Well, It's time for some more upgrades...

[Alan Timm]

It's that time of year again, and Gray's getting stripped down in preparation for some upgrades.

Here's what I'm doing to him next, with urls for the stuff I can link to.

 - Bob Houston's 3 axis articulated neck mod

http://www.thingiverse.com/thing:845793

http://www.thingiverse.com/thing:1243403

 - Gael's Stomach

http://inmoov.fr/top-stomach/

http://inmoov.fr/build-yours/

http://www.thingiverse.com/thing:1289287

 - Gael's V2 eye mechanism

http://inmoov.fr/eye-mechanism/

 - My Torso mod for the XBoxOne Kinect

http://www.thingiverse.com/thing:1289291

 - Monnerby's InMoov hand dual track servo pulleys mod

http://www.thingiverse.com/thing:1074604

I've had Top_Stomach printed out for a while, but I wasn't keeping watch for the rest, so I just recently fired up the printers for the Middle and Low parts.

I also need to take another look at the articulated jaw, I could never get the fit right, so it stayed closed. 

Should make for a fun couple of weeks.  Wish me luck!

Alan

[Dwayne Williams]

Good deal keep us all updated on your progress.

[revwarguy7]

Could you post a little more detail about the Kinect mod, or show a link to same?  

   Like what problem it solves,and how it does that, etc?   

   There's nothing on Thingiverse except a warning not to print it.

TIA, 

[Alan Timm]

Hey revwarguy7,

Gael's original design is for the XBox360 Kinect.  

I've been plugging away at the changes need to mount the XBoxOne Kinect instead.

I was using a very nice model of the sensor to base the changes on, and just got ahold of the actual Kinect this weekend.

https://grabcad.com/library/xbox-one-kinect-sensor-1

It requires changes to four torso frame pieces.  I'll also need to modify one of the Top Stomach pieces (TStomCovRightV2), and add an additional chest piece to hold the sensor in.

I'll most likely have the torso frame pieces tested and uploaded tonight.

Alan

[revwarguy7]

On Sunday, January 24, 2016 at 3:10:54 PM UTC-6, Alan Timm wrote:

Hey revwarguy7,

Gael's original design is for the XBox360 Kinect.  

I've been plugging away at the changes need to mount the XBoxOne Kinect instead.

I see - good luck with the Kinect One! 

[Alan Timm]

Ok, the design for the XBoxOne Kinect mod is up and finished.

http://www.thingiverse.com/thing:1289291

All the frame pieces have been test printed, and everything fits well.  That sensor model that I used was really, really accurate.

I like how the sensor mounts to the torso, it uses the TStomCovers for a bit of support, but SensorMount really locks it in.

I'm not that great at organic modelling, but I think I captured the general feel of the TStomCovers.  If anyone else wants to give it a shot, let me know.

I'm working my way up from the base, so I'll start printing out the final pieces for this mod next week as I wrap up the stomach parts.

Until then, here's a spiffy render.

[gael langevin]

Hey very good Alan,

Your modification look very clean and acurate!

I have the kinect in my workshop in a box waiting for tests, unfortunately it requires USB 3 connection and Windows 10. None of my computers are compatible for now.

The only one that could do the trick, is the Lenovo tablet that runs InMoov, but it is currently under Windows8.

The weight and size of the New Kinect is huge, not mentionning the size of the cables and power supply. It is one of the reason I didn't yet try to adapt anything for it.

I am glad you did, thanks for sharing!

It's added in the InMoov derivatives collection.

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

--
You received this message because you are subscribed to the Google Groups "InMoov" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.
To post to this group, send email to inm...@googlegroups.com.
Visit this group at https://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/d/optout.

[Alan Timm]

Thanks!

I'm looking forward to running the final prints for it this week.

I actually haven't gotten far enough into the new sensor except to verify that there are drivers for it.  

I've seen the demos though, that it looks like alot of fun, with alot of potentional.

I'm pretty sure that the SDK requires Win8 or later, but I haven't gotten that far yet.

https://msdn.microsoft.com/en-us/library/dn782036.aspx

Unfortunately, it does require an adapter for power, which isn't cheap.  

http://www.microsoftstore.com/store/msusa/en_US/pdp/Kinect-Adapter-for-Windows/productID.308803600

It also seems to be picky about USB3, requiring a USB 3.0 host adapter that supports Gen-2

https://msdn.microsoft.com/en-us/library/dn782035.aspx

I've seen a few articles on how to hack power into the sensor without the adapter.

http://stackoverflow.com/questions/25952296/how-to-connect-kinect-2-for-xbox-one-with-pc

Also appears to be supported in Linux under OpenKinect

https://github.com/OpenKinect/libfreenect2

Alan

[Alan Timm]

The finish work for the pelvis sub-assembly is done, and it's mounted to the pedestal. 

