Thierry-GearsManiac

It reminds me the intermittent mechanism found in this old GBC Cardan Lift version : In order to achieve the intermittent motion, any oscillating planetary mechanism (a differential is a special case of planetary mechanism) can do the trick. As I observed in several designs (usually GBC modules), the simplest LEGO implementation of an intermittent motion (now found everywhere) relies on an eccentric gear. Examples : Torso's Cardan Lift, Tomáš Ullrich's Draw Bridge, and countless of other designs. But, with any of these solutions, because the output gear's speed changes continuously, it never stops completely : close to zero for a more or less long time, then high on a brief period (but I currently don't master the theory or the factors which govern the steepness of the transition).

On my side, here is the video of the "rhomboid chain drive" passive arm in action : https://diode.zone/w/89d1z9XTv6yCJRfXFx6Ay6 The arm segments are 10 studs long (spacing between the centers of the axles). Built using half-thickness beams/liftarms (in order to stay at 3-stud thickness), they are slightly stiffer than an axle-and-connectors build, reducing pitch and yaw for the cradle. By chance (again), the chain didn't require tensioning gears (although I provided room for them, using triangle liftarms) and therefore exhibited no backlash ==> (almost) no roll. Tests with big gears pending : "Expert Builder" gears do in fact have a very low backlash : 1/2 tooth for 4 meshings from one end to the other, thus a very low angular backlash. (for information, the modulus is 1/3 stud, i.e. the tooth length on the pitch circle is pi/3 = approximately 1 stud) "splat gears" experiment not yet built, only quickly tested. With a modulus of 1/2 stud and using the biggest possible gears (Z14) as end gears and idlers, the whole arm length is no more than 2*14 = 28 studs when unfolded at 180°. Aesthetically speaking, these gears could be used for giving a "kid's toy" style to the device (one could build the frame as a wall of bricks of a single color, typically white, but I don't have enough of them) That's why I prefer to finish building of the module in its current form...

@Ankoku — KEN （レゴテクニック勢） (@kenken_lego) November 3, 2021 A revival of Akiyuki's Cycloid Drive (but with a much lower number of "teeth") !!! — くう (@ZVwQnK5poHS87nv) February 4, 2022 Although the trammel of Archimedes has already been implemented in this old mesmerizing standalone GBC (where used as a 4-way splitter), and in Riku Katsumata's Orbit Overlap module (where it spins on its center and ejects balls), I had not been able to find another use of it. I however tried to think about it when starting LEGO again two/three years ago (I would have liked to design a GBC module who would make use, almost this way, of two variations of this mechanism but I gave up when discovering how difficult it is to tame the balls (even on much simpler usual mechanisms). I however managed to prototype a "trammel of Archimedes without guiding rails" (it relies on gears), but it's too flexible for picking up balls, and therefore can only be used as a kinetic sculpture (perhaps should I finish and publish it, but it's a small design compared to everything here).

Although I ordered all my Z36 gears on Bricklink (a few years ago), I'll try to find traceability/identifying markings on both of them... EDIT20220227 : the Z36 gear on the left (probably the current version, same as pictured on Bricklink) has no markings at all, whereas the one on the right has "51436" and "01-02" deep in the blind holes around the center axle hole, on one side. Yes, I already encountered this kind of situation when thinking about other symmetrical designs involving the meshing of identical Technic gears (either directly or with a worm gear between them). But I don't think that LEGO will create tooth phase variants for every Technic gear size, otherwise many people won't notice these subtle difference among variants, mixing up them, and having trouble when paralleling them. With teeth numbers being multiples of 4, the phase remains the same for any 90° insertion orientation on an axle. Other gear families however contain gears with teeth numbers either odd multiples of 2 (all splat gears for example) ==> half phase per 90° step, or purely odd (all old "Expert Designer" ones) ==> quarter phase per 90° step.

As seen on your video, the stiff telescopic rack does still improve the reliability by stabilizing the cradle, more precisely by reducing its pitch (back-and-forth tilting) and yaw (left/right turning), despite the weight of the cradle and its load, and even in the absence of gear-based permanent roll (i.e. side tilting) control. The stiffness of the arm also plays an important role in both of our design variations. The only critical point that has to be solved (already solved in your previous design) is the roll cancellation at the entrance and the exit, thanks to a wall on the outer side and a funnel-like guide on the inner side (on the input structure).

Something else. Something never seen before between versions of Technic double-beveled gears : on the Z36 one, besides the difference on the rim (its thickness) and the hub (shape details), the phase of the teeth is inverted : "peaks" aligned with beams on the thick one on the left (probably the latest version) "valleys" aligned with beams on the thin one on the right (probably the older version ; only two of them in my inventory) However, the tooth phase doesn't change between versions for the Z20 and the Z12 (I placed the latest versions on the right).

