Build manual feedback

Hello all,

I am just starting to build my Mendel90, and I thought I'd post feedback on the build manual here as I find things.

1. More detail on degreasing would be helpful. Is it enough to just clean with windex and a papertowel? That's how I degrease the mirror after my kid smears it up.
2. When assembling the base, it would be nice to have a plan view of the bottom piece of dibond showing which hole is for what. When first getting started, all the holes look alike and it's a bit of a puzzle to get going. Ever after it will be easier because you'll only be looking at the holes in one little zone of the machine (and the holes will start getting filled up!)
3. The location of the second ribbon cable clamp is not visible in the instructions. You need to notice that there are two screws going into the gantry in the upper left and guess that the clamp goes on the other side.

Maybe more as I do more steps. I'll just keep posting in this thread.

-jeff

1. I am not familiar with windex but if it no longer feels greasy to touch then it is decreased. I use meths (denatured alcohol) but I think other solvents like acetone, isopropyl and white spirit should work. Also emulsifiers like washing up liquid would work, but being water based you would need to dry the rods well to prevent them rusting.

2. That is why the instructions don't start with fastening things to the base. Once you make up the gantry and stays and then offer them up to the base the holes to use become obvious. So far I have seen at least two videos where people screwed the fixing blocks to the base first. I don't know why as that makes it much harder to tighten the screws.

3. I will add another view from the font.

Thanks for the feedback.

---

[www.hydraraptor.blogspot.com]

It is still not clear for me what diameter of theraded rods are used and what is preferable.
STLs in dibond folder, which I believe the final version is, are for 8mm threaded rod, but STLs in mendel folder are for 6mm. So I am a bit confused what is better to print. Couplers seem to be all 6mm...

I think you have it the wrong way round. The Sturdy and Mendel versions use 8mm and the Dibond version is 6mm. One end of the Z coupling changes to match,

---

[www.hydraraptor.blogspot.com]

Ok Thank you
What is the purpose of two separate "corners" in x_motor_bracket.stl ?

They make it easier to print: [hydraraptor.blogspot.co.uk]

---

[www.hydraraptor.blogspot.com]

Regarding (2), the manual did a good job of explaining that I should get the vertical pieces assembled first, no problem. The tiny problem came trying to understand where the front/back of the base plate was, where the metal bars connected, etc. It was just a process of elimination, but I found myself looking for a 2d, plan view of the base to guide me.

More feedback:

4. When installing the hotbed, there's some confusing order of operations. At the end of step 12 (print bed assy) I had the hotbed attached to the standoffs. But in step 13, you need to access screws under the hotbed in order to finish the smooth bar alignment. I found this part of the build very fiddly, but I don't see much that could really be improved.

5. The ribbon connector nomenclature and pictures is really confusing. When I first read section 1, I thought, "ok, great, got it!". Then while building I found that each new ribbon connector reduced me to trial and error. I think a better naming scheme would be the number of mm's from center to center or the holes. Then I can just hold them up to my ruler and find out which one it is.

I'm really having fun putting it together, and looking forward to my first print.

-jeff

PS: Windex is an ammonia/vinegar based window and mirror cleaner, and it worked fine for me. See My Great Big Fat Greek Wedding for other uses...

Hello again nophead,

Just wanted to tell you that I've finished the build and calibration and gotten my first prints. Wow! So cool! And your machine is spot-on accurate, so high res. I'm really really impressed.

Here's some more feedback from calibrating and using the printer for a day or so:

1. having the .skeinforge directory on the USB key start with "dot" makes it hard to find on Unix. I understand why it's named that, but it still caused me problems trying to find it (I'm an experienced unix user and this still caught me up). It would be easier if the instructions said, "If you are using Skeinforge on Linux, install it using the instructions from the skeinforge website, then copy the folder Skienforge-Defaults-For-Mendel90 to $HOME/.skeinforge."

2. skeinforge is an ugly hairy beast. It would be really nice to have a little chapter with a walkthrough going from a STL file to a .G file copied onto the sd card and then printed. I struggled with this, and searched for skeinforge tutorials and found lots of tutorials for advanced features, but nothing saying, "just do it like this to get started".