My first crack at a wider pedestal used black 2" PVC, but Grey ended up wobbling about like a sad, jack-in-the-box.  The PVC was much too springy.

Picked up a length of 2" OD 1/8" thick aluminum tubing, and that works perfectly.

I'm also trying out the HobbyKing HK15338 digital servos for hip rotate.  

They're digital, which is nice, but they're 90 degree servos by default, and are not programmable.

But, they can be hacked for full 180 degree rotation by using a pair of 1.2K resistors 

Not bad for $16.

http://www.hobbyking.com/hobbyking/store/__42447__HobbyKing_8482_HK15338_Giant_Digital_Servo_MG_175g_25kg_0_21s_US_Warehouse_.html

  

[Alan Timm]

Final assembly for the pelvis and torso.

I'm working on the torso and chest cover panels this weekend.  They're all printed out, and they're getting sanded, primed, sanded again, and painted.

[Alan Timm]

Today was a good day for painting.  All of the chest and torso panels are finished and installed.

Once the paint cures I'll paint the embossed InMoov logo.

[Alan Timm]

Almost done!

All of the chest and head plastics are either replaced or refinished.  Some of those panels were printed a couple of years ago, and I've gotten a bit better at printing since then.  :-)

Unfortunately I couldn't use Monnerby's dual track servo horns with the 180 degree wrist mod.  I'm not sure why, but I had to revert to using the large horn for both sides.

So the arms are reattached.

I got caught up in messing with the eye mechanism design, so I wasn't able to finish the head and neck assembly this weekend.

So all that's left:

 - complete eye redesign, print and assemble

 - complete assembly of Bob Houston's articulated neck

 - wire the new joints into the head microcontroller

[Mats Önnerby]

Hi Alan. 

I'm sorry to hear that the dual track servo pulleys don't work with the rotational wrist. I think it's because the lines are stretched now and that makes the problem with the rotational wrist more visible. Since the lines are off center thru the wrist, they stretch on both sides when than hand turns. And with a 180 degree wrist that problem becomes significant. I think you also need to put springs in the hand, something similar to what Gael has done. http://inmoov.fr/hand-and-forarm/

The parts that you need is TensionerRightV1.stl and LeftTensionerV1.stl from this page http://inmoov.fr/inmoov-stl-parts-viewer/?bodyparts=Forearm-and-Servo-Bed and a set of springs. 

I see that you have some electronics in the hand. Is it a Teensy that you are using ? 

[Alan Timm]

Hey Matt,

No worries, I'm sure they work great with the 90 degree rotation wrist.  It's exactly the right idea to address the second point below.

There are 3 challenges that I see with the current finger design:

 - Need to programmatically override the max and min values for finger servos based on wrist rotation

Whether you're using the 90 degree rotation or rockin' the 180 degree rotation mod, because of the way the wrist is set up, the max and min values for each servo will be different. 

If you get it wrong you'll burn out the servo.  I have an open/close function for each finger based on a set of experimentally mapped values for different wrist rotations, and it mostly works.

There's alot of voodoo in play when I set up the finger servos, just to make sure that they don't bind the wrist and have enough slack in the 50% position to allow wrist movement, while also having enough room to fully open and close in all wrist postiions.

I think it got a bit worse with the new wrist design with the grid of line holes.  It was a bit better with the larger single hole, but at the expense of the lines binding as they approach maximum tension 

As long  as the servos are in the wrist, I think there's no way around this.

 

 - Lines become loose during varying stages of movement

 - Need to maintain constant tension on lines throughout range of motion

Here's where I was hoping Matt's servo horn design would really help, and it did for any single wrist position.  

Today's designs allow for quite a bit of slack for the non-tensioned side of each finger line.  It's only by luck and line "memory" that they stay in the horn grooves, and I've seen them slip out occasionally.  It seems that after a couple of actuations they kinda remember their shape, and more or less stay where they're supposed to go.  Maybe it's a bit worse for the 180 degree wrist mod. 

One option might be to reshape the horns into a variable displacement cam, which could maybe maintain enough line tension throughout the range of movement.  That may be a future enhancement for Matt's dual track servo horns.

