Prusa github stl hole sizes too large?

Having got my repstrap printing with reasonable accuracy, I have moved on to the next stage - starting to print a Reprap, in my case a Prusa. I downloaded the master from github, and printed the first makerbot plate as it is similar in dimensions to the abilities of my repstrap.

This plate has the main vertices for the frame, three belt-clamps and a bar-clamp. I have checked that the size is the same as the stl, and checked the sizes of the stl in Pleasant3D, an stl viewer. My problem: all holes seem to be 9mm rather than 8mm. The holes in the frame vertices are a loose fit for 8mm threaded rod, but are fine once the nuts are done up on the rods, but the bar-clamp has to bend a long way to clamp an 8mm threaded rod. I've looked at some of the other plates, and the individual stls, and they all seem to be the same - the holes in the original stls are all 9mm. Am I missing something here?! I created the gcode in Skeinforge 41, no scaling, used Repsnapper to send the gcode to Gen6 electronics. My repstrap is well calibrated, and has reproduced very accurately the stl; it's just the stl seems too big! Should I scale them down until they are the right size? Why aren't they the right size to start off with?

I found exactly the same. The holes in the STLs were miles too big when printed on my machines. I had to change the config file and print the calibration piece to get them correct.

These are the values I use:

//[m8_dia, m8_nut, m4_dia, m4_nut, m3_dia, m3_nut, bush_dia, mot_shaft, bush_rod, bush_outerdia, bush_length]
  [8.6,    15.4,   4.8,   8.5,     3.8,    7,      11,       5.2,       8.2,      16,            11],//metric

I have no idea what the correct SAE sizes should be as Googling SAE doesn't turn up anything useful.

It would seem that Josef's machine suffers from very bad segment pausing so holes come out a lot smaller. I find the same problem with a lot of STLs on thingiverse. At least with openscad you can correct it easy enough.

---

[www.hydraraptor.blogspot.com]

Thanks nophead, I thought I was going mad. I can't believe there aren't stls that are correctly sized. Hardly noob friendly. Guess I'll have to start learning Openscad.

Its because people tend to tune their models to suit their machines, rather than tuning their machines to make accurate renditions of their models.

---

[www.hydraraptor.blogspot.com]

Quote
nophead
Its because people tend to tune their models to suit their machines, rather than tuning their machines to make accurate renditions of their models.

Because of this Prusa has a configuration object.
Print out the object, check for size, make changes to configuration and then generate STL's for printing.
See this video which explains the process.

The idea is Super, since many people do not have their machines properly configured such that an 8mm hole in the model is really 8mm you simply change a few settings (based on your printed calibration object) and THEN generate the STL's for the Prusa RP parts to print.

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

Quote

The idea is Super

Sorry, Bob, that's a kludge. I still can't believe people don't even think about calibrating their machines to produce accurate measurements. Clearly it can be done (and I've done it for non-RepRap machines), but there's not even a suitable procedure in the Wiki.

Hint: a single calibration cube isn't sufficient, you need a small and a big one to measure scaling errors as well as path offset.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

BUT... Since MOST people don't properly configure their machines it is a simple and easy way for RepRappers who do not have properly configured machines to still print a Prusa Mendel set with accurate holes. I had to regenerate some STL's because the holes were way too big...

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

Quote
Traumflug
but there's not even a suitable procedure in the Wiki.

YES, that is the main problem!

My dream was to have a Skeinforge calibration function that would print out an object with different values of Skeinforge on one print plate such that one picks the best object (row and column) and then can enter the perfect value for a specific parameter. Such a thing would be very user friendly...

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

As a noob, having researched, designed and built my repstrap, learnt how to use Arduino firmware software, calibrated my repstrap, grappled with Repsnapper and Skeinforge, the prospect of learning Openscad is not overly troubling. Thanks for the link to the video Bob, I will have a look at it when I have time and create my own stls.

