Baby sized reprap

Hi!

Yesterday I had a crazy idea while I was biking

Most people (including me) like buy the whole thing together. Currently either the reprap darwin or the mcwire version means, that I need to collect the items separatly. Because ordering the pipes, or the steel rods over the sea its really expensive. So involving into the reprap community (and hopefully diving into the development too) means reducing the cost and time for the newcomers.

So the ideal solution would be to order everything from the same place (ex: rrrf.org), that means reducing the reprap physical size and weight. (which are the major part of the shipping cost).

And than came this crazy idea to me:
What if we could build a baby sized reprap machine, which would be able to build the real sized reprap.
So the building area would be as small as required for every brick.
It could print every piece of the big reprap, and after the user could transform this baby sized reprap to a normal reprap.
The electronics would be exactly the same, only the cartesian frame would be reduced.

And with a normal reprap the user would be able to print bigger objects.

What do you thinks?

Do it!

Yes I had a similar idea recently to basically make a self replicating boot bot to speed up the spread of RepRaps by transferring much of the replication work from the donor to the recipient.

I.e. somebody gives you a small set of RP parts to make the boot bot. You build it and then use it to make RP parts for a Darwin, or is successor. You then strip it down and reuse the motors, electronics and possibly rods and fasteners in the Darwin. You can then pass on the RP bits for the boot bot to someone else. No work at all for the donor to pass on one copy, only a small amount of work to make a new set of boot bot RP parts to pass on another copy.

For the boot bot I was thinking along the lines of the tripod parallel kinematics design that Viktor proposed here [builders.reprap.org], except made from FFF parts.

Simply three identical vertical axes, at the corner of a triangular frame made from rods like Darwin but one less corner.

Each corner (3off) would be:
   A base bracket
   A top bracket which holds the motor.
   Two vertical 8mm rods as guide rails.
   One threaded drive.
   A flexible coupler.
   A sliding carriage.
  Two parallel arms with magnets into HDPE cup bearings
A central tool holder
Six diagonal ties as we have now.
A stationary build table at the base.

Much simpler and easier to put together and more importantly: less time to replicate. It has fewer parts and also less types of parts as all three axes are the identical.

Downsides would be a smaller build volume, perhaps slower movement, more complex DDA firmware and only a single extruder but that is fine as long as it doesn't have any parts requiring support material.

Edited 1 time(s). Last edit at 08/02/2008 11:52AM by nophead.

---

[www.hydraraptor.blogspot.com]

It sounds like a cool idea, and scaling is something that will be great for reprap to be able to do (and bigger too), but in terms of weight saving, the only weight you are going to be cutting down is the steel rods, and you don't need get get them shipped from overseas, just go into any local hardware store.

Edit: written before reading nopheads post.

Also a good idea

Edited 1 time(s). Last edit at 08/02/2008 11:58AM by greenarrow.

... look here: [builders.reprap.org] in the post "Assembling the Tripod..." - here i constructed a 'babysized' parallelkineamtik bot as demonstrator for another project, but this 'toy' would fit the terms and is very easy in building and assembling - only the controlling software is a bit more tricky ...

Viktor

I'm planning a carry-on luggage scaled version of Tommelise 2.0 that will power all three axes with little Haydon 26000 linear steppers. I'm looking at a 150x150x150 print volume. It's not intended as a means of making parts for a larger machine, though I suppose that it could.

That's great. I've always liked the simple design of Viktor's machine.

Demented

From looking at the photos and briefly at the STL's the larger printed parts are:

x-carriage 46 x 72 x 92
x-motor-bracket 47 x 86 x 95
x-square-jig 23 x 54 x 105 (where is this part used?)
y-motor-bracket 49 x 58 x 58
z-motor-bracket 32 x 66 x 67

oops wasn't taking note of which axis was which

assuming their aren't any bigger parts a 120mm cube working volume should do fine
If the current version of Bits from Bytes extruder is used, the bottom mount plate/disk could be expanded to the same size as the vertical body about 50mm
so even if the 'box' size is bumped up to a 200mm cube this is tiny in comparison to Darwin or a McWire
Ignoring for the moment the steppers and the top top of the extruder poking the mini repstrap could fit in a shoe box

Hope I haven't missed something important here

I just don't get it...

why does the complete boot bot have to be small?
isn't it enough to have all the parts fit in a small box to get it shipped for cheap?
There's no reason for sending the stepper motors, the electronics or steel rods, studdings, nuts & bolts.
All those parts can be used with the darvin afterwards.