The one that I've been leaning towards would be to only actuate the close action, and have springs provide the tension to return the finger to the fully option position as the close servo is released.  That way there's always tension, there's never any slack in the mechanism.   I supposed if you added a spring to the grip actuation line you wouldn't have to worry as much about wrist rotation, although you wouldn't know exactly the position of the finger without it.

 - No force feedback to determine if you're overdriving the servos [grip strength, wrong end position, etc]

This is probably my biggest worry.  I've lost 3 shoulder servos to inadvertently setting an end position beyond limits, and poof.

The max and min end points for each finger are tied to the rotation of the wrist, and of course anything you're trying to hold.

Any time you set an end point beyond the max range, the servo will run full power until it melts. 

I did see someone attempt force feedback, and I also have a few ideas on how to use FSRs in line with the servo.

I'm not sure how much of this I would get to, I need a break after I get this set of mods done.

I suppose if I were to make this into a mod, I'd approach it like this:

Modify Servo sled to allow slight (1-2mm) movement in direction of line tension

Integrate an FSR (force sensitive resistor) into each slot in servo bed, to measure force up to 20 newtons (4 pounds) http://www.tridprinting.com/Electronics/#3D-Printer-FSR

Modify bracket to anchor 1 or 2 stage spring for each finger extension line (anchor joint 1 differently from 2&3 for nice, repeatable finger close movement)

Write code to back off finger grip tension when pressure approaches 20 newtons.  (With that, you may not even need to modify end positions based on wrist position, but probably would still need it if you wanted to know actual finger positions)

 

Alan 

[Alan Timm]

I've finished the changes and build for EyeMechanism2 and posted the files to Thingiverse:

http://www.thingiverse.com/thing:1328236

These eye supports are designed for the Microsoft Lifecam HD-3000.

I also made a couple of tweaks to some of the other files.

This V2 mechanism is sturdy, with no slop or play in the joints once everything is tightened down.

I think this means that the head gets reassembled tonight.  Almost done!

[nanomole]

Hi Alan - do you have a link to the cameras you used exactly ? . great work BTW !

[Alan Timm]

Hey nanomole,

Here is the exact camera I used for the build. 

https://www.microsoft.com/accessories/en-us/products/webcams/lifecam-hd-3000/t3h-00011

They were disassembled and the board mounted to each eye subassembly.

Alan

[gael langevin]

Very impressive work Alan!!!

I really like your color choice!

Let me know how those cameras are doing in the eyes, I maybe should add them in the BOM because it seems builders are now having a hard time to find the HerculesHD Twist.

Also your remarks are pertinent abot the 180 rotational wrist. Did you try the recent TensionerV1 along with the 180 degree wrist?

You should post your pictures in your InMoov album

I can't wait to see all assembled!

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

[Alberto]

How do you made that cleaned finished? Do you use sand, varnish..?

[Alan Timm]

Hey Alberto,

All of the white parts were printed in either white ABS or PLA with 100 micron layer heights

Then they were:

 - sanded either by hand or with an orbital using 220 grit sandpaper.

 - use duplicolor filling primer spraycan

 - more sanding

 - more filling primer

 - sanding then finish sanding with 400grit then polish a bit with plain paper towel

 - a few coats of duplicolor white enamel spraycan in a relatively dust free location

I didn't follow all those steps the first time I did the panels.  This time they turned out much better.

Alan

[Alan Timm]

Hey Gael,

Thank you so much!

Yep, the cameras work just fine.  Also they can both be plugged into the same usb hub.  (Depending on the bandwidth requirements sometimes only one camera can be active for each usb adapter, but it looks like either something changed or these cameras can both be used on the same adapter.)  Because of the way they are mounted, one of the images will be upside down, but that can easily be accounted for in software.

Because of the way the cameras are inset, you can see the inside of the lid in most eye orientations.  I suppose if I were updating the eye design, I'd try to move the eye cameras forward as much as possible to minimize it, although it doesn't seem so bad.

You know, I only recently noticed the tensioners, I'll take a look at them.

Thanks!

Alan

[Alan Timm]

Ok, now we're cookin' with gas.

Grey is fully assembled.

First, the good:

 - wiring was straightforward, with only one bad connection (loose crimp)

 - new neck assembly was a snap

 - eye mechanism fit into head assembly with no issues

 - great range of motion for the eyes now.

 - eye cameras both still work

 - simple to integrate the new servos into the existing microcontroller framework

 - jaw now works

Now the challenges:

 - the Bob's articulated neck design is really neat, but I'm getting too much play in the side to side assemblies and the joints tend to pop off.  I have a couple of ideas on how to address it.  I'll need to model the subassembly and play around with some things.

