And for the CNC router with extruder - that would of course work well. Not sure at what speeds those things work, you want to be able to move quite fast (like 120mm/s) for printing
@Vaclav_Hula - what if the 1 cubic meter item is mostly open space? (again, please note - I’m not trying to print a cubic meter plastic cube, I just want to have the flexibility to print items that have larger extents than the common 150mm x 150mm x 150mm…
I’m thinking more about adding FDM capabilities to a CNC router table now…
Price an accuracy are the imitation. Loki at this : http://blog.ponoko.com/2012/09/25/transforming-a-shipping-container-into-a-huge-3d-printer/
It give an idea
@Vaclav_Hula As long as you stay below 3m (for 6/8mm, 6m for 12 and above) cost for rods and threaded bars are linear with their length. 3m hardened and honed rods are about 50 Euro (tax included), so not a big deal. Even CNC quality trapezoidal threads are cheap - about 10 Euro per meter (up to 1.5m). And if you need sizes above, prices do not realy climb as long as you don’t pass like 20m length.
Mainly cos it takes too long to print bigger objects!
@Hans_Franke Axis from 12mm hardened smooth rods, one meter long, with .5kg extruder in center will sag by 0,4mm. Thats whole layer height and some. And it will oscillate up-down a bit.
@steeve_becker - HOLY CR@P! That’s… crazy! and cool!
@Ben_Norris - please define “too long”. I would be fine waiting for days for a print to finish…
The real reason is market, there is a much larger market for small foot print printer then a large printer, think of it this way a 1x1x1 printer would probably be around 1.3x1.3x1.3 and print times would be huge and cost of materials are much higher.
@Eric_Cha , regarding compensating for sag and gravity with beefier rods or support: it would seem to me that a software calibration for the “unstraightness” of the rods would be a decent alternative.
@Lee_Lemay - that’s an interesting idea. I could easily see putting a micrometer on the Z-axis and running it across the work surface to get a “variance correction” map of the work area to compensate for sag (and even a slightly warped work surface)…
@Eric_Cha - yar. It seems like instrumentation of the position of the extruder could go a long way towards removing many of the calibration issues that we face. I’m envisioning an inexpensive camera tracking the extruder, making the extruder controller a closed loop process.
@Lee_Lemay You will not succesully compensate the sag, as it will be highly dynamic depending on previous carriage moves. And inexpensive camera is nowhere near the resolution needed for closed loop positioning in larger machines - if you have 1m long axis with 5 full steps per mm (repraps have this resolution), then on camera with 5000x5000 pixels of resolution, after the skipped step the carriage would be 1px off, way bellow your ability to detect it on the move.
@Vaclav_Hula , those are great points. I have not done the requirements analysis for the problem, but it stands to reason that, if it isn’t being done already, the requirements are hard to meet.
If a step is 0.2mm, we’d optimally want > 10 pixels in that range, to make a decent measurement. At 1 meter away, that requires an angular resolution of ~ 0.011 degrees/pixel. That can certainly be achieved with the right optics - but can you actually see enough of the table to make it worth while?
This next bit is pretty fast and loose, with as much resemblence to reality as it has to a cocktail napkin, but let’s imagine we have the 5000 pixel camera that’s been mentioned. So, 5000 times 0.011 degrees is ~57 degrees. That seems like it would indeed be useful.
But whether such a camera system would be inexpensive, who knows. Specc’ing out the equipment to meet those requirements would take too much time. I’m sure that, since it hasn’t been done, it probably is not inexpensive. And maybe I’ve made some basic error in my reasoning.
I know someone who started making an A-frame printer that was 900mm x 900mm base (all rods were 1m long). First issue he encountered when he put the frame together: How to move the damn thing. It doesn’t fit in a car or station wagon and due to it’s weight, moving it around lots can deform the shape unless you REALLY strengthen the corners.
As for the time it takes to print, if you were happy with extrusion sizes of 1-1.5mm, it COULD be somewhat doable, but you won’t get great resolution of course. Your extruder would need to handle fairly high speeds though (due to the high flow rate), but conversely you might not need a gearbox on it if you’re using say 3mm filament.
As for a heated bed, you’re gonna need a LOT more power to keep it hot enough. For a 1m x 1m bed, you’d need 25 standard 200mm x 200mm heated beds (using that as a rough estimate, they’re 12V @ 11A each). In parallel that’s 12V at 275A, in series that’s 275V at 12A, or in paralleled groups of 5 in series it’d be 60V at 55A. And that’s just for a heated bed alone, forgetting the rest of your electronics (which by comparison, isn’t much).
Well, @Vaclav_Hula , you missed that my argument was about the quite high prices you cited (120 cm 10 times the price of 40cm).
Beside that, your calculation is (at least to my memory for the formula used) off by a magnitude (unless I screwed it). A one meter 12 mm steel rod, with a 0.5kg weight in the middle, will only bend 0.11 mm at maximum - and considering that the extruder is usually moved over two rods plus the threaded rod will also carry some load, so maximum dip will be less than 0.04mm.
@Hans_Franke That was not calculation, I measured it. Maybe i made error elsewhere, I will check it when possible. And the price increase was due to move to supported rods.