Our Mendel90 Build Experience and Mods
Posted by SanjayM
|
Our Mendel90 Build Experience and Mods July 17, 2012 08:08PM |
For the past few months me and a friend have been severely bitten by the RepRap bug, 5 months later,and after having built 4 different printers (1 kit, 3 self-sourced) and having done quite a lot of research and testing we thought we'd share on the forums with a bit of a review of the build process for a Mendel90 and hopefully incite some discussion about the design in general.
Firstly, a big thank-you to Nophead, we LOVE this design, when building one the amount of thought that has gone into this design really shows. The only real way to experience this is to put one together - everything just makes excellent design sense. I especially recommend his blog post entitled Mendel90 axes [hydraraptor.blogspot.co.uk]
Here is our latest build:
(Bit of a plug - The machine pictured is for sale here: [forums.reprap.org] )
First thing we noticed was how much more easily built this machine is than a threaded rod design. The 4 sheets fix together in a jiffy and the axes simply screw onto those. No aligning needed, no spinning nuts hundreds of mm along stud, no putting the frame together then realising you forgot a washer somewhere so have to take it all down to get it on. Things fix to and come off the sheets in seconds, all you have to get precisely right is a few simply right angled cuts. This also makes modding/maintaining/making changes really fast and easy.
Faults are quite easily found, you can get your head right in the machine with the open frame and see what's really going on.
The sheet also makes routing cabling really trivial, just screw it down to the sheets with printed clips as you like. It also means adding bits, mounting PCBs etc is similarly trivial. We intend to add a click-encoder and LCD screen controller in time, placing parts like this is dead simple.
Nice easy tidy cable management
The use of ribbon cable is fantastic! Do go for the proper multi-core rainbow coloured stuff, it takes a shocking amount of current (2A/strand) and makes organising what wires go where almost pleasurable! It loops really nicely to the moving axes and seems to enforce its own minimum bend radius quite nicely.
What to make the sheets of is an interesting question. We started with 12mm MDF, this works well mechanically, very strong!
We can stand on our machine!
We seal our MDF sheets with MDF sealant, and add a coat of paint to make it pretty. All fine, but all those coats of paint and dry time adds a non-trivial amount of effort. We looked at acrylic/plexiglass sheet, its certainly pretty, but it scratches, cracks and splits from experience. It is also very expensive compared to MDF. We've been looking at various sheet materials and been to some sheet plastic warehouses, what we have identified and are switching to is rigid pressed PVC sheeting (9mm), this is incredibly durable feeling stuff, super rigid and strong. It should take woodscrews with a pilot hole for a very strong hold, or can be tapped for machine screws. It's also heavy, which is a double edged sword with regards to portability vs stability. Price is quite good, we can get 4-5 machines out of an £80 sheet, so £16-20 a machine.
With regards to the printed parts, some could be quite challenging if you struggle with bridges - the X-motor-mount has some pretty big bridges going on. We printed our machines in PLA and this works just fine, even though I think nophead had ABS in mind. We sliced all our parts with slic3r, there were some issues, but I don't regret going slic3r over skeinforge.
Caveats! - The tiny parts that clamp the belt to the Y-Belt-Anchors are non-manifold according to slic3r, at least in the STL supplied in the 'printed' folder of the mendel branch from github. - They slice wrongly, and somehow seem to 'infect' other parts in the plate. This is solved by printing them separately from their individual STLs and not the supplied plates. No big deal, but odd? Is this a slic3r bug or a bad SCAD export?
Secondly, slic3r 0.8.4 handles holes and insides of arcs etc differently it seems, making holes smaller. This manifests itself quite insidiously and subtly - causing the inside of the C-clamp that holds the Z axis LM8UU's to be printed slightly too tight. - This in turn causes the LM8UU to sit slightly too spaced away from the nut-trap, which makes the Z-Threaded-rods and the Z silver steel sit out of parallel, slightly. This problem took a long time to track down!
We have switched from M8 Stud for the Z axis to M5 stud, this lets us couple the 5mm motor shaft to the M5 stud with just some vinyl tubing pushed over each and tightened with a jubilee clip. This gives great flexible coupling and eliminates Z wobble. It also gives you more E-steps/mm on Z at the expense of feedrate.
