Hey Jamie,
This is way awesome however I think it could be done even easier.
Our have a large height/width ratio. So a 7cm diameter cup can be 24-26cm high. Also when pushing them i to place we would run into the guide rails than run across the 800mm wide side.
I will post in seperste posts as I can only post 1 image at a time
So this is a row of 12 cups {one of the cups we produce}. The target array for this one is 12-11-11-etc. This repeats to a total of 20 rows. The cycle time is about 12s for every two cups.
If this align with some spacing it’ll look like on the picture. (I didn’t draw the prolongations of these 6 bars that keep them in place.)
Because the surface provided (card board tray) is relatively uneven they need to picked and placed not pushed. I envision something like this.
Mounted to the Z axis is a double carriage across 800mm (leaving only 100mm for x-travel but that is ok). This static bar is shown in green.
From this static bar there are protrusions into the cup shown in red.
Inserted in 3d-printed ASA guides are these fingers in cyan. In their default position there is a 20mm gap between the red protrusion.
To actually move these fingers I envision 2 pneumatic actuators in position 3/4 and 9/10 (so that each actuator moves a total of 6 fingers;3 on each side).
These actuators have a hall sensor feedback option and come in various configurations. I really just picked something out of the internet that had a 20mm stroke. The one shown here is obviously to big.
Then the actuator (mounted backwards on pic) extends it squeezes the cup in between the two fingers. The weight of the cup is 22gr, so even a slightly rubberized surface will be able to hold on to it.
As in my previous post, this gripper design keeps one side open, in order to stack the cups currently held tightly against the previous row.
that is a nice machine, but look at the footprint. And I wanted something that runs open-source for me to fix it myself if I have to.
That all looks good. The end effector is going to end up being pretty large (long) and not made of foam (so weight is a concern).
I was originally thinking the LR, with just a huge plate instead of the 611 plate. But the trouble is that you want to bring these cups in from the side, and the gantry on the LR is in a direct line, so it can’t easily reach over the incoming port.
The MPCNC could have more than one gantry though. There is still a lot of mechanical problems you’d have to work out. But here is a multiple gantry beast by Ryan:
If you were able to hang all of the actuator stuff off one side, then the LR could still work. It would have to reach over the line of new cups to pick them up. The rollers would need to stay clear of the incoming cups.
Let me just clarify. You have cups filling up some sort of table, you want a machine to palletize them or just make a single layer?
You are asking for a one step but it might be easier for more. You have some sort of machine that is dispensing cups, you could pickup or load directly from there. your end goal is a layer, but what is next?
Do you see what I am getting at? If you had a machine to make a layer, move down, add a sheet, add cups, move down repeat. Sometimes the best solution is not to replace a person exactly but to skip steps humans need that robots do not, (bend over and entire pallet). Also remember it does not have to be as fast as a human as it will run constantly and with no breaks ( I understand speed it always better but baby steps speed is easy to add after you have something that works).
So I guess what exactly does the cup output look like and what does the end result look like?
I am seeing more of a guide-way fed directly from the out put machine that simply steers each cup into place and drops it. Then moves the entire layer down, add slip sheet, repeat. Easy to signal when no cup moves, and easy to signal if it hits a cup in the wrong place, then human intervention.
Very similar to this, just much smaller.
Hi Ryan,
yes the machine is supposed to do only one layer. I think I mentioned this in my initial post. After each layer the machine best gives an accustic signal to wait for the OP to reset the stage and hit GO.
Output is about 10-14s for 2 cups. The rest is math.
That’s a lot of useless complexity, moving parts, mass (hence inertia) and expensive to manufacture parts compared to just succion cups.
Plus you have to account for the fact that the cups will tilt when you’ll pick them this way, so when you’ll release them the top of the cups will move back by a few millimeters so in the end they will not be touching the previous row, which will add a small distance that will add up at each new line until it will cause some more serious problems.
Just go with succion cups, what’s the point of making things complicated?
Ok, let’s consider this as I think you have some valid points. Maybe you can answer some open topics.
I made you a quick 3D drawing as it will be easier to explain this way:
Just attach this thing somehow to your Z axis and make a system of tubes going to each vacuum cup’s rods.