For the Mini figure, I unfortunately own very few of them (including accessories). Here is a close up picture of my current mechanism : (removed from the module, rear view ; the black 4x2 L-shaped liftarm was attached to the frame ; a few extra #44 connectors nearby = the parts used for the elbow) In my build, the spacing between the axles of adjacent Z36 gears is 4.5 studs (or here, approximated by the hypotenuse of a 4x2 right-angled triangle). Therefore each arm segment is 9 studs long, and the maximum allowed radius of the unfolded arm must stay below 18 minus a few studs (i.e. never undergoing a 180° angle). However, your drive mechanism design is much bigger : the half width and height of the square traveled by the cradle pivot is 12 studs (frame), plus approximately 4 studs (distance between a sprocket's center and the pivot mounted on a tread link passing on it). The biggest distance from the cradle's pivot to the center is therefore 23-24 studs, which means that you'll have to either build a very much longer arm system (i.e. more than 12 studs for each arm segment = more than my bulky turntable-based early prototype), or reduce the size of the drive mechanism. Therefore, if we keep its size as-is, in order to increase the length of the arm segments, one must either : add extra idler gears, at the expense of a slightly higher backlash. For example, adding two Z20 gears per arm segment would increase their lengths to 14 studs. (EDIT20220224 : use the (big) "splat" (a.k.a. "flower petal") gears, hoping that their higher/worse linear backlash will be mitigated by their bigger radius/diameter, keeping the angular backlash low enough on the end gears. The other big gear families ("Expert Builder" and "Samsonite") probably have the worst linear backlash). switch to the folding diamond-shaped (or rhomboid) chain solution (I plan to show it in action in an upcoming video), because the upsizing is only a matter of adding chain (or tread) links, which doesn't increase the backlash ; the only challenge is to adjust the optional tensioners. On another side, your use of the sliding rack (a pair of parts I don't own yet) does give me some extra ideas : trying the telescopic shaft solution (or even the use of the sliding Z8 pinion (11955)).

OK, but this arm alone won't prevent the cradle's axle from spinning with respect to the frame. On my side, slowly working on my prototype, I managed at last to make a video of it in action, using the geared arm variant (rebuilt with Z36 gears only) : https://diode.zone/w/aNr5BcPZWSpTCHq2VeYzCk I just noticed I forgot to show how impossible it is to tilt the cradle (at least beyond the backlash), but the moving of the passive gears speaks for itself. The bar supporting the minifig shows the circular translation of the elbow gear.

@Doug72 In your proposal, there seem to be too many degrees of freedom, so that the joint holding the cradle will be able to rotate freely on some range of angles. it's almost like having two successive standard hinges (i.e. three bars) instead of geared hinges (although the kinematics are slightly different). You can try it by attaching the center end on a stand and manipulating the other end with one of your hands. @aeh5040 Parallelogram linkages (i.e. two in series) could have done the trick as well : I've already thought about them. But if one wants to prevent the parallelogram from turning over (and failing by getting crossed in the absence of a 3rd link like in Schmidt couplings), then the center point would have to be outside the rectangular tread and the two arm segments long enough. A Schmidt coupling could indeed work, but because it constrains the input and the output to be single-sided and on opposite sides, it would result in a rather bulky build : from the rear to the front, there are : 1. the input stand holding the input "star" ; 2. all the intermediate parts of the coupling ; 3. the output "star" ; 4. the tread (traversed by the axle joining the output "star" to the cradle) ; 5. the cradle ; 6. a stand holding the frame of the tread drive (otherwise it would float in the air !) Instead, if some Schmidt coupling flavor is still desired, one could still design a hybrid arm system where the "arm" would be the first half of a Schmidt coupling and the "forearm" would be any single geared arm segment flavor (spur gears, chain etc...), so that the output axle can now extend on both sides (to the cradle in front, and to the tread behind).

Here is my (very slow) beginning : First test of the core mechanism, 20*14 frame, with the bevel gears-based arm : Because of the thickness of this variation, I had to recess it into the rectangle while giving the sprockets a double-sided support and preventing the arm from colliding the tread. By chance, the tread length fits perfectly without the need of a tensioner. Unfortunately, the whole arm is too short and reverse-folds itself when almost fully extended. Therefore I reduced the height to 12 studs (but a tensioner is now required). Here is a quick-and-dirty height test (after building a cradle as well) : Unfortunately, there is too much backlash that may later prevent a reliable alignment between the cradle and the pickup area : rotation backlash caused by the bevel gearing twist backlash and forearm flexibility because of the single-sided support of the gears I therefore plan to give up on recessing the mechanism, to dismiss the bevel gear solution and to revert to the 14-stud height : I already rebuilt enhanced versions of the spur gear arm system and the folding rhomboid chain, only 3-stud thick and with proper bracing (i.e. double-sided, thanks to half-thickness beams/liftarms or specific axle-and-connectors build) : coming later...