 - there is a slight interference between the eye subassembly and the jaw assembly in some positions.  need to verify where it's hitting, and push an updated eyesupport bracket design.

 - the jaw geometry doesn't seem to allow the jaw to close all the way, need to update one of the brackets to give a little more clearance.

Also in both head sets that I've printed, and both eye mechanisms, the eye mechanism mounts 1-ish mm to the bot's right.  I see the same offset when I model the subassembly.  My clearances between the eye and the faceplate are very tight, and the right eye may rub a bit.  I may try to update one of the brackets and push the update.

Alan

[nanomole]

looking very good alan - I did the same with my inmoov and painted - looks good when finished. just started prining your eye mech mods.

On Saturday, January 23, 2016 at 2:39:07 PM UTC-8, Alan Timm wrote:

[Bob H]

Hi Alan, Did you put a small screw thru the joints? This will keep them from twisting off.

[Alan Timm]

Hey Bob!

This articulating neck design is genius!  

It took me a while to understand how you modeled the piston adapter, but I think I finally get all of the features.

The first idea was to pin the adapter to the ball joint like you suggested, so here's my first attempt at adding a track into the adapter for a 3mm socket head bolt.

The front neck servo is already constrained to a single axis of rotation, so the adapter is the only place where there can be a twist.

The side to side servo assemblies can specifically twist in the direction of the screw because of the way servo pivot works.

Maybe that's why the original ball joint was so tight in the first place?

I'm going to take a crack at the servo pivot / servo holder combo to see if I can constrain movement to either one or two axis.

If it all works the the way I envision it, everything should be nice and tight and constrained to only what's necessary..

Alan

[Alan Timm]

Ok, I've finished my changes for the upper portions of the articulated neck.

The sample prints worked perfectly, so after a couple of minor tweaks the final prints are running on the printer with 100 micron layers .

I'll publish my updates to my thingiverse project  here once I get all the prints off the printer and they assemble without issue.

http://www.thingiverse.com/thing:1243403

I'm still working on the servo mount.  Still not sure if it needs one or two degrees of freedom (I knew I should have paid more attention in geometry class).

[nanomole]

might be time to warm the printer up again soon then.

hey alan - just finished your eye mech mod - works a treat and lots more freedom of movement - thank you!

[Alan Timm]

Hey nanomole,

Excellent, I'm so glad they worked out for you.

I can't take any credit for the design, that goes 100% to Gael and his EyeMechanism2:

http://inmoov.fr/eye-mechanism/

I just made a couple of changes based on what I needed for my InMoov.

That's the really cool part about releasing your designs as open source, it encourages people from the entire community to download, build, and make changes to the designs.  

I have learned so much from following Gael and building my own personal InMoov, and I look forward to applying everything I've learned to other projects in the future.  Thanks Gael!

Alan

[nanomole]

Yes Gaels design is awesome- thank you so much for sharing!

[gael langevin]

Woohooo!

Nice job here Alan!

I'm glad you are enjoying your builds!

I like the three joint articulated neck that Bob has done. I really should print as well for my next InMoov.

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

--

[Bob H]

Alan, your build and mods are looking great! Everyone sharing their modifications is what makes the InMoov project so great, of course none of this would be happening without Gaels genius and sharing. We are all looking forward to seeing your InMoov in action. I am working on another modification based on a ball joint - more on that later.

[RobotGuy6]

Hey Bob... I am glad you're back working on inMoov... Looking forward to see what you create next...

[Scott Hodges]

Hi Alan,

OK, so you helped already with a couple of things here, like fixing the off-axis eyes with your helper mix of Gael's V2 eye mech.   I'm also reading that I will probably need to use tensioners instead of the two tier pulleys I have now.  

Here's a couple of questions.  In the close-up of your arm servo-tray, it looks like you have fabricated a servo to dual cat-6 cabling set up, is that right?  Are you running cat-6 to another controller outside of the arm or just using it to consolidate wires within the arm?  What is the board you have in there?  A Nano with a servo shield?  An SSC-32?  

Your bot looks terrific and I will gladly give you credit for any ideas that I borrow.  :-) 

Thanks,

Scott

[Alan Timm]

Hey Scott,

Thanks!  I'm glad you're able to make use of some of my mods.

To run less cable, I feed each servo power from a separate wiring harness, and only pass the signal and position wires back up through to the forearm controller.

And you're absolutely right, I hacked up an old CAT-5 cable.

Each servo is hacked to provide position output as a 4th pin, which is fed to an a/d pin on the board.