Due to nophead using slightly different homing setup on his custom electronics we have rearranged the endstops a little, to get the home position to be bottom-left-front as most reprap software expects. We do this by arranging the Y bar clamps so the switch is at the back. and we have taken then homing switch off the X-motor-housing, and added a place for it on the X-Idler. We'll get the STLs for this little mod on thingiverse soon.
New EndStop Positions Circled
Out of cheapness/convenience we have consolidated fixing hardware to pozi-drive pan head machine screws and strayed from the prescribed BOM a little. Some things are a little awkward to do up, but its quite manageable really. One change we did have to make due to this stems from the fact that it is impossible to do up the bar-clamp of the X-ends nearest the frame. To get around this we just drilled an 8mm hole through the gantry through which you can poke a screwdriver to reach the screw that tightens that bar-clamp.
Right, it's getting late, and that's all I can think of!
Next post will be on our hot-end - All metal, no PTFE/PEEK, just brass, A2 stainless and aluminium, very strong, simple and should hit obscene temperatures for polycarbonate etc. It's a beauty!
Firstly, a big thank-you to Nophead, we LOVE this design, when building one the amount of thought that has gone into this design really shows. The only real way to experience this is to put one together - everything just makes excellent design sense. I especially recommend his blog post entitled Mendel90 axes [hydraraptor.blogspot.co.uk]
Here is our latest build:
(Bit of a plug - The machine pictured is for sale here: [forums.reprap.org] )
First thing we noticed was how much more easily built this machine is than a threaded rod design. The 4 sheets fix together in a jiffy and the axes simply screw onto those. No aligning needed, no spinning nuts hundreds of mm along stud, no putting the frame together then realising you forgot a washer somewhere so have to take it all down to get it on. Things fix to and come off the sheets in seconds, all you have to get precisely right is a few simply right angled cuts. This also makes modding/maintaining/making changes really fast and easy.
Faults are quite easily found, you can get your head right in the machine with the open frame and see what's really going on.
The sheet also makes routing cabling really trivial, just screw it down to the sheets with printed clips as you like. It also means adding bits, mounting PCBs etc is similarly trivial. We intend to add a click-encoder and LCD screen controller in time, placing parts like this is dead simple.
Nice easy tidy cable management
The use of ribbon cable is fantastic! Do go for the proper multi-core rainbow coloured stuff, it takes a shocking amount of current (2A/strand) and makes organising what wires go where almost pleasurable! It loops really nicely to the moving axes and seems to enforce its own minimum bend radius quite nicely.
What to make the sheets of is an interesting question. We started with 12mm MDF, this works well mechanically, very strong!
We can stand on our machine!
We seal our MDF sheets with MDF sealant, and add a coat of paint to make it pretty. All fine, but all those coats of paint and dry time adds a non-trivial amount of effort. We looked at acrylic/plexiglass sheet, its certainly pretty, but it scratches, cracks and splits from experience. It is also very expensive compared to MDF. We've been looking at various sheet materials and been to some sheet plastic warehouses, what we have identified and are switching to is rigid pressed PVC sheeting (9mm), this is incredibly durable feeling stuff, super rigid and strong. It should take woodscrews with a pilot hole for a very strong hold, or can be tapped for machine screws. It's also heavy, which is a double edged sword with regards to portability vs stability. Price is quite good, we can get 4-5 machines out of an £80 sheet, so £16-20 a machine.
With regards to the printed parts, some could be quite challenging if you struggle with bridges - the X-motor-mount has some pretty big bridges going on. We printed our machines in PLA and this works just fine, even though I think nophead had ABS in mind. We sliced all our parts with slic3r, there were some issues, but I don't regret going slic3r over skeinforge.
Caveats! - The tiny parts that clamp the belt to the Y-Belt-Anchors are non-manifold according to slic3r, at least in the STL supplied in the 'printed' folder of the mendel branch from github. - They slice wrongly, and somehow seem to 'infect' other parts in the plate. This is solved by printing them separately from their individual STLs and not the supplied plates. No big deal, but odd? Is this a slic3r bug or a bad SCAD export?
