Extruder design problems

Been a while since my past blog.  Its been quite a mission to get a working extruder built, but i'm nearly there.  The last stepper motor just did not have enough grunt and the MakerBot stepper I have  is reserved for Mendel, so i've bought one from MindKits, 2.3Nm or there abouts vs MakerBot rated 2.6Nm, good enough I reckon as RepRap suggests a minimum of 1.4Nm.

With the new motor attached I paced the bits I planned on using against the plate and motor.

This caused a few problems and I also realized that I also need a way of forcing the filament onto the pinch wheel, this resulted in.

 

Unfortunately I forgot to take with me to work the copper bracket which will hold the extruder, as such I forgot and drilled a hole where the bottom M8 bolt is.  How idiotic of me.  

The next day took the bracket with me and moved the parts around, and found a few parts lying around that were scrapped, plus a washer, so with more drilling I came up with.

 

This design enabled me to put in a spring loaded filament pusher, for lack of a better term, and I have polymorph around, and so moulded  some into a bracket for an M4 screw, washer and spring.

The location of the bracket was important as I want it far enough away from the copper bracket which serves a double purpose as a heatsink, and polymorph is malleable at 60 degrees.

The design also allows the filament to be guided into the pinch wheel for when it moves laterally on the wheel.

Though this is by no means a stretch from being completed, for one the bolt for the M4 filament pusher is too long, the spring too long, the M8 remaining bolt is too long, holes need to be drilled underneath and a bracket to lift up the copper bracket, how ironic, a bracket for the bracket.  the pinch wheel assembly also needs to glued together and onto the shaft, the plan for doing the pinch wheel to the shaft of the motor is to use 680 LocTite retaining compound, which from experience from work is extremely difficult to break, often requiring a butane torch directly at the joint for over a minute to be enough for over torque to break.

I am still trying to build this thing to the best of my ability, with hurdles around every corner, very limited funds to work on it, IE none since mid November 2009, and i've still got to get around to migrating the wine I have into a clean demijohn for its 6 month brew cycle.

Ah well, its all fun!

Hopefully, by March it will be up and ready for calibration, Z axis is waiting on the extruder to be mounted so that can be finished.

Two steps forward, one step back

Thanks to suggestions from Wade and NopHead on the RepRap forum, I found the problem. I feel like an idiot, especially when I am an electronic technician at work and doing pretty well, but completely missed the obvious on my own project.

It was suggested that I didn't wire up two extra lines from the I2C bus to the extruder board, which I did, and it still didn't work. So there I am staring at the firmware, and looking at the picture of the Mendel on the RepRap Wiki, when something clicked, verified something that NopHead mentioned, then looked at the wiring diagram closely.

Its true, i'm an idiot. The I2C isn't implemented at ALL, and the two lines that would actually send pulses from the mother board on what would be I2C to the stepper override pins, crazy huh?
So i'm pleased for a moment that the LEDs are flashing, except one.

I hooked up my DC motor with filter and on the side that had just one LED flashing, the DC motor did nothing, on the other h-bridge on the extruder board, it worked fine.

Another thing was, just in case, I put on a new 644p on the motherboard and removed the Sanguino 1.0 that I had bodged on, and using a specific command for AVRdude, which HAS to be run in the location of the hex file (something that people keep forgetting to mention) I was able to successfully program the bootloader, as was an old old problem where programming the virgin chip directly through Arduuino IDE first would always fail. However using AVRdude to program it the first time, Arduino IDE programs the bootloader just fine thereafter, weird. Another reason I hate Arduino, but not Atmel.

AVRdude come with the latest WinAVR install

The command I used, which I also posted on the wiki is
avrdude -c usbtiny -p m644p -U flash:w:ATmegaBOOT_644P.hex

Run under the directory

C:\Program Files\RepRap\arduino-0017\hardware\bootloaders\atmega644p>

Though with AVRdude, you can only use it through the DOS command terminal

There is a program that does it through GUI, I tried it, but it crashed and has not worked since
AVRdude-GUI

Even though I can't justify the expense of buying a replacement h-bridge chip for the extruder board, as local supplier Farnell is out of stock and means that i'll have to import it from far away, means that for a ~NZ$7 chip would cost upwards of over NZ$50 in shipping alone, but I still need my RepStrap done ASAP. Fortunately I bought a V1.2 stepper driver, essentially a L27/L298 combo that everyone knows and loves.

The H-Bridge chip is removed and modded the extruder board in a way that I can use the screw terminal to attach Step and Direction to the V1.2, though I have forgotten that I will also need power for it as well, and I have no spare plugs on my power supply, i'll fine some #8 wire somewhere and bodge it.

Annoyed

I'm quite annoyed from this past weekend, I still cannot get my extruder board to recognize commands sent from the host software to turn on the extruder.
My thinking is that I may have a dud chip, but I can't totally believe that, especially when I use Zach's test code which puts ADC signals to various parts of the board, yet though, when I insert my own code, which is simple digitalWrite, the extruder h-bridges are at 1.8V.
No one i've asked either cares or knows what is going off.

It is however entirely possible that I got a dud chip, I got one for the 644p, found the tx line 100 ohm internal short, how the hell does this happen.

I was going to post some pictures, but there is virtually nothing to see yet to make it worth while, all I have is a wooden frame, with the Y axis sitting loosley on top of the wooden frame, same with the x axis, though its lower down.
Both the X and Y axes are old, mismatched A4 flatbed scanners that I scored off www.TradeMe.co.nz for $5 for both.


I did buy another demijohn last week, hoping, and failed in the weekend, to transfer my wine to a clean and sterile demijohn so that all the dead yeast at the bottom won't affect the 3rd and final fermentation cycle, which could take 6 months or more, for my plum wine. Why plum wine? Cause I have a very large plum tree that fruits an amazing amount of fruit over every Christmas week, though last year it was 3 weeks early, great for getting me started for the first time on wine making.

Hmm, my blog posts seem to be rather long.

RepStrap log start

I've started this blog, cause I feel like I need a place to jot down what i've done, and am doing.

To start off, even though my cartesian robot is about half complete, i've been working it for several months now, when I have either had enough money or time to devote to it.

I was hoping to have this thing built and working before the New Year started, so much for that plan.

So while its holiday time, and with the materials I bought at the start i've managed to get the inside frame nearly done for the z axis, just got to add a couple of bearings to it.
I'm designing the Z axis from scratch, the first attempt failed, in that it had way insufficeint constraints and mean't that it flailed everywhere when the threaded rod was turned. I am now using long drawer type slides attached to a frame, I won't know how this is gonna turn out till I get the outside frame done, that's the frame that moves up and down, the inside frame is stationry.

For the robot, the X and Y axes are made up of two old A4 flatbed scanners, bought them off TradeMe forNZ $5 for the two, what a sweet deal for two drive mechanisms, 2 opto end stops and high resolution 1.8 degree steppers.

I'm using MakerBot electronics, as there wasn't much point redesigning that which has already been done, though several people have apparently coded up their own g-code driver boards, if I get board I may try this by turning Adrian Bowyer's 5D Arduino code to true embedded C.