[Arm-netbook] screwed up the Riki200 plotter design

[Benson Mitchell]

On Tue, Jul 25, 2017 at 2:28 PM, Luke Kenneth Casson Leighton <lkcl at lkcl.net
> wrote:

> > But even if I'm wrong about the rigidity... why bother?
>
>  pulley system - doubling.
>
Okay, but you aren't getting that with EtchXY (see below). And you can add
it to my proposal just as easy.

>
>  mass is also equal for x and y.
>
Which is also true for my proposal.

>  corexy has the weight of the x-gantry.  moving in Y has more inertia that
> X.
>
Yeah, I'm not suggesting CoreXY -- I do get why that's quite unsuitable for
what you're doing.

> The thing that made
> > CoreXY special is the combination of non-moving motors with a simple
> (thus
> > cheap and lightweight) gantry. Once you've committed to the more complex
> > (thus expensive/heavy) dual-gantry setup, as seen in both your Riki200
> > design and Etch-XY, I don't see any benefit to be had from long timing
> > belts wrapping around a half-dozen pulleys; there's a much simpler way to
> > drive each axis independently with non-moving motors.
>
>  not "and guarantee rigidity and add a pulley doubling system and also
> guarantee equal mass distribution" as well.
>
The mass distribution seems fine; motors, shafts, and pulleys are all
non-moving. The moving parts are the gantries and extruder platform, all
just the same as you have them. Likewise rigidity seems pretty solid, with
one belt per block.

As for pulley doubling, it's simple to add. I just didn't go into details
because I'd assumed you were willing to give it up, since you're talking
about using EtchXY which lacks it.

Instead of anchoring the ends of the loop directly to the moving blocks,
just put pulleys there, and bring the end back to the fixed chassis to
anchor it. Going back to your original diagram for the Riki200, it's just
like the bottom 20% of that diagram, but the belt wraps 180 instead of 90
degrees around those chassis-mounted idlers. (And of course, it's rotated
90 degrees out of the page.)

If that was unclear, say so -- I'll come up with a sketch.

> For the X-axis, you put two shafts parallel to the Y-axis, at the left and
> > right sides. They each have two timing belt pulleys (at the top/bottom
> > ends), supporting one loop of timing belt to drive each green block. One
> > shaft is coupled to the motor, the other is an idler.
>
>  i know the sort of thing: i've seen it in use: it's used in the
> ultimaker-2 and also in an open design pick-and-place amachine.
>
Ah, good.

> the amount of force on the belt is considerable.  with the EtchXY design
> the force on the belt is halved due to the pulley system.
>
But the force is only halved for the Y-axis in EtchXY. Look closely -- the
X-axis is anchored directly to the green blocks, so the accelerating force
is just the sum of red and blue belt tension. (The distribution of force
between the red/blue belts (50/50 at center, progressively worse towards
limits of travel) does help vs some other designs, but that benefit applies
to the parallel-shafts system, too.) And given you seem to be building a
square printer, you should be accelerating pretty much the same amount of
mass around in X and Y -- so you have to design (choose belts and/or limit
acceleration) based on the axis without mechanical advantage.

(It would be different if your build volume is way out of square, such that
the yellow-block gantry masses twice the green-block gantry -- then you
have mechanical advantage right where you need it, so the same acceleration
in X or in Y give similar cable tension -- but AFAIK you're not doing that
sort of build.)

 by the time it's all assembled, the 2 belts, 4 pulleys, 4 shafts,
> then 4 sets of rails/rods, it really does add up very quick in terms
> of ccomplexity.  then you have to CAD design it, make sure that
> everything fits, that's a month's work right there...
>

Yeah, but you've got all the rails and rods, exactly the same -- it's
really just 4 short belts vs. 2 long belts, 4 long shafts vs. many short
shafts, and especially the pulleys. I think as a result of whatever
misunderstanding has you thinking I'm adding extra rails, you're also
overestimating the design work involved to redo it -- it should only be a
little more radical than redesigning for EtchXY.

> Etch-XY has 8 short shafts (idlers on the fixed chassis
>
>  M5 16mm bolts, 2 M5x18mm washers.  $0.02 each.  625 bearings.  saves a
> lot.
>

Wait, 6 of those 8 are tooth-side-in -- and you're still wrapping them
around bearings instead of timing belt pulleys? I wouldn't have thought you
could get away with that! But if so, that does make a bunch of the
savings/simplification I thought I was getting illusory.


> > -- not counting the motors or idlers on the moving gantries), all
> > with parallel axes (Z-axis), while the simple solution has 4 long shafts
> in
> > pairs (X-axis and Y-axis)
>
>  plus 4 pulleys

Actually 8, really (2 per shaft, 2 shafts per axis, 2 axes), whereas I
thought you had 10 pulleys all told (two on the yellow blocks, 6 fixed, and
2 on the motors), based on where the belt wraps tooth-side in, whereas
really you have... just the two on the motors?

Although, now that I think... hey, if you can wrap timing belts around
bearings, so can I!

Keep the drive side (per axis) one long shaft + 2 pulleys on the drive
side, but for the idler side use M5 screws/625 bearings just like you're
talking about -- so it is just 4 pulleys after all. (Oh, and a couple
flexible couplings, or yet more timing belt pulleys, to couple the motors
to the drive shafts.)


> and 4 sets of rods/rails
>
Again, you've already got all the rods and rails -- that stuff would really
be _exactly_ like you have it. You're "only" redesigning the 4 green/yellow
blocks to put the idler axes horizontal rather than vertical (pretty
simple, I think), and completely redoing the motor/pulley mount brackets at
the 4 corners of the frame (not really simple)  -- point is, all the linear
rails and rods stay in exactly the same places they are.

I don't think the shafts should be a big deal (Sure, not as cheap as M5
screws, but still can be pretty cheap... nothing fancy, just cold-drawn
rods), but I see where the pulleys add up. The one thing that might make it
worthwhile is that it dodges the skew problem completely -- each block is
completely controlled by one belt.

Perhaps the best answer is to modify EtchXY to give you pulley reduction on
both axes? That change is pretty straightforward. But I don't see what to
do about the skew problem...

so adding pulleys to reduce the force on a standard "cheap" 6mm GT2
> timing belt, that's important, because now you can try going twice as
> fast but still use... cheap 6mm GT2 timing belt.  otherwise you would
> need to use GT3 and 8 to 10mm, that's no longer "cheap".
>
>  make sense?

Yeah, absolutely. I'm sure there are reasons not to gang two or three 6mm
GT2 belts on an extended pulley to get more strength?

Benson Mitchell