RepRap: Blog

Chris Olah and I have been developing a fork of the Prusa Mendel at Hacklab.to over the past couple months with 3 major goals:

Labels: Mendel

I've been having fun watching the cambrian explosion of variations on Mendel.

This one is particularly significant.



LaserCut_Mendel by Kimberly and Lambert Andrus of TechZoneCommunications.com llc.

LaserCut Mendel has the same metal hardware and other paraphernalia as vanilla Mendel. But now you can make one in the privacy of your own home using your laser cutter!!! (If you don't happen to have a laser cutter, you can buy a LaserCut Mendel from Kimberly in the RepRap For Sale Forum, or come out this weekend to the Bay Area Maker Faire, and buy one from her in person).

Kimberly Anders also sells a version of the RepRap Gen 3 Electronics, the Generation_3_Electronics - Tech_Zone_Remix. These boards have generated quite a bit of interest in the community and you can buy them in this thread in The RepRap For Sale Forum. In the best RepRap traditions, our fellow users are helping document them here: Generation_3_Electronics/Tech_Zone_Remix/How_to. (Discussion/Support)

Labels: generation 3 electronics, lasercutter, Mendel, techzone

I received some aluminum plates (one for my rapman and other for my mendel) and some 50W resistors (still waiting for the 100W ones) tied it all together and it worked just like Chris shown on his super cool blog :), anyhow using the max power (~12A ~500W) I can push to the bed it can overshoot the max temperature those resistors are specified to handle so I needed some way to control it. Initially I used the simple thermal switch (like you can see in TA oven) but as I want to be able to change the temperature from the motherboard I decided to replace it with small electronics.
I do not have a SSR lying around but I have bunch of "no name / no marking" 25A 600V triac's that are perfect for the job, I used the MOC3043 (optocoupler with triak output and zero crossing detection) to separate the electronics from the AC going in to heat bed (I'm using 41VAC), the microchip PIC16F819 is used to drive the LCD, read NTC and work as I2C slave. NTC is connected with 10K resistor to maximize reading range between 20 and 180C. NTC used is same one from BFB hot end's (GT-204).

Labels: controller, heated bed, I2C, Mendel, PIC, RapMan

A little while ago, Nophead made a Dibond heated bed. Dibond is a sandwich of LDPE between two aluminium sheets, and is very flat.

I thought I'd have a go too. Instead of the TO220 resistors he used, I used nichrome wire taped down with Kapton on the back.



Then I insulated that with crack-filler foam, cut down to about 10mm thick with a bread knife.




It seems to work well. Here are the larger Mendel parts printed on it in PLA with it set to 50^oC. Their bottoms are f. as pancakes.



I used a piece of ordinary aluminium sheet clipped on the top with bulldog clips to give me a removable tray with good thermal conductivity. That is what has the blue tape on here. The tray is flexible, but the Dibond holds it flat.

I thought it'd be clever to use the 5v supply out of my PC PSU, as that's not being used for anything else. But the current (16A) is a bit silly - connectors and so on get warmish. For the next one I'll run it at 12v and about 7A.

Dibond is rated up to 80^oC, which means it's fine for PLA, but might not get hot enough for ABS.

I've integrated it into the host software, the G Codes, and the firmware and updated the copies in the repository. I have to go to Cardiff tomorrow to give a RepRap talk to the British Computer Society, but I hope to put all the details on the wiki over the weekend.

Labels: dibond, heated bed, Mendel

I got irritated by the thought of a large chunk of wood that could be replaced by printing. So I'm tinkering with these modular panels about 100mm a side, scaled to fit the frame 3 abreast using the same fittings as their wooden counterpart.

Here you can see an Arduino with a simple prototype board on it. This holds 2 of TIP122 drivers for the heater and an experimental DC extruder. Alternatively it can hold 2 EasyDriver stepper controllers. Basic, but hopefully reliable.


