Starting on the bicep printing and assembly
I was actually expecting to start on the assembly and testing of the hand and forearm but the servos I had ordered to be delivered as soon as I got back from vacation did not arrive. I ordered six MG996R servos from Hobby King but they never shipped and it appears that they were terminally out of stock. So, as soon as I got home I ordered another set of six by way of Amazon. They shipped almost immediately, albeit from Germany, but they got here in about 7 days.
In the mean time I decided to proceed with printing and starting to assemble the bicep and I immediately sank into a new session of discovering how many ways a newbie with 3d printing and a newly assembled 3d printer can use up rolls of filament. One of the really important concepts I began to understand is that almost every piece to be printed has a seemingly new set of potential screw ups embedded. Just for fun here is a list of some of the things that went wrong:
3d printing boo-boos
I discovered that I had been trying to print using a .35 nozzle with slicer set to a .5 nozzle. One of the myriad settings that I missed. This was causing real bad stringing and clumping.
I really did not understand about setting the starting Z height correctly primarily because trying to set this accurately with the Z stop switch holder that is standard on a Prusa is a nightmare. I finally sorta set the height somewhere and the used the Z offset up and down to find something that worked. This was not a very effective way to operate. Later on after another week or so I build a screw adjustable Z height bracket and used a better switch holder from Thingiverse and then learned to actually get the four corners of the bed to be level. I ended up with a good working height to be about 4 sheets of laser printer paper.
I did not grasp the importance of the belt tension ( y belt this time around ) and was seeing some prints that looked like pictures I saw online of y belt tension too loose. Tightened up the y belt and printing improved a bit. For some reason, checking the x belt didn't occur to me. A while later I made a little adjustable belt tensioner for the y belt. That really resolved issues with keeping that belt tight.
I was having problems with the sides of my prints with ripples. After poking around on line I discovered that this is called 'Z wobble' and can be identified by holding something printed up beside one of the z screw rods to see if the ripple pattern matched the threads. The pattern matched. I set about trying to figure out what was wrong and how to resolve it. I tried a variety of things and finally decided to replace the z threaded rods. That resolved most of the issue but there is still a slight detectable wobble pattern. I'll get back to fixing that when I get some time.
In the process of getting printed pieces to stick to the bed and not warp I have gone through a bit of trying different forms of black magic. Seems like black magic anyway. This period started with me using Kapton tape on glass with a coating of abs/acetone on the tape. I went through phases of this working very well and then not working at all. I tried varying the hot end and bed temperatures, printing directly on the glass, printing on the glass with different thicknesses of hair spray and finally back to Kapton with abs/acetone. What I did not realize at the time was that a lot of the problems with sticking were related to Z height and bed leveling. I now print on glass with Kapton and a fairly light coating of the abs/acetone mix. Another thing I now do is to spray the clean glass with windex before applying the Kapton the push the bubbles out while the Windex is still wet. Then when the Kapton is dry I print on it. Works to a treat.
One of the problems I am still having is how to get a good print of things like to worm gears for the bicep and shoulder. If I don't use support I get good tops and saggy underneaths that make a really unusable print. If I use support I get so much crap stuck to the undersides and edges of the gear faces that I have to whittle till my fingers hurt. I have seen people say to use a fan to cool the plastic quicker but I haven't seen any examples that are close enough to my combination of equipment and models.
Finally, the x axis pulley gave up. It cracked and lost a tooth. The indications from that were so obvious that I couldn't miss them. Having the belt skip a tooth every rotation gives a rather distinctive pattern.
The bicep
Hope all of that wasn't too boring. Getting up to speed with the printing and assembly of InMoov is a fascinating process and I can't help talking about the process.
Anyway, I went through the process of printing out all the bicep pieces for the right arm (some were printed multiple times) and started the process of assembly. Since I was following the bicep assembly help section the first thing up was to modify a couple of hs-805bb servos for continuous rotation and for external potentiometer. A problem I ran into with the procedure is with pulling out the motor and small pc board. No amount of coercion would allow me to pop out the motor and board. I later found that the motor was securely glued in place. What I did was to unsolder the motor from the pc board and solder it back when I was finished. In order to unsolder the motor there are three soldered points to undo. In the following picture they are the two points on either side of the bottom of the motor and the point on the right side and a little above the right motor point. The third point is a ground connection. You will need a solder remover tool to do this de-soldering.
When I did the first servo I did not realize that the third point (ground) was connected to the motor and I broke the wire loos from the motor. To reassemble the servo I had to clean out the solder, solder a wire to the frame of the motor and then solder it back to the pc board with the other two points. Here is a picture of the bottom of the servo with the wire soldered to the motor and before the replacement of the pc board.
The rest of the procedure is essentially the same except that it is not necessary to remove all the gears, only part of them.
Bicep rotation gear assembly
The most troublesome part of assembling the bicep was the arm rotation assembly. Between the problems with trying to get a decent print of the worm gear and the really tight tolerances for the whole assembly I spent many hours fitting and whittling and fitting and whittling, etc. It was and actually still is a real challenge getting this part to work correctly.
Once I got an assembly that would operate sort of smoothly I decided to spin it for a while using my electric drill. I build this little attachment using the end of a broken worm gear and a bolt.
Then the assembly was assembled. I did put the cover on before running.
and then I chucked it up in the drill and ran it for a while going in both directions. I then opened it, blew out the dust and checked that the works still worked.
The bicep assembly
After messing with the arm rotation assembly I did the rest of the bicep assembly. Most of this was reasonably straightforward work. The instructions were easy to follow as long as I followed them. Here is a partial assembly with the attachment piece for the forearm. Actually if you look closely this is assembled incorrectly. The bicep assembly is mounted to the rotation plate upside down. Part of the instructions that I did not pay close attention to was to test the assembly before glueing it together. This one went in the trash and another was printed and assembled.
Here is assembled arm before adding the bicep skin and just before connecting the servos to the computer to do some more testing. On the left edge of the picture the Adafruit servo board and Raspberry Pi are just visible. Also, the Eagles on the tube.
That's about it for the bicep. Next blog post will be about the finishing of the hand and forearm. I have printed all the parts for the torso and had finished about half the shoulder parts before my x pulley broke (see above if you haven't already). I am really excited about maybe getting a robot together soon. I need all the friends I can find (or print).
Later
No comments:
Post a Comment