Another user even found a practical GBC application for this mechanism ! Awesome and mesmerizing !

Building only from Technic (i.e. studless) parts : I didn't notice this at the first time. Was it for aesthetics purposes and/or for a self-challenge ? Because building fully studless seems challenging for my design skills and my studded background (before my dark age... whereas I now take advantage of both design strategies). Perhaps am I going off-topic.

In order to eliminate (or at least to mitigate) the falling balls problem, would it be a solution to replace the cradles' outer fingers by walls = kinds of "armrests", like in the original design ? On my side, I think I should take up the challenge of building a one-high-speed-cradle variant (without reverser of course, and with the goal to test my horizontality mechanism in real conditions), but I expect a lot of falling balls anyway !

Understood (and I haven't thought about this very simple solution which fits the existing design). The front/back slight tilt may be caused by the 0.2L excess height of the guide, which is 2.2L high (1L for the white beam below, 1.2L for the brick height of the guide). Replacing the white beam by two plate layers will remove this excess height and may solve this little issue.

Nice variation of the reciprocating sweepers often seen on GBC videos (the ones with a ratchet for directing the back travel). The originality in your module is the linear oscillation thanks to a 2-bar reciprocating arm (instead of the use of a big rotating arm).

For the loading and the unloading areas, the setup from the original module seems to do the job. The only remaining concern may be if its lowest height will allow the height of input bin to comply to the GBC standard. Do you see any further difficulty ? On the video of your first version, I saw the carriage tipping when travelling through the top side because of the guide under its underside : for guiding the carriage there, its top side should have been horizontal and flat, and with a rail above it. And now, as I managed to build most of the mock-ups of the translation mechanisms I thought about, I'll introduce them to you (reminder : the key idea is that the gear ratios must be +1:1 for output/elbow and elbow/input) 1. The double arms : above is my first build (spur gears), and below is another one, bevel-gears-based (just notice the Z28 at the elbow, whose both sides are used). 2. The rhomboid chain : Inspired by this video, at 13:10 : https://youtu.be/PM4_9JnTq-E?t=790 In fact, this idea is a variation of a double arm with +1:1 chain drives on each segment. The chain perimeter will stay constant for any folding state of the rhombus linkage because the input, output and idler sprockets are of the same size. The linkage can also be parallelogram-shaped (less aesthetic because of the symmetry loss) or even "kite"-shaped (symmetrical) (note that the use of torque-limiting gears for the idlers is purely aesthetic : this special function is not used) 3. The single extensible arm (exploded) : At both ends, you have 90° transmissions which, when combined, achieve a +1:1 ratio. Between them, an extensible axle with a big enough ratio (in general, slightly higher than 2:1 because the overall ratio between folded and unfolded state --including the non-extensible ends-- is around 2:1). Two solutions presented : above : a double sliding coupler (several implementations exist ; mine uses several 3-stud half-thickness liftarms on the center element and https://www.bricklink.com/v2/catalog/catalogitem.page?P=98585 (looks like the hub of a Z36 gear) on the ends) below : the mix of a Sarrus and scissor linkage. Similar to this : https://www.youtube.com/watch?v=DasxDJJF2hk . Uses both kinds of 120° connectors I didn't build the latest mock-ups at the right dimensions (i.e. the ones which would fit the real square tread) : they were built quickly, only to show their way of working.

I now see the key elements for a good guidance of the carriage : the contact between the underside of the carriage and the ground "rail" (the series of tiles) for the horizontal travel at the bottom the two lateral guides for the vertical travels; with funnel-shaped entrances/exits for the self-aligning of the carriage For the top horizontal travel, no need to maintain a tight horizontality of the carriage, because its low center of gravity may be sufficient. Understood. The problem is that the geared sections of these parts won't find any such particular use in my mechanical arm : all Z40 gears rotate freely with respect to all the arm's segments. I will indeed happily show a few more mock-ups (or paper-drawn sketches at least).

As promised, here is a quick mock-up of the "double circular translation arm" for keeping the carriage horizontal without guides : The 5x7 frame represents the fixed center and the quarter ellipses represent the carriage. The achievable distance range (from 13 to 21 studs in your model) required gears bigger than Z40 (if we minimize the number of idlers), hence the turntables (here Z56, but one could use Z60 turntables with Z36 gears instead). It's quite challenging to make this mechanism as slim as possible (in depth and in width) while minimizing the backlash at the same time (less idlers and/or good bracing). Other implementation variations may follow at a later time (likely after Christmas, due to a lack of time). Just tell me if you will still be interested in seeing them in your topic.