Slic3r is much easier to understand at first glance, but I understand you don't recommend it ("doesn't maintain geomettry"). Can you explain more what the problem is?

3. The instructions for how to calibrate the printer are excellent from a mechanical point of view and terrible from a software point of view. For someone who's never touched an Arduino in my life, I was totally at a loss. I had to overcome many little problems and only managed to compile and flash the new firmware by chance in the end. Some problems I had:
a. wasn't sure where to find the hardware/Melzi dir and then where to put it in Arduino for Linux
b. didn't know what to choose from the Tools->Hardware menu. (Which of the two melzis?)
c. had no serial port menu (greyed out). fixed by adding my unix user to dialout group, then log in again.
d. didn't know which programmer to choose (solved by trying each in turn, and randomly pressing the reset button and the upload button until something happened).
e. didn't really have any confidence what to do with jumpers (auto reset and usb/vreg), and I still don't know if what I did was essential, pointless, or dangerous. But somehow the melzi got new firmware anyway and didn't explode, so...

I think a much better way of calibrating would be for the melzi firmware to read from a file called calib.txt on the SD card. I will look into if that's possible and post patches if I manage to get it working. Including a short length of smooth bar in the kit for calibration would be welcome as well. I'd like to follow a process more like this:

1. use pronterface to lower the extruder until it just touches the bar as you roll it under
2. type MXXX D, where D = the diameter of the bar you are rolling under the extruder. The software would do a Z-home operation, and count the number of steps needed to find z-home. It would then echo "Z-MAX measured as: Z mm", where Z = steps to home * mm/step + D. Then you'd copy and page Z-MAX into calib.txt.

Thanks for a really fun build experience, now I'm going to go play. Already printed some cookie cutters for easter, which my wife loved.

-jeff

I too have just finished my Mendel 90 kit build and must say the build manual was excellent as far as the hardware was concerned it was a lot easier than I was expecting it to be and had few problems. The only issues were caused by my own poor soldering but using a multimeter I corrected them easily. The android print supplied printed fantastically as good as any I had seen online so was very happy, installing the software was fairly straightforward on a windows 7 laptop and calibrating the E_STEPS went ok after a quick email to Nophead as I couldn't see which line to edit in config h even though it was staring me in the face .My issues now are all related to using skeinforge and pronterface as the printer keeps setting the wrong temps and trying to print outside the build area. I couldn't agree more that a guide on going from an STL file to a final printed object on the Mendel 90 would be brilliant and am supprised that there isn't already something similar online considering the popularity of these excellent kits. Altogether I couldn't praise nopheads Mendel 90 kit and build instructions high enough.

Gareth

I noticed that EEPROM_SETTINGS is disabled in nophead's Marlin source code. Is this because the Melzi doesn't support it? Or to save size in the binary?

I would like to move Z_HOME_POS and E_STEPS_PER_MM into the EEPROM.

-jeff

The EEPROM will work if you enable it. You probably need a later version of Marlin to set the home pos. I think it is broken in the version I use.

I don't enable the EEPROM by default because it gets very confusing when you download the firmware and the settings don't change.

Also, I find it easier to edit a text file with names in it than send random Gcode commands.

---

[www.hydraraptor.blogspot.com]

I enabled the EEPROM and EEPROM chit chat in the version of MARLIN which came with Nophead's kit so that I could make use of the MARLIN M206 command - specifically and only so that I could tweak the Z height using M206 Zz where the value of 'z' effectively brings the bed closer (+ve z) or further (-ve z) from the nozzle. Unfortunately, this was not as straightforward as I thought it would be - enabling the two #define lines for the EEPROM was easy but the M206 command did not work. This led to me making a few changes to the MARLIN firmware which now works as I believe it was meant to..

The reason I wanted to use M206 was so that I could swap the glass on the heated bed. I have two pieces of glass, one plain glass which I use most of the time for PLA and a second which is covered in Kapton tape for use with ABS. (I have yet to try ABS juice). One sheet is thinner than the other so when I want to use the glass prepared for ABS prints, a quick M206 Zz command and I am up and running without having to amend the firmware source code and upload it. If I intend using the "ABS glass" for a period of time, I can store the M206 value by using M500.