However, for many that don't have my time and patience, it is hardly "a simple and easy way", and not being able to produce working prints is a failure at the final hurdle. Many people are buying partially-built or complete machines, as printing is what they want to do, and using a CAD program to generate stls to send to Skeinforge to upload the gcode in Repsnapper to send to the calibrated firmware to print is a convoluted tool path that only technically-minded individuals will follow. I only have to look at my 11-year-old nephew to know this; he would love to build his own, and is interested in the output, but the steps in between are beyond his ability and patience presently. Just like when using computers, most people don't program them; they just want to use other peoples' programs to do what they need to. An analogy I have used is that the reprap community is like the homebrew computing community of the late-70s; people built and programmed their own back then, which is why it wasn't for everyone. It is now, because of ease of access to consistent results. I agree with you all, that people should have properly calibrated machines. To have a reprap in every home, they need to be produced to a baseline calibration, so the model is the exact size the physical object is intended to be and the reprap is calibrated to reproduce this. Circulating stls that are randomly sized for arbitrary machines is meaningless. Circulating accurate CAD files is better, but the hoops the user has to jump through to get them printing are a barrier to mass-uptake. Anything in between opening an accurate model in a software package and printing it, with as little human interaction as possible, is essentially a kludge!

Incidentally, I'm not intending to criticise this nascent community, just a little frustrated at the gradient of the learning curve. But excited that I'm getting good results. Over time, I'm sure better software solutions will percolate from the high end down to the homebrew community. Many thanks for everyone's help!

droftarts Wrote:
> An analogy I have used is that the reprap community is like the homebrew computing community of the late-70s; people built and programmed their own back then, which is why it wasn't for everyone.

We can agree this is a necessary step on the road to be accessible to everyone. The problem is we're not quite there yet, but some people seem to think that, and complain a lot about it. It's not bad to want it to be available to all, but RepRap is a DIY project, and as with all DIY you have to get your hands dirty. You might even go so far as saying you have to WANT to get your hands dirty. Some people don't want that and maybe, just maybe RepRap isn't for them?

It's kind of like on linux, if you want the newest cutting edge programs, you have to compile from source, if you (like me) don't know how, you have to use the what's pre compiled.

Also, I don't see how a motivated 11 year old couldn't put together a RepRap if he had some guidance along the way.

>Circulating accurate CAD files is better, but the hoops the user has to jump through to get them printing are a barrier to mass-uptake. Anything in between opening an accurate model in a software package and printing it, with as little human interaction as possible, is essentially a kludge!

Absolutely. That's why projects like [cloudscad.com] is interesting.

About the 9mm vs 8mm holes: I find it much easier to assemble a machine if the holes are 0.5-1mm larger than the rod, and I have seen no difference in the end results.

---

--
-Nudel
Blog with RepRap Comic

Quote

The problem is we're not quite there yet, but some people seem to think that, and complain a lot about it.

Very true. People simply compare a RepRap to e.g. a Stratasys, because both can be bought. At least it looks like that.

It's even likely we don't get accessibility to everyone anytime soon, because nobody is actually working towards that goal. People work on solving their own problems, on realizing their own ideas. If developers ever work on common accessibility, they do it to get rid of those mentioned complains.

It's even disencouraging to have a open "just works" design, because the 3D copyshops will pick that up and found their business on it. Developers invest time to do development, copyshops invest time to do marketing. Guess where the money goes.

After all, RepRap is a development project and a very interesting marketing experiment.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Fair enough, I take all your points on board. Personally, I don't have a problem with it, I was just grumbling that people post models that are machine specific. I realise there are lots of hoops to jump through when you are an early adopter. My nephew read this thread and said he still wants to build one, so I guess I'll be the one to guide him! He'll pick it up quickly anyway.

droftarts, glad I didn't disencourage you. I just realize they can look pretty harsh. My thoughts here are more on the philosophical side than a comment to your personal project.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Second that Traumflug. My post was also not meant to be harsh, and is not intended to point at any specific person, but more at general principles and views I have observed.

One thing we can probably all agree on, is it does take a LOT of time to get into every facet of RepRap. And I do think it would be impossible to put all that knowledge into one single guide/book/library, especially since new things are happening and changing every day. We do have the wiki, and I guess we (as the individuals in the community) could be better at updating the wiki, which would solve some of the issues.

