Slic3r to .85 from the default .79, and so far it looks OK.

As soon as it's done, I'll post some pics.

You don't really have to spend a fortune on a Ball screw.  There are many inexpensive option out there.

The ball screw I bought only cost me $45, and that included shipping  I got it here:

http://www.ebay.com/itm/SFU1204-Ball-Sc … 4605057686

As I mentioned, I have a friend who owns a machine shop, so he machined the end down to 8mm, but some of the Chinese eBay vendors offer to machine the ends to your specs, for an additional fee (usually ~$20 extra).  So it might be worth looking into it.

There are also a couple of Korean vendors which specialize in used CNC equipment.  I bought 2 ISSOKU (Japaneses made) miniature ball screws which I plan on using in my CNC quality home made printer.  Those were pulled from a CNC machine so they already have machined ends.  While they are used, they are smooth "like butter".  You can hold the screw vertically by the ball nut, and the weight of the screw alone will get it spinning and rolling down.

Granted the Chinese made screws are not quite as slick, but they're 1000 times better than a plain 5/16" screw and nut.

Nice work.

I find that layer issue can be a couple of things; Extruder Stepper Steps per mm have changed (filament tension, mood, whimsy...), Retraction Settings need tweaking, Flow rate is different due to conditions.

Turn on/off retract on layer change to rule out retraction quickly...

Pirvan looks awesome. I had those filament issues because of retraction, and our default settings trying to run the extruder speed too high. lowered my max extruder speed to 15mm/sec and problem solved.

I tried turning off the retract on layer change, and that gave me a lot of strings and blobs.  So I tried nlancaster's suggestion about slowing down the retract speed.  Mine was set at 75mm/sec, so I put it down to the Slic3r default of 30mm/sec and that worked a lot better.

What I don't understand is why the print keeps starting in the same place, considering "Randomize Starting Points" is checked.

Well,  this one didn't go so well.  Some it looks fine, but there were lots of problems:

• Half way through the print, the legs broke off, and after that the printer was just oozing plastic in mid air, where the legs should have been.

• Even though the chin and the hands behind the head had support, it was pretty mangled up.

I think my problem has do do with the extruder fan.  This is something relatively new for my setup, and I had used it sparingly before with (relatively) good results.  But in all those cases, I had turned the fan on manually and only on certain layers with overhangs and/or thin extrudes.  This time around, since I left the printer going over night, I left the fan running through the whole print (no G-Code support yet).  And I think it contributed to the mangled printout.

However, as you can see from the the pics, the surfaces are pretty smooth so I'm happy with that, as this is what I was looking for when I added the ball screw.

I had the same thing. Not sure what is going on. That is my first time using that option since slic3r 9.10 (or that I actually paid attention).

The lady looks like she was coming out good too. What resolution did you print that on? (EDIT : nevermind. file name says .1 - I will learn to read one day)

OK, so I uploaded the STL file of the Girl fixing her hair to a public storage, you can download it from here:

http://7pnd2n.1fichier.com/en/

I uploaded the complete, uncut model, that way you guys can print it the your favorite way.  If you plan to cut it, I would cut it right at the waist, it's the easiest seam to hide.

Meanwhile here is a print I did a little while ago.

Here is a print of yet another girl.  Like the others, this is a 3DS model so it requires quite a bit of work to get it printer worthy, and even then it's nearly impossible to clean everything. 

On this model the hair was impossible to fix, because it literally has not volume/thickness, and many of the strands intersect each other.  Of this reason, the slicer has a very hard time dealing with it, and it comes out really crappy.  I tried Netfabb Studio, MeshLab, but none of these can do much about the hair.  They fixed 95% of the main model, but none of the hair... 

Perhaps some of us here that are well versed on this subject should start a thread on how to clean up models especially those that are familiar with Meshlab and its various filters.  I was able to figure some out, but I haven't scratched the surface yet, so I'd really like to see a good tutorial.

I also have started to add my own home made supports, since neither Slic3r nor Kisslicer can do the type of support I want.  The way this model was cut, it only needed support for the chin, hands, tips of the shoes, and back of the hair.  Slic3r would have wrapped the model in scaffolding all around, which would have been a bear to remove.  Kisslicer can't even handle the model, so it's useless for this type of work.

So my approach is to add my own strategically placed supports.  I start with a model of a very thin (.25mm) bar which I load along with the main model, then I rotate, stretch (or shorten) and then place it where it's needed, then duplicate it and add it in other places.  Yes, it intersects the model, but it provides the needed support, and since it's very thin, it's easy to remove and only leaves very small marks behind.  Here is a picture of the unsliced model on the Repetier platform.  Notice the supports under the hands, the back of the hair and the tip of the shoes.

Last but not least, this is a thread about the ball screw, so in terms of improvement, I think it helped quite a bit, and it really provided much needed rigidity.

I looked at the McMaster Carr page you linked to, and frankly, I would never spend that kind of money, you're looking at nearly $300 for the screw and nut, that is way to much to spend on this. 

Additionally, you are looking at imperial/SAE sizes.  The problem with that is that you'll have to deal with weird round-off  errors which occur in the translation from inches to millimeter.  In fact the is the reason why many change the "Z" rod from a 5/16 to a M6.  This way, the stepping values become whole numbers.  You'll no longer have to use values like 0.1058 for your layer height, just input .1000 or .3000.

Anyway, buying a ball screw at McMaster, even though I'm sure the quality would be very good, is prohibitively expensive, so I would go on eBay and buy one of the ones made on China.  You'll spend less than 1/3. 

The Ball screw I bought is a SFU1204.  That means it's 12mm diameter and has a 4mm pitch.  The 4mm pitch translates into a microstep value of 800 for the printer, and produces nice even layers without the round-off effects.

Here is the link to the ball screw I bought:
http://www.ebay.com/itm/SFU1204-Ball-Sc … 19d2677b00

The same guys sell pre-configured ball screws, but they are configured for CNC machines so they have a very long area that is turned down with special keyways, and you don't need that.  Technically, you need only about 5/8" (16mm) at each end machined down to 8mm so it fits the 5mm/8mm coupler.  The other end would fit nicely in a 608ZZ bearing (top bearing support block).

If you can find a coupler that supports 5mmx12mm, then you probably not need to do any machining other than cutting the rod to length.

Another thing you can try, is to contact the vendor before buying, tell them you want to buy the item, but want them to cut it to length and machine the ends to your spec before shipping.  Many of these guys will do it for and additional $20-30.

Whatever you decide though, get a metric ball screw, not SAE.  You'll also find that the SFU1605 is the most common, but that is simply HUGE, and the ball nut is also proportionally huge.  In this particular case, bigger is not better.  The 1204 is just right.  If you can find a reasonably priced 1005, that would be nice too, as the smaller screw also has a smaller nut.

BTW, assuming you want the same setup as what I have, the total length of screw you need is 11" (about 278mm).  On that, you machine each end 5/8" so you have a total of 9.75" threaded length on the 11" rod. 

Let me know if I can help with anything else.  If you need the adapters I made, I can post the STL.

Good luck.

If you can post the STL for the adapters, that would be great! I am going to give this conversion a try. Maybe Solidoodle should buy from these guys to begin with and save us all the trouble.