Here's a quick shot of how I approached it.  power and ground are soldered to each side of a 2x4 male header, and signal and sensor are similarly soldered to their own 2x4 male header.

The whole thing is insulated with a hot glue gun, then wrapped when everything is assembled.

I opted to use the Teensy 3.x for my build, as it supports 24 total a/d converter pins, which I thought would come in handy.  They've been great so far.

https://www.pjrc.com/store/teensy32.html

I hesitate to share the carrier board, only because I didn't do the position feedback correctly, and although it works, it doesn't work very well.  It works very well on the bench, but once you get 27 of them together running off of a 6v lead acid battery, things start to get a bit weird.

https://oshpark.com/shared_projects/ERkzaZlk

I've been working with the RSSC robotics guys, and they've suggested I implement an RC (resistor capacitor) into each a/d input.  I'll also most likely need to feed in the position pot + lead as a 2nd a/d input in order to address all of the issues.

I'll get started on that after I finish my neck mods.

[gael langevin]

Hi Alan,

Are you running your InMoov on another software than MyRobotLab or Rohs?
If not, you might want to consider getting the Nervo boards which drastically simplifies the power and connections. If I'm correct,  you started your InMoov before we developped the Nervo board with Marten and Leon, so maybe it is the reason you went for a personal path.

Markus also has gone in a personal wiring path and did some great things. Otherwise the EZ system also has a simple way to connect each servo.

On your posted mods of the eye mechanismV2, I like how you biffed up some parts to make it even more sturdy and I noticed you modified the off centered eye. Just to let you know, it was done on purpose, I didnt like to see InMoov's pictures with cross eyed effect. Because we can adjust in the vision tracking of MyRobotLab the midle point of the camera, I decided to give a slight off set of the eyes which gives a more human look.

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

[Alan Timm]

Hey Gael,

That's a really good point, and I wouldn't recommend that anyone copy my microcontroller design at this time.

While there is potential to do some pretty neat things (closed loop position feedback + moving servo pid to the microcontroller) I haven't actually succeeded yet.

I don't know much about electronics, and so let's just say that my current carrier board design "needs some work".  :-)

New bots should definitely leverage your Nervo boards, and MyRobotLab, as it makes everything so much easier.

Thanks!

Alan

[Scott Hodges]

All good info.  Thank you very much.  I do a moderate amount of animatronics work and have put some similar arrangements together but I try to keep it really simple.  I usually run a separate power-ground pair and then just run the servo signal up the cat 5.  You can run 8 servos/signals/triggers on a single Cat-5/6 cable that way and the servo chatter is reduced because power is separate.    I hand-built a couple of PCBs to do it.  Sounds very similar to what you are doing.

Don't worry, though, I already ordered my Nervo stuff and am REALLY looking forward to using MRL so I don't have to write Python, OpenCV and C++ or robot-C stuff from scratch again.  

Thanks for the info!  

Scott    

[Alan Timm]

Quick update:

Grey left the house again for a quick talk and demo down at the Robotics Society of Southern California:

http://www.rssc.org/

https://groups.yahoo.com/neo/groups/RSSC-list/info

Trish was kind enough to post a quick video from the meeting here:

https://www.youtube.com/watch?v=TbSUXzdT4oY

Unfortunately I wasn't able to give a demo, as the pedestal adapter snapped while transporting him from the car.  

So the video shows him propped up against the back wall.  I also almost snapped off one of the arms when I caught him.

There are only 3 areas where I've run into problems and either stressed or broken Grey:

 - bicep, attach to shoulder (snapped 3 times)

 - forearm, where it attaches to bicep (snapped once)

 - shoulder, between shoulder rotate and bicep rotate (snapped once)

I've already posted a beefed up version of the forearm hinge, I'll be posting an updated bicep design, and I'll also be beefing up the shoulder assembly a bit.

Once I'm done with these mods, it will be alot easier to transport Grey, and he'll be alot more tolerant of occasional abuse.

Alan

[Alan Timm]

Here's an updated design for the bicep assembly.

http://www.thingiverse.com/thing:1347113

I've snapped the arms a couple of times already, and they always snap where RotMit mates to each ArmSide high.  

This version beefs up the bicep frame to 10mm, while including recesses for all original components.

It also allows you to anchor RotMit to the ArmSides using countersunk screws.

The only other part I needed to update was ServoHolder and Base.  

All of your other original parts should fit without issue, including the covers.

I'll publish a similar update for the shoulder assembly soon.

I should be able to put these on the printer later on this week.