My attitude is to learn new things when I need it. Do the toolpath I currently use work adequately? Fine, no urgent need to change it, or learn new programs. Do a smd rs485 IC need to be fixed? Well, I have to learn to solder SMD components, and so forth. It's impossible to learn everything before you begin, and if you try it, much of the information will be dated when you need it.

---

--
-Nudel
Blog with RepRap Comic

No worries, Traumflug. I have printed the calibration object, watched the video, and downloaded and installed Openscad, and now need to change the calibration and start creating my own stls. As I already have a reasonable working repstrap, it would be very difficult to discourage me when I'm this far down the road!

Quote

YES, that is the main problem!

The good thing about problems is, one can start o solve them: [reprap.org]

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Traumflug Wrote:
-------------------------------------------------------
> The good thing about problems is, one can start o solve them: [reprap.org]

A good start, Traumflag! If we can get the experience of other seasoned reprappers on that page, it would be great! *nudge* nophead? *nudge* (unless it's already written up somewhere else - link/copy onto the wiki page?)

I am not sure what I could add. The main reason holes come out way too small is due to segment pausing, which I have never suffered from because I use 100Mb Ethernet instead of serial over USB..

Even if the machine is perfectly calibrated, holes will still come out too small. I fix that in my software using the technique here: [hydraraptor.blogspot.com]. To fix in in SF I think you need to use the Stretch module, but I have never got to grips with that.

---

[www.hydraraptor.blogspot.com]

Quote

The main reason holes come out way too small is due to segment pausing, which I have never suffered from because I use 100Mb Ethernet instead of serial over USB..

I consider it a wide spread myth RepRap electronics suffer from too low hardware communications speed. For example, Teacup buffers commands in it's own RAM and can feed that buffer with some 40 lines of G-code per second over the serial line. It's virtually impossible to drain that buffer, short of sending heaps of zero-movements.

Quote

Even if the machine is perfectly calibrated, holes will still come out too small.

If holes come out to small, calibration isn't perfect Your blog post has a number of quite good observations, let's hope they go into standard RepRap software soon.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

Traumflug Wrote:
-------------------------------------------------------
> I consider it a wide spread myth RepRap
> electronics suffer from too low hardware
> communications speed. For example, Teacup buffers
> commands in it's own RAM and can feed that buffer
> with some 40 lines of G-code per second over the
> serial line. It's virtually impossible to drain
> that buffer, short of sending heaps of
> zero-movements.

You can easily exceed 40 segments per second when extruding a curve at say 50mm/s. Also serial over USB can have big latencies, so if you wait for the OK before sending the next line the through put is much lower than the baud rate would suggest.

I am not saying it is impossible to make it work with the current hardware, but a lot of people are using firmware / host combinations that cause holes to be much smaller than they should be. That is why Thingiverse has a lot of STL files where the holes are far too big.

> If holes come out to small, calibration isn't perfect

No that is not true. You can calibrate the machine for exact XY motion and compensate for plastic shrinkage and eliminate segment pausing and holes will still come smaller than the STL. You have to compensate in software for the fact that when you drag filament in a circular path it makes a circle smaller than the path the nozzle takes.

---

[www.hydraraptor.blogspot.com]

Here's the configuration.scad numbers I ended up with:

//[m8_dia, m8_nut, m4_dia, m4_nut, m3_dia, m3_nut, bush_dia, mot_shaft, bush_rod, bush_outerdia, bush_length]
[8.2,15.5,4.5,8.6,3.5,6.8,11,5.2,8,16,11],//metric

Not too far off nopheads. I have the M8 hole diameter small so I can drill it out. As I have a Bowden cable, I always end up with a blob on each layer as it stops and retracts (perhaps I should change the path order from Loops > Perimeter > Infill to Perimeter > Loops > infill?), so have to drill it out anyway. I don't seem to have observable segment pauses, but looking at the stls and the path Skeinforge produces, I would say there are 13 facets on a 8mm and 6 facets on a 3mm hole.