You buy the electronics, motors and metal parts.
And build your boot-bot.
Then you print darvin parts (maybe boot bot parts), disassemble the boot bot, put it back in it's box and ship it to the next.
Everything that's really needed is the extruderhead and an easy cartesian.

true, the tripod is nice, but I think you'll have to use different electronics and software to run it, don't you?
So you have to ship the electronics too. more unnecessary weight.

'sid

The electronics would be the same: three stepper controllers, the only difference would be more complex firmware. That may mean a bigger micro, but you can get more powerful micros for less money anyway. Since driving a cartesian machine is just simpler there is no need to build a second set of electronics, just swap the firmware.

---

[www.hydraraptor.blogspot.com]

Hi sid,

... same electronics, other software ...

The main advantage of a BRR (read 'babyreprap') would be lesser friction and so smaller motors, shorter rods, belts, studdings ... and such.

So it's something cheaper and more portable/shippable ...

Viktor

using smaller motors isn't cheaper

If you have to use smaller motors you will have to ship the motors around, or you'll have to keep them and the next one will have to buy additional motors, no matter how cheap they are, they are useless afterwards, because you send the BRR awy after you're done prining the darvin parts.

Okay, same electronics (except the controller chips) hmm.. not perfect, but good enough
So, only three chips to send not the complete set of electronics.. that's good.
but the motors...
StepperMotors a fragile enough to get useless because of some -what were they called- paking gorillas?
If only parts that cannot break easily are shipped the set will do it's work a long time.

I like the idea of a small pakage that travels around the globe from one reprapper to the next, having some kind of passport where everyone adds his "visa" and let people know where the bot has been before.
But to keep the pakage travelling it has to be dropsafe, stepper motors are not, most electronics are not, even rods that need to be perfectly straight are not.

That's why less is more in my opinion

'sid

> StepperMotors a fragile enough to get useless
> because of some -what were they called- paking
> gorillas?

Are stepper motors really that fragile?

How small do you want to go?

I could see something using 4 servos: 3 for positioning, 1 for filament extruder. The 3 positioning servos could be chained together robot-arm-style with the extruder mounted on the last segment. Instead of 3mm welding rod, you could use nylon fishing line. The heater would need to run hotter but it could be much smaller and lighter.

No idea how precise or repeatable something like that would be, though.

Quote

Instead of 3mm welding rod, you could use nylon fishing line

I remember looking at nylon when reading up on thermoplastics and looking at other things which might be extrudeable. Isn't its glass transition point *significantly* higher than any of the other materials people have tried?

The problem with using a baby-size printing head is that suddenly everything is going to take waaaaay longer to print out. If you have a full-sized grownup printing head, you could just stick it straight onto the full-sized reprap once you've printed out the structural parts.

Quote

The 3 positioning servos could be chained together robot-arm-style with the extruder mounted on the last segment

Robot arms are very tricky to make, no? If you want something with plenty of precision which is still easy to print and assemble, I don't think arms are going to do the job at all.

That's not to say that an armed reprap is a bad thing... I came across this [pergatory.mit.edu] (pdf) a while back when trying to learn what 'parallel kinematics' meant. It at least gives the impression that using parallel kinematics in a SCARA robot improved accuracy; in reprap's case it might mitigate the fundamental inaccuracy of the system.

Its something I intend to revisit once I have some kind of manufacturing capacity to make the arm parts, though I doubt it is really something that you'd want on a cheap'n'cheerful repstrap that is only really intended to blast out a couple of dozen parts before being cannibalised and retired.

Ru Wrote:
-------------------------------------------------------
> Instead of 3mm welding rod, you could use nylon
> fishing line
>
Have you priced that stuff on a per lb basis?
>
New Image Plastics offers .020 Micro Rod in a variety of plastics at reasonable prices, if I recall correctly.

[www.plasticweldingrod.com]

Hi Ru,

... look through the links in this post: [forums.reprap.org]

... especially the different kinematic-concepts here: [155.69.254.10]

There are many different ways of enhancing the accuracy with parallel kinematics or special setups.

My favorite is the bionic approach, where we print elastic morphing structures with embedded hydraulic structures - as discussed here: [forums.reprap.org]

Viktor

... here: - [www.maschinenbau.hs-magdeburg.de] - is a collection of different parallel-kinematics with movies showing them in action ...

Viktor

just had a flash of an idea while reading this.

along with the baby RepRap, a firmware which does *nothing* but print RepRap parts.

ie, build the BRR, thread in the filament, turn it on, get RepRap.

of course, you'll have to have enough filament for either a constant supply or be able to pause production from time to time. also, the finished parts should be removed by hand before the next one starts... a buzzer and a pause timer would help there, unless you want to watch it without sleep, like Thomas Edison, until you know it worked....