Alan

[RobotGuy6]

Alan, those are really nice... Do you print in PLA (that could be your reason for your breaking stuff) as printing the arms that size would be a serious challenge using abs... Now I wish I bought an Up Box printer LOL Great abs printer... I am pretty good with ABS but even with my skills I don't think I could successfully print them in one piece using abs on my Wanhao Duplicator i3...  I can use Meshmixer to slice them in half but that means using acetone to fuse them back together which defeats the purpose here......

[Alan Timm]

Hey RobotGuy6,

Thanks!  The arms snapped for two reasons.

1.  I printed them with 3 perimeters, 1mm tops and bottoms, 50% infill, which isn't enough for the amount of stress applied to the puzzle pieces. All of my new prints are done with 6-8 perimeters, 2mm tops and bottoms, 25% infill

2.  The way the armhigh mates with rotmit, there is alot of force applied to a relatively small surface area, that's why I increased the width to 10mm, and included countersink holes.  That's as strong as I know how to make it, without totally redesigning the whole subassembly.

Most of the original bot was printed in ABS, it's been only very recently that I've been printing parts in PLA. 

With objects at these sizes, I see no reason to favor ABS over PLA, and my recent experiences seem to confirm this. This mod will be printed in PLA once my plastic arrives.

And I'll break out the ArmSide to two pieces shortly, and I'll include 100-200micron gap between the pieces so they're easier to assemble.

Thanks!

Alan

[RobotGuy6]

That's awesome Alan... I am just getting into using PLA as well. My hesitance is based on available adhesives... As you know acetone bonds abs basically stronger than the actual part... With PLA I am still looking for a really go adhesive for it... An alternative is PETG... which is PLA with much more strength... Still, finding a good adhesive is an issue...

[Alan Timm]

Hey RobotGuy6,

I have two recommendations for you.  One I've personally used, and one I keep hearing about and will be ordering shortly.

I've had great success with simple PVC cement from your local hardware store.

This stuff bonds very well to PLA, takes about 2 hours to set, but is very thick, so it ends up changing slightly the dimensions of your bonded parts.

http://www.amazon.com/Oatey-31017-Medium-Cement-4-Ounce/dp/B000CCIP08

I've also been hearing that the pros use this stuff.  It's an acrylic cement.  They also recommend the applicator.

I'm pretty excited to try this out. It's coming highly recommend from many makers, and seems to bond PLA really well.

http://www.amazon.com/SCIGRIP-10799-Acrylic-Solvent-Water-thin/dp/B00466V8F0

http://www.amazon.com/Gaunt-Industries-HYPO-25-Applicator-Weld/dp/B00EZRQOAY

Alan

[Alan Timm]

The beefier shoulder parts have been uploaded to thingiverse:

http://www.thingiverse.com/thing:1347747

Same changes as for the bicep.  Frame width increased to 10mm, and added countersunk holes to reinforce the joint.

[Alan Timm]

While I'm waiting for the last of the neck parts to print, I made a few small changes to my bicep remix. The test parts fit together very well, so I'll be putting these on the printer next.

This beefs up the bicep frames to 10mm, and anchors them to RotMit using screws.  I also tried to round off every edge I could find.  "Fillet All The Things!"

As soon as I get these off the printer over the next few days, I'll update the thingiverse project with the final files.

http://www.thingiverse.com/thing:1347113

[Robert Doerr]

Those new parts look awesome.  I really like the idea of using the screws to anchor into the RotMit part.  Since I haven't assembled that section yet I'll have to print out these new parts to try.

With the Extra Fillets on the edges does that change how it matches up with the covers?

[Robert Doerr]

With the thicker parts is the space between them the same?  If so then the shoulder cover may need to be updated/remixed to match:

http://www.thingiverse.com/thing:760302

[Scott Hodges]

Alan, this bicep looks awesome!  Couple of questions:

Where was your bicep breaking?  Above the forearm hinge?  

Do the bicep covers still fit on this or do they need to be altered also?  

For the reinforcer, is there a reason it does not fit into a slot on the outboard end?  I've always thought this was odd and that there should be a lower piece should have a slot there.  

Thanks,

Scott

[Alan Timm]

Hey Scott,

I've had breakages in only exactly 3 areas:

 - Where ArmHigh mates with RotTit in the bicep (several times)

 - Where the forearm cover mates with the bicep frame (once)

 - Where the shoulder assembly comes together (once)

Part of the problem is that I printed them with 3-4 perimeters, 1mm tops and bottoms which isn't strong enough.  