They interconnect with 2xM4 16mm screws and trapped nuts. Objects can be anchored on pillars with 40mm M3 or M4 screws, or against the module's surface with 30mm ones. Here are some preliminary prototypes. I've attached them to the current "Simplified Kiwi Mendel" known more affectionately as Lemon Slice.

Speaking of slices, I caught my hand on a falling mandoline and have to rest my left hand. fortunately thumb works so I can sneakily enter blogs from bed on my phone...

Vik :v)

Labels: kiwi, mandoline, Mendel, module, panel, printed, reprap

I'm using RapMan 3.0 for a while now and it is a great machine. I did tweak it a lot, but however I tweak it, and however I change more or less important parts or the way some things work on this Darwin based kit the few things I cannot change and they bug me big time. Major issue I have with it is size. And as I'm printing a lot of PP/HDPE there is no way I can use that big table (PP warps 10 times worse then ABS)...

From RapMan

now it would be cool if I only used this desk for printing but that's not the case. I found solution in Mendel, the second generation reprap machine. Mendel is super cool reprap, it has small footprint, it requires less power... but all the extruder designs I seen for Mendel are not comparable to those of rapman's designed and sold by BFB. Getting your nozzle to 260C and extruding at high pressure will render PFTE tube based extruder tips unusable, and standard extruder drive has no way of applying enough pressure. dr. Bowyer recently published YAGE that IMHO had potential to push with enough strenght (I did modify the drive a bit and added some very sharp 2mm pitch pulley as drive instead of that M4 inset) but still, the PFTE tube design and 260C don't really go together, not to mention glued PFTE and high pressure so I looked at the hot end piece of rapman's extruder. BFB sells hot end kit for some reasonable price, and as I have few spare and I know this hot end can take a beating (the first one extruded kilograms of different filaments and is still in prime condition) I decided to give it a go.

From MENDEL

Labels: extruder, Mendel, RapMan

I have finally managed to print tiny little ball-chain gears that work with 3.3mm and 3.5mm diameter ball-chain and still fit on the NEMA17's 5mm output shaft. The trick is to print the gears in two pieces.

As you can see in the photograph, I print two 4mm thick sections of gear and put them on the shaft with an M5 washer sandwiched in between. Make sure the teeth are lined up on both gears. This gives a channel to guide the ball-chain down the centre of the gear, and grips the sides of the balls adequately. As the gears age, it will also stop the balls grinding their way too far through the PLA, though I must admit that my experience with ball-chain Z axis gears suggests this will not be a major issue.


The gears themselves are designed to be "spiky", but the vaguearities of the printing and rendering process lop the spikes off and leave enough by accident to produce a functioning gear! SCAD and STL files are here: http://www.thingiverse.com/thing:2009

As the gears are stuck on the shaft of the NEMA17, and the original Mendel Y Motor Bracket is thicker than the length of the NEMA17 drive shaft, a new motor bracket is needed. Also, the ball chain cannot take tight corners well, so a 608 bearing is mounted on the frame cross-member to act as a pulley. The same arrangement is fitted to the other side of the frame making the Y Idler Bracket obsolete. There are knock-on effects on the bed design and X Carriage which I'm still working around.

Vik :v)

Labels: ball chain, gear, Mendel, reprap, small

Well, guess what? Supplies of suitable belt in New Zealand have completely failed to materialise. This comes as no great surprise to me, but must typify the situation in developing parts of the world. So, I am going to do a beaded belt (aka ball-chain or bath-chain) version of Mendel, using no fancy belts at all. The only problem is that the X & Y gears need to be very small to have enough resolution.

Well, the improved stepper drivers we're using have a slightly higher resolution than the old ones - not as high as the 16 micro-steps or more claimed by the chip specs, but useful. Also my printing skills have improved. This means that it may be possible to match a slightly larger chain with a slightly larger printed drive gear and better 0.1mm accuracy.

If that fails, it's time to break out the frickin' big "laser" ...

To the workshop!

Vik :v)

Labels: beaded belt, frickin' big laser, Mendel, reprap

Mendel comes in at £395 for the lot (in the UK).