OK... And also, if the two arms end up being too bulky, then they perhaps may be replaced by a single extensible one : 90° gearings at both ends and an extensible shaft between them (if one can design its sliding coupling with a sufficient stroke). I'll eagerly follow your work...

The chain tension indeed does match perfectly with the fixed sprockets' positions (I can't find the right way to translate) : it's often hard to adjust by trial-and-error the spacing between sprockets in order to achieve the right tension, especially when a spring-loaded tensioner is not an option like here. I wonder if some people already wrote some formulas for this, for the various kinds of chains and the corresponding sprockets, knowing their dimensional features. The support for the horizontal sides (especially the upper one) seems also nicer and "falling" at an exact position too. And meanwhile I found another crazy idea for maintaining the horizontality of the carriage, which this time could even add a nice visual effect without hiding the main mechanism : it relies on two "circular translation arms" in series (I should post some sketches later because pictures always speak better than words), one end fixed at the center and the very opposite end on the carriage's pivot.

I can imagine the difficulty of keeping the ball carriage horizontal all the time, especially when traversing the corners, when it disengages/reengages between horizontal and vertical guides. On my side, I would have thought about another solution, but with the drawback of hiding the nice simplicity your main mechanism : the ball carriage would slide on a horizontal rail, and this horizontal rail would be able to be lifted up and down (of course passively, by your chain-driven carriage) while kept horizontal, thanks to some left/right synchronization mechanism (several solutions : double rack-and-pinion where the left and right pinions are synchronized by an axle, adjacent parallelogram linkages, scissor mechanism, pulleys and cables, accordion door structure etc...). Of course, these mechanisms are passive (they don't drive the carriage by themselves, otherwise it would defeat the purpose of your main mechanism). but one of them would prove to be the most compact and reliable.

Unfortunately, I have never made build instructions for this associated feature. I only have the Stud.io 3D model for it, see the download link above. (ans I even can't work again on it or export a more visible picture because a LINUX update broke the working of Stud.io under Wine and I failed to make it work again) The only extra thing you need to do is to rebuild one of the robots mirrored, then to adjust the position of the single-bevel input or output gear in my new drivetrain. Perhaps I can help you on specific points first : a difficulty to find which sub-elements to remove ? Or to see how to install the mod ?

Challenging, I totally agree ! That's why I failed, even in pure theory. And real building would be even tougher, as I experienced in an advanced mechanism from the beginning of this year (a bell-meter which required a lot of redesign and fine-tuning... but was eventually still not enough reliable). For actuating the switch off and on with enough force, the mechanism made for lifting the ball receptacle would have to simultaneously rewind some kind of energy storage device (a counterweight or a spring/rubber band), then trigger it when the ball receptacle reaches its down position (without falling), making it turn the switch off and stop mid-race, locked. Then, the last ball (either by following another path as you suggest, or because of the total weight of the ball receptacle) would unlock the energy storage device, which would, thanks to its remaining potential energy, turn the switch on again. Some kind of clock escapement anchor would be mounted on the switch in order to move it back-end-forth during the discharge of the stored energy, but the rewinding phase should not act on this anchor. With my current LEGO skills, I would be only barely able to build some prototype of the core of such a mechanism in a reasonable time, but not a complete and highly reliable GBC module. Another difficulty I see is that the ball input should be able to work with no stirring or delivery mechanism, which seems hard to achieve, otherwise the motor should stay permanently on for driving it, ruining the purpose of the core mechanism.

Original idea ! And rather advanced mechanism for sequencing the successive states. On my side, although I already thought about such principle (but only in a theoretical way), I couldn't manage to find a mechanical solution, given that I gave me an extra design constraint : powering down the motor while the platform is waiting to be fully loaded, i.e. turning it off when it returns to its starting position, and turning it on again when fully loaded.

The pushing/pulling force that moves the vertical gutter horizontally back and forth was applied to one end of the gutter only : inequal forces between the top and the bottom (plus some possible friction at some of the slidings) may result in a "rotational" move of the gutter which would indrease the friction forces and lock it. In this new design, this force gets applied to almost the center of the gutter, so it's better balanced between all slidings (think momentum / torque). And if I ever had to build this module, I would double the crank : the original one at the bottom plus one at the top (plus a sturdy chassis for it). (alternate solutions based on synchronizing the top and the bottom thanks to linkages --like for the space bar on a keyboard--, racks and pinions, pulleys and cables could do the trick as well, but would be too bulky)