Very easy to do, can be done in a few hours 
Suction cups are usually a long hollow threaded rod with interchangeable cups attached. They exist in any possible diameter so anything that is smaller than the size of your cups should be ok. As you can see, they also exist in several different height, so that they can better comply if, for instance, your tray isn’t very flat (See the difference between “S1” and “S3” models in the picture hereunder).
I don’t know where to source that in your country, but in China it’s very easy to find:
Yes, they would be grabbed from the top, as you can see in my previous drawing. That’s why I said earlier that you will need to select a fast an powerful Z axis motor if you want the machine to be fast. But it’s easy to solve.
I don’t really know much on the matter. I know that some systems exist in order to convert compressed air into vacuum, or maybe just a simple vacuum pump could work, I suggest you to look at what they use in the industry to power those kind of things it should be fairly common. 300g is really nothing, pretty much anything should work as long as it’s designed to run all day long.
Hi I guess this is an option to consider, I’ll have to think about it.
But one question, It would be great if I could find a suction pad of lets say 10cm x 80cm (with unused wholes self-closing). That way I could pick them all up contacting the cups top edge, this would also leave any cups with wholes in their walls standing (this would be ok)
So, my thoughts: I agree that a suction system might be a good option. Consider this, the suction can be applied to the top of the cup, it is not necessary to get to the bottom. That way, one array of suction holes can accommodate both the 15 and 16 count rows. No need for long tubes going to the bottom. and no need for two arrays or even shifting. There is actually no need for actual suction, just negative atmospheric pressure which is easy peasy to create with a fan or a venturi
Would you need a venturi at each cup so that any cups with holes or missing cups would not affect the pressure of the other cups? Still probably pretty simple solution.
And I thought we just got weird about “nema 17”…
Some people just like to be pacific.
@NaszNiemiec, @jeffeb3 Hard to describe what I see in my mind, but in its simplest form: the cups come off the machine in a row as seen in the pic. A flat bar, with an array of ports connected to a negative air pressure device… (Could be just a squirrel-cage fan, and manifolded hose assy) on a rotational arm, lowers over either 15 or 16 cups (pretty easy to control this). A solenoid valve opens the ports to the neg AP… ambient air pressure holds them in place, the arm lifts, rotates 180 degrees and then lowers them into place in the cardboard flat,the solenoid closes and allows the AP to balance, releasing the cups. The arm rotates back to pick up the next row, the cardboard flat advances one step, and the whole process repeats. The issue of 15, then 16 cups is managed by placing 31 ports on the bar corresponding to the overlap spaces in the 15/16 counts, thereby always applying the neg AP to the cups, no matter where they sit. in the 15 count position, or the 16 count position.
But if there was a broken cup, or one cup didn’t have a great seal, the ambient air would ruin your negative air pressure really quickly.
Also, I don’t think it needs to rotate. In the picture showing them all in the box, if the bars and row of cups coming in was parallel to the rows going out, you’d just have to pick them up, and move them. No need to add a rotation to the mix.
I like the idea of having a positive pressure pump blowing air past 16 venturis (They are generally pretty cheap, I think. No moving parts). Then have 16 small tubes going from there to the suction cups (or just holes in some soft material, like Jamie’s prototype). The whole thing might be kind of loud…
At that point, it would be down to just the details. How often do they fail. How catastrophic is it (Does it flood the whole shop floor with cups?)? Writing some simple gcode by hand. Making sure the gantry won’t run into the infeed.
Like I said Jeff, hard to describe the picture in my head. Nothing wrong with the multi venturi idea I see your point about delivering individual NAP to each cup, but I guess I feel that it’s energy wasted and noisy. Think about the surface area of these cups… just a neg 1#/sq inch would actually exert big force on these cups… as long as the ports weren’t huge, even a broken or missing cup would not cancel out the neg AP supplied to the others, especially using a high volume source such as a squirrel cage fan. Secondly, if it was mission critical, sensors could detect if there was a problem with the array (missing cup, etc) and abort at 90 degree rotation and drop that defective array into a bin for manual attention later, and then just repeat the failed cycle. I’d argue that there is no need for Marlin at all, just PLC’s and some mechanical components. I’ll argue that a rotational idea is way simpler than a linear motion. I guess I should have drawn a picture LOL