Alan

Edited 3 time(s). Last edit at 03/13/2013 11:10AM by Alzibiff.

jra Wrote:
> 1. having the .skeinforge directory on the USB key
> start with "dot" makes it hard to find on Unix. I
> understand why it's named that, but it still
> caused me problems trying to find it (I'm an
> experienced unix user and this still caught me
> up).

It needs to have the dot for Windows because you can't rename it to have a leading dot in Explorer. I don't want to make it easier for Linux users at the expense of making it harder for Windows users.

I will add a warning that it is hidden in Linux. As is printerface.ini in Win8 apparently.

> It would be easier if the instructions said,
> "If you are using Skeinforge on Linux, install it
> using the instructions from the skeinforge
> website, then copy the folder
> Skienforge-Defaults-For-Mendel90 to
> $HOME/.skeinforge."

You don't need to install Skeinforge, copying it is sufficient.

It says to copy the config to ~/.skeinforge, which I believe is the same as $HOME?

>
> 2. skeinforge is an ugly hairy beast. It would be
> really nice to have a little chapter with a
> walkthrough going from a STL file to a .G file
> copied onto the sd card and then printed. I
> struggled with this, and searched for skeinforge
> tutorials and found lots of tutorials for advanced
> features, but nothing saying, "just do it like
> this to get started".

I think all that is missing is "Press the Skeinforge button and select the STL file you want to slice." I will add that.

>
> Slic3r is much easier to understand at first
> glance, but I understand you don't recommend it
> ("doesn't maintain geomettry"). Can you explain
> more what the problem is?

With skeinforge the outline of the object is the same as the outline of a slice through the STL (inset by the half the filament width). This isn't the case for slic3r. It creates gaps where there are no gaps. For example with the Mendel90 carriage it doesn't connect the third bearing mount to the rest. It makes a carriage with a gap in it, an isolated bearing mount and two separate single filament walls where the connection should be. With skeinforge you get one outline that goes all the way round so the third bearing mount is connected. In fact the object is a big D shape with some slots in it which leave the bearing attached, not two separate objects with some bridging pieces.

Other things it does is add random spiky bits in corners, etc. See [github.com] and [github.com] for typical issues.

I tried it a long time ago but holes were changing in size from one version to another and simple things like a sloping roof had gaps in it. I started following the issues on github, waiting for it to become stable, but it never seems to get any better.

I have tried kisslicer, which is very easy to use, but holes come out too small. I have yet to try Cura. That is also easy but I don't know if it is accurate.

If you want to print Yodas and Owls it doesn't matter what size they are but I need to print things that are geometrical.

>
> 3. The instructions for how to calibrate the
> printer are excellent from a mechanical point of
> view and terrible from a software point of view.
> For someone who's never touched an Arduino in my
> life, I was totally at a loss. I had to overcome
> many little problems and only managed to compile
> and flash the new firmware by chance in the end.
> Some problems I had:
> a. wasn't sure where to find the hardware/Melzi
> dir and then where to put it in Arduino for Linux

The manual states "After installing Arduino copy the Melzi folder from the SD card to /usr/share/arduino/hardware using sudo cp -r." That works for Ubuntu.

> b. didn't know what to choose from the
> Tools->Hardware menu. (Which of the two melzis?)

The manual states "Use the Tools / Board menu option to set the board to “Melzi W/ ATmega1284p 16mHz”."

> c. had no serial port menu (greyed out). fixed
> by adding my unix user to dialout group, then log
> in again.

I didn't need to do that on Ubuntu. Which flavour of Unix are you using? I can't really support different operation systems. Anybody that uses Unix should be capable of solving their own problems. If not they are using the wrong OS as it isn't user friendly.

> d. didn't know which programmer to choose
> (solved by trying each in turn, and randomly
> pressing the reset button and the upload button
> until something happened).

You don't need to select a programmer to load the firmware. It is for programming the bootloader with an ISP programmer.

You just press the download button. That's it with Melzi as it has a functional auto reset circuit.