Assumptions are:

- Using RS (UK) for bought in standard stock (fasteners, bearings, bars, studding and belts)
- 4 off NEMA 17s, @ £20 each
- RP parts @ material cost of £20 (assumes no commission)
- Thick sheet cost (inc cutting cost) @ £20
- Electronics are Zach’s Gen3 @ £110

Raw data can be found in the Mendel assembly data sheet on the wiki. A tabular summary is also available on the wiki.

Here’s a graphical breakdown:

Labels: costing, Costs, Mendel

I have done some file reorganizations in the RepRap subversion repository to tidy things for the Mendel release.

Under the location https://reprap.svn.sourceforge.net/svnroot/reprap the (partial) directory structure of the trunk is now (the links are for browsing):

trunk/mendel/electronics - the designs for Mendel
trunk/mendel/firmware
trunk/mendel/mechanics

trunk/darwin/electronics - the designs for Darwin
trunk/darwin/firmware
trunk/darwin/mechanics

trunk/reprap/host - the Java host software (common to both)

The Darwin designs used to be under trunk/reprap, and they still are for the time being to avoid breaking links. But eventually I will remove those.

Some of the Mendel data is already there, and we'll be adding the rest shortly.

Labels: Mendel, release, version 2

Here's my toothed belt design for Mendelssohn using the Darwin-style moulded gear. I've turned Mendelssohn on its side, so you can see the Y axis workings clearly. I'm just using the NEMA17 to drive this at 250mA, so it won't get hot enough to melt the CAPA. The belt is the E-SKY model helicopter belt, and there is just the right length once you've cut the loop.

The big silver idler is a 608 bearing and 2x ridiculous washers as per the Z idler. The less heavily loaded small belt guide is an M4 soft, black plastic spacer between two M4x11 washers (plenty of lubricant). The belt is clamped by printed pillars (shout up if anyone wants the design files BTW), each held by 2 M4x40mm screws.

The base itself is 230x230x9mm MDF, supported on 16 x M4x40mm countersunk screws clad in 20mm lengths of stiff rubber tubing. These go into 4 PLA pads, which can also be hand-formed from CAPA. These pads have a very low friction, and can be levelled individually to cope with quite gross manufacturing flaws :) The axis moves at 50mm/s very smoothly with plenty in reserve even at 250mA. Yes, bearings will move much smother and faster with less wear. But this seems to work for now so I'll move on to butchering the X axis.

Vik :v)

Labels: helicopter belt, Mendel, Mendelssohn, y-axis

I am particularly fond of Vik's bath-plug drive, but it's too lumpy for the X and Y axes. His model-helicopter belts will work well for these, but we can't print the drive gears for them, as the teeth are a bit too fine.

Toothed timing belts are a perennial problem for RepRap: we can reprap drive gears for the 5-mm (0.2") pitch ones with no difficulty, but not for a finer pitch. But the 5-mm pitch belts generally have a minimum width of 10mm, whereas we really need 5mm-wide belts. Otherwise the whole thing gets too chunky.

I think I finally have a solution: split a 10mm-wide belt to give two 5mm-wide ones. I made a jig...

...that you can put the belt in, stab it with a box-cutter/Stanley blade, then pull the belt through. The AoI file for this is in the Mendel section of the subversion repository here. Because timing belts have cords (usually kevlar or steel) running along them lengthwise to stop them stretching, the split works particularly cleanly as it cuts between two cords. And 10mm-wide belts with 5mm (0.2") pitch seem to be universally available and cheap. You can see the result at the top of this post.

I've designed a belt drive gear (here) that you can also see on the NEMA 14 motor at the top. This should work for all three axes of Mendel.

Incidentally, when reprapping something like this with a low cross-sectional area there can be problems with each layer of the part getting too much material deposited on it if you use the same reprapping settings as for larger parts. To avoid this:

1. Print the part along with other things of the same height or higher (this effectively increases the cross-sectional area), and
2. Increase the layer thickness slightly. I use 0.3mm for normal builds, and 0.5mm for tall thin things.
   