I mostly got away with it until I started moving him about from room to room, and when he started leaving the house with me to go on talks and demos at the RSSC.

All of my new parts are being printed with 6-8 perimeters, 2mm tops and bottoms, 15% infill.

But I also wanted to beef up these interfaces to make breakages even less likely.  By widening the frame pieces to 10mm, they're not only stronger, but they're thick enough now that I can add in anchor screws easily.

And you know, I kinda noticed the reinforcer thing, but didn't do anything with it.  I'll give it a mating slot in the frame for the next half that I print.  

Hey Robert,

I was careful to minimize the reprints necessary for the remix.  The new frame pieces have the same external dimensions, they just go inward for the extra width.  I created voids for all of the internal bits so they still fit without issue.  

The existing covers still fit snugly.  But I'll definitely recheck before I push the updated files.

The shoulder remix definitely changes some things around with regards to your nice shoulder cover.  I'll let you know as soon as the new models are verified.  I think there may be a way to implement a similar cover with the new shoulder panels.

Thanks!

Alan

[Peloche]

Hi Alan,

Congrats for your all your nice work.

Concerning Stronger Bicep Frame i'm wondering if anyone could print biceps parts ArmSideV1_outside_Grey and ArmSideV1_inside_Grey as they are in one piece only, Gaël parts were twice HighArmSideV2 + lowarmsideV1, have you got a large printer ?

Thanks a lot,

Serge,

[Alan Timm]

Hey Serge,

I'm actually planning on publishing multiple variations of Strong_ArmSideV2.

I'll push the complete part, a two part ArmSide, as well as versions for round and square potentiometers.

It's not hard to generate the variations, that way anyone could make use of the updated design

Alan

[Robert Doerr]

The shoulder cover wasn't done by me but it is a nice part.  If there can be a relief in your part so that the internal clips still fit then that part can stay the same.

I must say that your updated parts look awesome and should help a lot.  The original ones I printed for the shoulder/bicept used 4 shells and 30 percent infill.  I think I'll use the updated parts you made on my robot.  So the original parts won't be wasted I can still assemble those but use that portion or the arm assembly aw a demo to show off just what can be done with a 3D printer.

There is another RSSC member that I know (Thomas M) that is also building an InMoov. 

[Peloche]

Thanks Alan, 

Could ArmSideV1_outside_Grey be printed  on a Creator Pro ?

I made a test with your PivConnectorV1_Grey, it's my first part printed in blue.

I used the following settings:

[Alan Timm]

Hey Serge,

It looks great, but hang on a bit, I haven't run test prints for that remix yet, and I'm planning on rounding up some of the edges to make it more organic.

Everything I'm doing can be printed on your creator, and I'll post the arm part as two pieces as well.

It'll take me a day or two to run the rest of the bicep parts, then I'll get to the shoulder.

Alan

[Peloche]

Thanks Alan, take your time ;-))

Have a nice evening,

Serge,

[RobotGuy6]

Hey Alan... just a note... I did print the PivConnectorV1_Grey and rightPivTitV1_grey... the PivConnectorV1_Grey is now too wide to fit properly. The cutout for the PivWorm interferes with centering PivConnectorV1_Grey onto the  rightPivTitV1_grey anchor lugs... Hope that made sense...

[Alan Timm]

Hey RobotGuy6,

Sorry about that, I didn't check for interference before I posted the files.  I didn't expect any interest in the shoulder remix so I rushed it.

I'm trying not to recreate RotMit and RotTit, so as a result there is a very slight alignment issue with this assembly.  Also the RotMit back mount is about 1mm off of center, and I can't fix it without recreating the entire part.  So I just worked around it now.  You can see it in the render below.

I have another day or so of bicep parts, then I'll switch gears and verify the shoulder parts.

http://www.thingiverse.com/thing:1347747

[RobotGuy6]

No need to be sorry man, it's great work... Cheers

[Alan Timm]

I made some changes to the fillet, and test printed the files.  They look and fit great except for the 1mm offset for the RotMit mount, which you can see in the render below.  Once they're printed, the offset is not as dramatic.

I may come back and recreate the part later, but for now it works.

I also smoothed out the fillet.  It prints well and is more uniform.

Parts have been uploaded to the remix here:

[Robert Doerr]

These look great!  So are these new arm parts ready to print or are you still testing/tweaking a few things?  Looking forward to printing these....

Robert

[Alan Timm]

Hey Robert,

Both the Strong Shoulder remix and the Articulated Neck remix are complete, I've printed out all the parts and confirmed that they're good.