> e. didn't really have any confidence what to do
> with jumpers (auto reset and usb/vreg), and I
> still don't know if what I did was essential,
> pointless, or dangerous. But somehow the melzi got
> new firmware anyway and didn't explode, so...

Quote

When you are directed below to update the firmware do the following: -
• Ensure the auto reset jumper at the top of the Melzi is fitted.
• Use the Tools / Board menu option to set the board to “Melzi W/ ATmega1284p 16mHz”.
• Use the Tools / Serial Port option to select the correct USB port.
• Press the play button.
• Wait for ”Done uploading” to appear.
• Save your changes with the save button.

There is also a section describing what the jumpers do:

Quote

Make sure the auto reset jumper is installed at the top of the Melzi. This is required for loading
firmware but has to be removed for stand alone operation from the SD card. The power jumper in
the middle of the board should be set to VREG to power the Melzi from the PSU. The USB position
can be useful for loading firmware before the board is installed but the 5V at the end of a USB cable
is a lot less accurate and more noisy then the regulator output.

I don't see how it could be clearer.

>
> I think a much better way of calibrating would be
> for the melzi firmware to read from a file called
> calib.txt on the SD card. I will look into if
> that's possible and post patches if I manage to
> get it working. Including a short length of smooth
> bar in the kit for calibration would be welcome as
> well. I'd like to follow a process more like
> this:
>
> 1. use pronterface to lower the extruder until
> it just touches the bar as you roll it under
> 2. type MXXX D, where D = the diameter of the
> bar you are rolling under the extruder. The
> software would do a Z-home operation, and count
> the number of steps needed to find z-home. It
> would then echo "Z-MAX measured as: Z mm", where Z
> = steps to home * mm/step + D. Then you'd copy and
> page Z-MAX into calib.txt.

All that firmware effort to avoid:

Quote

When the nozzle is just touching your feeler object, take the Z value with M114. Subtract the height
of the object from that and then subtract the result from the Z_HOME_POS in the firmware.
For example, if it thinks it is at 7 when it is actually 8 then Z_HOME_POS is too small by 1mm.

Why is it any easier to put a number on Calib.txt on the SDcard than it is to put it in configuration.h and press the download button?

There are things I can make clearer in the manual but nearly everything you mention is already in there.

Edited 1 time(s). Last edit at 03/13/2013 11:36AM by nophead.

---

[www.hydraraptor.blogspot.com]

Quote

.My issues now are all related to using skeinforge and pronterface as the printer keeps setting the wrong temps and trying to print outside the build area. I couldn't agree more that a guide on going from an STL file to a final printed object on the Mendel 90 would be brilliant and am supprised that there isn't already something similar online considering the popularity of these excellent kits.

The temperatures are set in start.gcode because I use the unattended start procedure so can't use the temperature module in Skeinforge.

The configuration I supply will print in the centre. Mendel90 has the origin in the centre of the bed. The pronterface.ini I supply make pronterface display the origin in the centre. The skeinforge config has Multiply ticked with one row and one column and Center X and Center Y set to 0.0. That makes it slice anything centred at the origin, which is in the middle of the bed.

Other hosts and slicers can be set up the same way. Certainly slic3r, Kissslicer and Repetier host all work with the origin in the middle. For some reason Reprap convention is to have the origin in the corner, but then you have to slice with a bed size dependent offset and you cant actually go to the corners on a lot of machines because of the clips holding the glass.

As mentioned above I think the manual has going from STL to printed object in it. It is only missing press the skeinforge button, which as there are so many buttons, it really does need to be stated.

Edited 1 time(s). Last edit at 03/13/2013 03:46PM by nophead.

---

[www.hydraraptor.blogspot.com]

I agree that all the bits and pieces are there in the manual, and the fact of the matter is that I did get it working. But coming off the very clear build and testing steps, I felt like the calibration and firmware upload steps were really confusing.

I really didn't feel comfortable, on top of all the other things I was trying to learn and understand, changing constants in a .h file and compiling and sending a new firmware to the Melzi. Perhaps the problem is that I do embedded linux development for my day job and I know how easy it is to brick something.

Anyway, thanks for listening and responding. I understand when you say "I don't see how to make it clearer". I've said that before when people get confused with my documentation.

-jeff