The drive gear does need the motor shaft to be filed to have two flats:


The way to do this is to:

1. Put Blu-tack in the bearing hole to prevent iron filings getting in the bearing,
2. Clamp the shaft, not the motor; carefully get it parallel to the vice by eye,
3. When filing, check frequently: if the flat shape is a rectangle, it's parallel to the axis; if it's a trapezium, it's not.

With a little care you can get the size right within 0.1mm.

Labels: cogs, gears, Mendel, timing belt

I've got the Z axis going on the Mendelssohn design. In the picture here (RepRap is tilted so the bottom faces the camera) you can see the ball-chain drive connected up to the two Z threaded drive bars. The shiny pulley is a 608 bearing with some great big washers either side of it. Side note: If you put M4 nuts on an M4 bolt, the exterior diameter of the nut is 8mm and it fits perfectly as the axle for a 608 bearing.

The 4.5mm beaded chain gears are the same as used on the RepRap Child/Phoenix machine that built Mendelssohn. Printing a gear with a grub screw and captive nut was too hard, so I devised a clamp-on gear that fastens onto the NEMA17 shaft using an ordinary hose clamp. The tensioning pulley makes installation of the Z drive chain much easier on this prototype than the Darwin.


If you want to see it in operation, here's a video of it. I've stuck a mole grip on the X axis to add a bit of weight, and in case you can't see the ammeter it's running at 250mA. It ran like this for 20 mins, cycling up and down without overheating the motor or the EasyDriveV3 stepper driver. The OLPC is just powering the Arduino board - though some magic is needed to stop the OLPC powering down its USB ports at inconvenient moments. Once that's sorted I'll be able to power a RepRap using Ralith's minimalist RepRap command line utilities.

Vik :v)

PS I've since made the Y axis move back & forth using helicopter belt but need to print new bits to do it properly. People are asleep now, so I'll leave that for tomorrow.

Labels: Mendel, Mendelssohn, reprap, z axis

Well, you've seen the nice diagram from Ed so here's what the framework looks like when you put it all together. One change becomes immediately apparent - there is no longer room for a cat to hide inside or under it. Things seem to basically fit, though it might be an idea to wrap tape or plastic shim around the Y bars to ensure they're gripped firmly.

There is a little flexion at the top of the assembly. It is possible that with the right speed of head movement at the right height we could get some vibrations building up, however I have not fitted the board holder or the Z axis guides, which may well add more rigidity.

The extruder holder is a simplistic design, basically because there are not enough 625 bearings in my part of NZ to build a Mendel (Gary just cleaned out the local supplier!). I found PLA on silver steel slides just beautifully, so I'll be using that approach for the prototyping. For the vertical guides I'll probably just push some CAPA in for the moment.

So, now comes the tricky bits - saddles, gears, pulleys etc. This prototype will probably experiment with 3 types of belt drive - 4.5mm ball-chain as successfully used on Darwins for the Z, E-Sky EK1-0503 model helicopter belt (sadly not quite long enough for the proposed X axis) on the Y, and if I can print gears for the 3.5mm ball-chain that'll go on the X axis. If not I'll install the standard Darwin X belt.

Vik :v)

Labels: Mendel, Mendelssohn, reprap

A structure that could be the beginning of a Mendel has been seen lurking in the workshop. It must be stressed here that Mendel is not yet even in alpha status, so changes in design are almost guaranteed. But thanks to the miracle of Open Source, those who enjoy building stuff that hasn't been designed yet can join in the fun.

Because this one is being printed by Phoenix, the original Child RepRap, I've decided to call it "Mendelssohn." This will make Mendelssohn a 3rd generation RepRap. Although it looks - and is - a lot smaller than the Darwin design, Ed's cunning artifices have made a machine with a greater printable area than the original Darwin. It'll be a lot easier and cheaper to put together too - after we've fiddled with it for a bit.

Now to convert the rest of the pieces into gcode...

Vik :v)

Labels: Mendel, Mendelssohn, reprap