---

David

that'd be damn shitty

Why:
First thing to print should be a small testpattern, to make sure the machine itself is still aligned properly after the shipping stress.
Noone can tell how often you'll have to repeat that step.
If you cannot change the order of what file is printed when, you may run into serious issues with the filament.
You most certainly does not have a coil of filament that's large enough to print all at once without bending the filament (as it tends to turn in the extruder)
So maybe you have enough filament for ... let's say an optobracket but the next piece is the carriage, now you're nipped.
And you cannot make sure, that you will receive every printout perfectly.
So you may need to to a piece twice, how do you cope with that then?
See... You'd have to have a display at least to choose the part from a list.
The display itself may not be the problem at all, but it has to have firmware, that has to be debugged properly...
And yet another:
You send the brr around, at that very time there is a significant modification of one (or more) parts.
The person that receives the brr cannot print but old parts...

Even worse: if it can only print darvin, it'll get's useless if Mendel is coming around.

A BabyRepRap would be nice, but it should be able to print all the parts that fit into it's workspace, and shouldn't be preloaded with any darvin or whatever

'sid

Quote

Even worse: if it can only print darvin, it'll get's useless if Mendel is coming around.

Because firmware must be carved from stone and is immutable!

The flaw here is that building a firmware which contains STL data (or even magically preprocessed data, similar to what goes into Forrest's EEPROM buffer) is going to be BIG. Waaaay bigger than will ever fit on a little uC.

There is also the hassle of rebuilding it when you wish to move to the latest version of a part, and indeed the very need to rebuild it at all which you probably won't want to have to do if it can be avoided

It's a silly idea because it doesn't make the builder's life (or developers' lives) any easier

I don't get this. If you graft a little SD card reader onto your baby reprap machine you could hold the complete parts descriptions of dozens of generations of reprap machines. A 1 GByte SD card costs about $10 and they get lots bigger than that.

The firmware can be exactly the same as goes on a big reprap. Indeed, when you've printed out the parts for the big one you ought to be able to just move the control boards and extruder over to the big one.

Ru Wrote:
-------------------------------------------------------
>
> Because firmware must be carved from stone and is
> immutable!
>

I guess the idea was, to make it as simple as possible for the "receiver" of the BRR to print a machine.
that'll include not to do anything but to feed the beast and to hit the start button.
It'll exclude to edit firmware

I guess for the "standard" user it's easier to just select some random stl and hit okay, than to reburn firmware to the machine (and even if he will, noone knows what version of the reprap the next user intends to print... how often can you flash the chip without the need of a replacement?)

True Forrest, very true.
And I do think that's what we all want.
Just feed some files via SD Card (or such) and hit a button, no need for a PC or something.
But that'll be a very long way I guess.
And that's where the need of a display is lurking again

So for now, I think it's just a small cartesian (or parallel kinematik) System with a printhead (maybe with or without electronics)
just big enough to print all necessary files, and small enough to fit in a shoebox...

Any suggestions on the machanics?
(something cheap maybe, because it may break during it's travel around the globe )

'sid

sid Wrote:
-------------------------------------------------------
>
> True Forrest, very true.
> And I do think that's what we all want.
> Just feed some files via SD Card (or such) and hit
> a button, no need for a PC or something.
> But that'll be a very long way I guess.
> And that's where the need of a display is lurking
> again
>

Shouldn't be. A Pic 24F would do the job pretty easily and handle the display and keyboard as well, I expect. The only hassle and it's a hassle, not a major obstacle, is that I'd have to buy a development board to design the electronics rather than cobbling together a Euroboard stripboard prototype.
>
> So for now, I think it's just a small cartesian
> (or parallel kinematik) System with a printhead
> (maybe with or without electronics)
> just big enough to print all necessary files, and
> small enough to fit in a shoebox...
>
> Any suggestions on the machanics?
> (something cheap maybe, because it may break
> during it's travel around the globe )
>
Tommelise 2.0 scaled down to a 150 x 150 x 75 mm print volume would fit inside a decent sized briefcase without a lot of trouble. As I have it now with a 300 x 300 x 150 print volume it fits inside a airlines-acceptable suitcase.

Parts for Tommelise cost about half of what Darwin's do.

Quote

I don't get this. If you graft a little SD card reader onto your baby reprap machine you could hold the complete parts descriptions of dozens of generations of reprap machines. A 1 GByte SD card costs about $10 and they get lots bigger than that.

