GearHeads Forum

Hi Kirk:

  Yes, the calibration can be pretty coplex. ..BUT once calibrated I think useing a known scale factor as the final step
is a valid one. Since there are so many variables, the system is likley to be off a small amount, such as mine is with that
htd3 , but the scale factor is liek a final step in the sizing of whats going to happen based on final calibration results.

  I was aware of most of the links information, but not all, I appreciate the link. For inside holes and such GT uses about
300 vertices to reduce the amount of vertex wrap on inside holes, and about 3000 points on outside contours for the same reason.
This is one of the reasons not all models are the same quality. Then theres the printer softwares capability to do arc fitting
on contours which can fix both inside and outside poly's.

  I think its mostly software that differentiates the various printers, and its getting better all the time. Im watching it morph
with interest.....

Art

Art:

Calibration actually involves two things.  First you have to calibrate your printer so the head moves accurately, then you have to calibrate the width of your extruded filament, i.e. your extrusion rate, so it's actually what your slicing program assumes.

With commercial pulleys and belts (screws, whatever) you know the real numbers and calculate what the correct step size is.  For reprap style printed parts, they're often not quite right, and you have to take out your dial gauge and accurately (well, as accurately as you can) measure the actual movement and adjust your step size.

Extrusion calibration takes into account how far the extruder moves the filament - usually governed by how far the hobbed bolt (or other displacement type filament driver) digs in.  It's easiest to just measure how far the filament moves.  You'll have to measure the diameter of your filament, which can change from roll to roll (or within a roll if you're unlucky), at several places and average it.  Since it can be oval you'll have to take that into account - usually you're told to average it around the diameter  You're really trying to determine the circular equivalent of the oval cross sectional area, but the average seems to work.  I've forgotten how to figure an oval area, anyway.  After all this you print a single wall object and find the average thickness of the actual extrusion, and if it's not right there should be a fudge factor you can put in the slicing program.  If not, you can fudge the filament diameter.  There are some people who use cross sectional area to set their extruder and filament, but I don't use or understand this technique.

The common practice of printing something, measuring it, and adjusting the appropriate step rate only works for that one size because you're compensating for two variables.  It is a good confirmation the other steps were right, but not a good calibration tool.  Good enough for Mr. Jaws, not so good for gears.

Inside holes have some other problems.  [url=http://hydraraptor.blogspot.com/2011_02_01_archive.html]This[/url] says it better than I can.

You can fudge the scale to get it right, but if you get everything tweaked just right you won't have to.

Hope this is helpful.  I've spent way to much time reading about and fooling with this stuff.

Kirk

Dan:

  I havent run enough testing to say for sure. What I suspect is its a matter of the printer itself and the current calibration of that printer. Since HTD's are soooo sensitive ( any small timing belt pulley really..), Im thinking the same cal factor will work on all models until you recalibrate a printer. I printed 3 to find that scale %. I suspect its good for only my machine, numbers will vary between units. Once you know it, I think you can then just use that scaling value in future. I'll know that after a few dozen more.  :)
  Since Im pretty sure the stl's are accurate to several decimal points, Im getting pretty confident that the scaling is the way to ensure your printer will do a good gear. Get a timing pulley to work and you'll KNOW your printer is now cal'ed for gears. That linear accumulation of error is a great way to test for that, if a pulley fits a belt, you have figured out your machine. Smaller the belt, the tighter your calibration is..

My printer just failed me twice today.. seems noise or something as the platform just dropped to the bottom , but after about 50 models, thats not a bad failure rate. Think Ill make a servo driven one that a bit more robust.. :)

  Art

Art,

That shrinkage you mention, is it a constant for a given printer-material combination that could be taken note of and implemented on any print? Or is it merely specific per given STL file?

Dan

Ken:

  Yes, and no. My printer, and I suspect most printers, rely on the STL triangulated model
to do its placement of material. All the programs I have seen have scaling, which in essense is a form of "Comp light" in that while it can scale the materials to the right size, if you increase an items outside dimension you are also scaling the inside dimension. Its why mathmaticallly correct models are a necessity
if your doing any mechanical type of work, the sizing should be proper and then the printer calibrated to proper size.
 Even then, the issue of shrinkage and resolution of the printer play a strong part. The T5 pulley for
example sems to work very well, but the belt had to be run back and forth to help "run in" the gear and remove ticks and aberations unseen by the eye. Pulleys are probably one of the lowest tolerance items,
while gears are high tolerance. Slight backlash is ually available in a gear, while a pulley has "linear growth" of error. For example, if a pulley tooth on an HTD belt is 3mm, and the printer prints
3.05mm's after 10 teeth of wrap your already .5mms out, 1/6th of a tooth and the belt will no longer sit
on the pulley. The wrapping of a pulley by the belt makes a pulley very suceptable to this linear error
accumulation. My HTD pulleys, ( the first two ) dont work after about 10 teeth, scaling the model in the
printer makes it much better but you have to print several shrinking as you go to find your sweet spot. Once you know the scaling factor to use though, you can resuse it on all pulleys since that scale is actually the variance of real world to printer world accuracy.

So quick answer, yes and no. :)

Art

Hi Art,
Does your 3D printer have any offset capability, kind of like cutter comp on a CNC?


Ken

Hi Ken:

  Ive found the timing pulleys are the hardest to do accurately, the T5 seems to work great, the HTD3 however wont run in properly, but with a very small 3mm tooth profile it doesnt have to be out much not to work. I am going to try shrinking the print slightly as I can see the teeth are very slightly too large, after 8 teeth or so the belt begins to rise out of the mesh.. a sign of a small error..

Ill let you know how the HTD's work out, but Im pretty happy with the T5's..

Art

Sure look great, if time permits one will be setting on my desk.

Thanks for posting the sample prints.

Hi Guys..

One more. A T5 pulley made to see if I could match a belt laying about in my shop.
This took 9 minutes, was printed on low res, fast mode, and seems to fit fine. As you can see its pretty
damn small..

Art