As almost all physical parts of the machine are produced as of now, the final assembly of the whole machine can start!
As the machine consists of distinct more or less independent units that will collaborate, there first needs to be a base-plate to mount on it! An OSB-board of 30 x 40cm builds the base for the machine. Such boards look pretty good despite being raw to all extents. It is expected an overall height of less than 20cm for the whole machine.
The Bin Units are assembled by gluing the yellow bins to the black base-plates. On putting the (in the meantime modified as to the prototype) template disks into the Bin Unit, a slight wobble in diameter could be encountered. It was too remarkable so that blaming the print-out tolerances was not the way to find the guilty party. In turn on having a look to the Tinkercad design, the reason was clear: For some (not understandable) reason, the hub of the plates was not concentric to them: A small fix to do, but some hours of subsequent 3D-printing to expect and getting the original printed out disks sorted out due to QA reasons … because all parts are printed with 25% infill only, the waste of 3D filament stays in bounds.
Mounting the both sensors per unit and the motor and the central hub is pretty straight forward. Finally, each unit is screwed to its base spacer block so that it sits high enough above the mounting plate.
After having mounted and adjusted the Bin Units and the central Carousel Unit, the machine starts in getting its final shape!
Next step is to assemble the motor driver boards (the PCB’s for these boards in the meantime have a population / count in its several evolution flavors of flies / rats / mice / cockroaches): And just as a cook starts to prepare his pizza as of his recipe, the ingredients need to be there to equip the boards, “roasting them for 4 persons”:
Also for these boards, a slightly different but fully compatible variant was done: Originally, the intent was to wire the boards power supply and I2C connectivity star-wise from the main unit outward, but that would have meant to have some more cabling laying around in the machine. That is why a more bus-wise cabling was intended then, thereby “daisy-chaining” the Bin Unit boards to the overall cabling bus. In order to make that smooth, the boards must support a 2nd connectivity port so that the connection to the next one can be done using just plain vanilla Dupont cables. As there already do exist boards in the original variant, these ones just will be used at the end of the but / on single connectivity as in the Carousel Unit. Finally, here are the for pieces needed for the machine:
After uploading the firmware using an USBasp programmer, the boards were ready for usage.
What also followed was the assembly of the Head Unit, already labeled as Mischa as being the name for this machine:
The Head Unit is almost as impressive as an ATAC-unit (but still misses the matching movie…).
One could ask what these both red and black jumper wires on the display do mean:
The answer is that they are the result of an irritating bug: On an early hardware-only test of the assembled board (after having checked proper isolation between paths and the proper connectivity of the parts), it turned out that the board remained dead (no Led or display backlight lighted up), but there was some significant current draw at the same time bringing the supply voltage down. As current * voltage = power, the third part of this formula also soon manifested itself by getting the display unit slightly warm.
Despite immediately having disconnected the power, a bad feeling came up that this incident might have had consequences: Due to the symptoms, there must have been some short-circuit like effect on the board, and such symptoms unfortunately often can be as lethal as e.g. heart attacks. As the board itself was checked before and while assembling for short-circuit effects, only a component could cause this behavior. As the microcontroller was not yet soldered on, mainly the display unit remained being the candidate of trouble. Checking the circuit diagram and the unit datasheet soon unveiled the disastrous error: The power supply voltage was reversed!
So in order to fix the board, the both circuit paths to the power-supply connectors were disconnected and these both jumper wires were added in order to supply the power as required. But a subsequent test unleashed that the display unit had gone forever.
Thus, there was the need to replace this as “solder-and-forget” intended unit against a new one. De-soldering the old unit was no joy at all, but after replacement, the hardware-test passed as expected: Beside the backlight / Leds going on, there could be observed the driven pixels of the display and the contrast was adjustable:
That’s all to expect with a hardware-ready board not yet having any kind of firmware on top.
Next step was to do the wiring on the machine in order to interconnect all those small electronic parts. Finally, the Revolution PI was mounted to a small cap rail as commonly found in professional environments on the plant floor. Thus, the Revolution PI is a device being capable / ready-to-mount to be used in industrial environments as e.g. also professional PLC’s are built for.
Thus, the final machine looks like that (just be aware that the Head Unit is not yet wired and screwed on as it still needs to be programmed – later on, there still will come connections towards the Bin Units and the RevPI):
Now, everyone can imagine and see how the final machine looks like. What follows is the firmware programming for the Head Unit, what is yet another manifold topic for itself…
To be continued…
Da Vinci Space 
You must be logged in to post a comment.