Very true. But not what the original poster said.

Quote

A Pic 24F would do the job pretty easily and handle the display and keyboard

Now that would be interesting. I vaguely recall someone saying that the firmware to access conveniently formatted (ie FAT16) flash memory was quite large and inconvenient?

Ru Wrote:
-------------------------------------------------------
>
> Now that would be interesting. I vaguely recall
> someone saying that the firmware to access
> conveniently formatted (ie FAT16) flash memory was
> quite large and inconvenient?
>
That was Nop, and he's right. He said, iirc, about 900 lines of C code. He's talking about starting from scratch with a bare bones compiler, I expect.

With the Mikroelektronika Basic compiler that I use, however, all that stuff is reduced to a set of library routines that I call to do I/O and read and write files from my Pic.

Yes I did say about 1000 lines of C to handle reading, writing, directory lookup with long filenames, FAT16 and FAT32. I don't really count that as large and inconvient though. Certainly not something I would pay money for.

---

[www.hydraraptor.blogspot.com]

Is it AVR-DOS you're talking about?
then the library has 4000 lines of code..
[members.aon.at]

'sid

[edit]
1 file handle requires 1200 Bytes SRAM, plus stack, so a chip with 4k or more is recommended.

Edited 1 time(s). Last edit at 08/26/2008 08:15AM by sid.

No I am looking at some code I wrote at work.

---

[www.hydraraptor.blogspot.com]

sid Wrote:
-------------------------------------------------------
> that'd be damn shitty
>
> Why:
> First thing to print should be a small
> testpattern, to make sure the machine itself is
> still aligned properly after the shipping stress.
> Noone can tell how often you'll have to repeat
> that step.
> If you cannot change the order of what file is
> printed when, you may run into serious issues with
> the filament.
> You most certainly does not have a coil of
> filament that's large enough to print all at once
> without bending the filament (as it tends to turn
> in the extruder)
> So maybe you have enough filament for ... let's
> say an optobracket but the next piece is the
> carriage, now you're nipped.
> And you cannot make sure, that you will receive
> every printout perfectly.
> So you may need to to a piece twice, how do you
> cope with that then?
> See... You'd have to have a display at least to
> choose the part from a list.
> The display itself may not be the problem at all,
> but it has to have firmware, that has to be
> debugged properly...
> And yet another:
> You send the brr around, at that very time there
> is a significant modification of one (or more)
> parts.
> The person that receives the brr cannot print but
> old parts...
>
> Even worse: if it can only print darvin, it'll
> get's useless if Mendel is coming around.
>
> A BabyRepRap would be nice, but it should be able
> to print all the parts that fit into it's
> workspace, and shouldn't be preloaded with any
> darvin or whatever
>
> 'sid

Good points! In that case, I'll update my idea with a little forethought, which i did not do, at all, in my previous post.

As for the filament twisting out, I've seen solutions for that on the main site and other places. It is a challenge, but not an impossibility.

Rather than a firmware; a program on a USB disk that loads sequential designs into the machine and pauses after each part completes. Attach that USB disk to a special interface on it's own PCB to choose the parts to be made. This could have a small LCD counter to show part numbers and a few buttons: CONTINUE; PAUSE; CANCEL; END; PREVIOUS; NEXT. This connects to the BRR's USB interface.

The idea here is, if your machine is in *great* working order, you don't have to give up too much personal time to make a series of parts. I don't know how long it takes to clear off a part, load another design, and set the machine printing again, but I'm sure that pulling the part off, and pushing CONTINUE is fast(er) and simple(r).

Pros:
1. Before running this program the BRR could be attached via USB to a PC for testing/calibration.
2. Individual parts could be selected (so long as the user knows the part numbers; a list of which could be printed on conventional paper).
3. The sequence can be interrupted at any time and the machine attached to a PC again for testing/calibration.
4. This program could be used for other construction projects involving a large number of parts. (ie, attach the USB disk to a PC and upload numbered designs such as 0000.stl~9999.stl to a certain folder such as /designs/)
4.1 This could also be used for the real RepRap and I think someone else was working on such a thing...
5. Use your PC for other tasks or leave it off to save power costs.

Cons-
1. Belling the cat. This requires a new PCB design and software for that. It also has to output information to the BRR just like a PC would so they can use the same USB port on the BRR/RR.
2. Parts. Perhaps the LDC panel could be ripped from a cheap calculator? I'm still a beginner in electronics... I can't imagine what one would need to make this work.

Also, as someone pointed out--firmware is firm, not hard.

---

David