Secondly, slic3r 0.8.4 handles holes and insides of arcs etc differently it seems, making holes smaller. This manifests itself quite insidiously and subtly - causing the inside of the C-clamp that holds the Z axis LM8UU's to be printed slightly too tight. - This in turn causes the LM8UU to sit slightly too spaced away from the nut-trap, which makes the Z-Threaded-rods and the Z silver steel sit out of parallel, slightly. This problem took a long time to track down!
We have switched from M8 Stud for the Z axis to M5 stud, this lets us couple the 5mm motor shaft to the M5 stud with just some vinyl tubing pushed over each and tightened with a jubilee clip. This gives great flexible coupling and eliminates Z wobble. It also gives you more E-steps/mm on Z at the expense of feedrate.
Due to nophead using slightly different homing setup on his custom electronics we have rearranged the endstops a little, to get the home position to be bottom-left-front as most reprap software expects. We do this by arranging the Y bar clamps so the switch is at the back. and we have taken then homing switch off the X-motor-housing, and added a place for it on the X-Idler. We'll get the STLs for this little mod on thingiverse soon.
New EndStop Positions Circled
Out of cheapness/convenience we have consolidated fixing hardware to pozi-drive pan head machine screws and strayed from the prescribed BOM a little. Some things are a little awkward to do up, but its quite manageable really. One change we did have to make due to this stems from the fact that it is impossible to do up the bar-clamp of the X-ends nearest the frame. To get around this we just drilled an 8mm hole through the gantry through which you can poke a screwdriver to reach the screw that tightens that bar-clamp.
Right, it's getting late, and that's all I can think of!
Next post will be on our hot-end - All metal, no PTFE/PEEK, just brass, A2 stainless and aluminium, very strong, simple and should hit obscene temperatures for polycarbonate etc. It's a beauty!
|
Re: Our Mendel90 Build Experience and Mods July 18, 2012 05:01AM |
I notice you haven't used the ribbon cable for the bed. I believe the arrangement you have used, with a thick cable and nothing to stop it flexing were it meets the carriage will lead to it burning eventually. I have had a couple cables burn doing that. What happens is when you have a thick cable it needs a much bigger minimum bend radius. I think it is a cube or fourth power law. If that is not enforced over the full length of the cable you get more bending at the ends. The strands break one by one and the remaining ones then overheat and melt / burn the insulation.
The ribbon cable is many many times more flexible but I also recommend the polypropylene strips. Without them it will bend more at the ends than the middle and this effect multiplies as the flexing weakens the bit that flexes most. It may look fine at the moment but I have run machines 24/7 for a couple of years and that is when these things start to fail.
> With regards to the printed parts, some could be
> quite challenging if you struggle with bridges -
> the X-motor-mount has some pretty big bridges going on.
I have a new version of that part with a removable support structure that I will release soon.
>
> Caveats! - The tiny parts that clamp the belt to
> the Y-Belt-Anchors are non-manifold according to
> slic3r, at least in the STL supplied in the
> 'printed' folder of the mendel branch from github.
> - They slice wrongly, and somehow seem to 'infect'
> other parts in the plate.
According to Netfabb it is the anchors themselves, not the clips, that are non-manifold. It is due to the edge of the nut trap being coincident with the face of the webs. Again that is fixed and will be in the new release.
> Secondly, slic3r 0.8.4 handles holes and insides
> of arcs etc differently it seems, making holes
> smaller. This manifests itself quite insidiously
> and subtly - causing the inside of the C-clamp
> that holds the Z axis LM8UU's to be printed
> slightly too tight. - This in turn causes the
> LM8UU to sit slightly too spaced away from the
> nut-trap, which makes the Z-Threaded-rods and the
> Z silver steel sit out of parallel, slightly. This
> problem took a long time to track down!
I would not use Slice3r for anything that has to be geometrically correct at the moment. Skeinforge has been getting this right for years now.
> Due to nophead using slightly different homing
> setup on his custom electronics we have rearranged
> the endstops a little, to get the home position to
> be bottom-left-front as most reprap software
> expects. We do this by arranging the Y bar clamps
> so the switch is at the back. and we have taken
> then homing switch off the X-motor-housing, and
> added a place for it on the X-Idler. We'll get the
> STLs for this little mod on thingiverse soon.
The original scheme works fine with standard electronics and firmware and avoids a set of wires to the idler end. All it means is the front of the machine is actually the back. Alternatively the firmware can be configured as having max endstops instead of min.
> Next post will be on our hot-end - All metal, no
> PTFE/PEEK, just brass, A2 stainless and aluminium,
> very strong, simple and should hit obscene
> temperatures for polycarbonate etc. It's a beauty!
>
I look forward to seeing that as I also think all metal is the way to go and in fact pioneered it many moons ago.
[www.hydraraptor.blogspot.com]
The ribbon cable is many many times more flexible but I also recommend the polypropylene strips. Without them it will bend more at the ends than the middle and this effect multiplies as the flexing weakens the bit that flexes most. It may look fine at the moment but I have run machines 24/7 for a couple of years and that is when these things start to fail.
> With regards to the printed parts, some could be
> quite challenging if you struggle with bridges -
> the X-motor-mount has some pretty big bridges going on.
I have a new version of that part with a removable support structure that I will release soon.
>
> Caveats! - The tiny parts that clamp the belt to
> the Y-Belt-Anchors are non-manifold according to
> slic3r, at least in the STL supplied in the
> 'printed' folder of the mendel branch from github.
> - They slice wrongly, and somehow seem to 'infect'
> other parts in the plate.
According to Netfabb it is the anchors themselves, not the clips, that are non-manifold. It is due to the edge of the nut trap being coincident with the face of the webs. Again that is fixed and will be in the new release.
> Secondly, slic3r 0.8.4 handles holes and insides
> of arcs etc differently it seems, making holes
> smaller. This manifests itself quite insidiously
> and subtly - causing the inside of the C-clamp
> that holds the Z axis LM8UU's to be printed
> slightly too tight. - This in turn causes the
> LM8UU to sit slightly too spaced away from the
> nut-trap, which makes the Z-Threaded-rods and the
> Z silver steel sit out of parallel, slightly. This
> problem took a long time to track down!
I would not use Slice3r for anything that has to be geometrically correct at the moment. Skeinforge has been getting this right for years now.
> Due to nophead using slightly different homing
> setup on his custom electronics we have rearranged
> the endstops a little, to get the home position to
> be bottom-left-front as most reprap software
> expects. We do this by arranging the Y bar clamps
> so the switch is at the back. and we have taken
> then homing switch off the X-motor-housing, and
> added a place for it on the X-Idler. We'll get the
> STLs for this little mod on thingiverse soon.
The original scheme works fine with standard electronics and firmware and avoids a set of wires to the idler end. All it means is the front of the machine is actually the back. Alternatively the firmware can be configured as having max endstops instead of min.
> Next post will be on our hot-end - All metal, no
> PTFE/PEEK, just brass, A2 stainless and aluminium,
> very strong, simple and should hit obscene
> temperatures for polycarbonate etc. It's a beauty!
>
I look forward to seeing that as I also think all metal is the way to go and in fact pioneered it many moons ago.
[www.hydraraptor.blogspot.com]
|
Re: Our Mendel90 Build Experience and Mods July 25, 2012 08:27AM |
rhmorrison Wrote:
-------------------------------------------------------
> Dark Alchemist Wrote:
> --------------------------------------------------
> -----
> > I have the Nema 23's already and no Nema 17's
> from
> > my CNC project. My problem is getting the
> mounts
> > for the 23's as everything I have seen on the
> net
> > has the mounting for 17 and not 23.
>
> Take a look at Andis PowerMendel.
Google translator here I come.
edit: Rock on but it takes a Mendel to make that.
Edited 1 time(s). Last edit at 07/25/2012 08:29AM by Dark Alchemist.
-------------------------------------------------------
> Dark Alchemist Wrote:
> --------------------------------------------------
> -----
> > I have the Nema 23's already and no Nema 17's
> from
> > my CNC project. My problem is getting the
> mounts
> > for the 23's as everything I have seen on the
> net
> > has the mounting for 17 and not 23.
>
> Take a look at Andis PowerMendel.
Google translator here I come.
edit: Rock on but it takes a Mendel to make that.
Edited 1 time(s). Last edit at 07/25/2012 08:29AM by Dark Alchemist.
Sorry, only registered users may post in this forum.