Articulated Neck Remix:  http://www.thingiverse.com/thing:1243403

Strong Shoulder Remix:  http://www.thingiverse.com/thing:1347747

Tonight or tomorrow I'll add the rest of the instructions for the neck mod, and post a youtube clip of it in action.

I'm still working on the bicep parts.  I'm going to even out the fillet features a bit, and get them on the printer over the next day or so.

Thanks!

Alan

[Peloche]

Hi Alan,

Very nice, many thanks, will print them tomorrow ;-))

Serge,

[Robert Doerr]

I'm still using the original neck design (may decide to switch later) but will get going on the shoulder pieces.  I'll wait on the Bicept pieces until they are ready.  All these look great.

On Wednesday, February 24, 2016 at 11:31:50 AM UTC-5, Alan Timm wrote:

Hey Robert,

Both the Strong Shoulder remix and the Articulated Neck remix are complete, I've printed out all the parts and co.  nfirmed that they're good.

[revwarguy7]

You are doing some very nice work, Alan, and I just wanted to say thank you so much for sharing it!

[Peloche]

Hi Alan,

I've printed the new "Strong_PivConnectorV2", it's nice, what do you think about the left downside corner (red circle on the picture), is it large enough, the thinest part is 2 mm width ?

Thanks,

Serge,

[Alan Timm]

Hey Peloche,

Yeah, that's the skinniest of the features, it's 4mm at the center, and tapers to about 2mm on the outsides.  The original was 4mm side.  It's probably a wash between how strong that particular feature is now vs how it was in the original design.

As these are anchored now, it's less of a concern.  If I ever get around to recreating PivTit, I'd move that tab in a few more mm.

Alan

[Peloche]

Hi Alan,

Thanks, could they be glued (acetone) with PivMitV1 ?

Serge,

[Peloche]

PivMitV1 is not the good side, it's RotTitV2 isn't it ?

[Alan Timm]

While I guess you could acetone weld the pieces together if you're printing in ABS, I'd recommend using a handful of #8 flathead screws.  

Also, I just pushed a small update for PivMit, Now it should be possible to use all the mounting screws to affix PivTit to the gearbox.  I added a small recess that should allow you to slide PivMit into the shoulder subassembly then secure with screws.  

http://www.thingiverse.com/thing:1347747

Thanks!

Alan

On Thursday, February 25, 2016 at 7:51:40 AM UTC-8, Peloche wrote:

Hi Alan,

[Peloche]

Thanks Alan, RightPivMit is printed at 81% ... i'll see for left shoulder with your last update.

You are doing a very good job Alan, it's time-consuming, thanks.

When i see all the design done by Gaël it's incredible what he has done, so many parts and so accurate, he is really The Creator, hats off to Gaël !

Serge,

[Dwayne Williams]

Than you for the shoulder edits Alan,

They have needed that final tweek for quite sometime.

[nanomole]

Hey Alan - i must say i am impressed with all these updates......great work and thank you for sharing !

[Alan Timm]

Ok, so turns out you can't daisy chain the HiTec HS645MG or the HS805MG.

I used a pair of HiTec HS-645-MG for the neck side to side, and they've let loose their magic smoke.

Same with the Pair of HS-805-MG.  They actually blew a few nights ago.

For the Neck, I'm going to move it to a single cheap HobbyKing HK-15328D, and gut the other one just for show.  The horns don't fit though, so I'm gonna have to design and print something for it.

For the Waist side-to-side, I'll use my remaining HobbyKing HK-15338 pair to replace the Hitecs.  I think they'll daisy chain just fine, because it's what I used for the pair of waist rotate, and they haven't blown (yet).

Alan

On Saturday, January 23, 2016 at 2:39:07 PM UTC-8, Alan Timm wrote:

It's that time of year again, and Gray's getting stripped down in preparation for some upgrades.

Here's what I'm doing to him next, with urls for the stuff I can link to.

[Bob H]

That's strange that you've had problems with daisy chaining the servos. Are you using a "gutted" servo (only the motor left in it) with the rotation stop removed for the slave servo? I've got 3 combinations like this. The only time I've had a problem is if the servos can't move to the position specified in a script. That's usually caused but a joint being too tight or binding.

[RobotGuy6]

 I did the same as Bob H and also have absolutely no issues chaining the servos together. Be aware as Bob mentioned you have to remove the board a pot from one of the servos... All that should be left is the motor and gearbox.. The first servo controls the second servo as well with it's control board.... 

[Alan Timm]

Yeah, I'm kinda confused by it as well.  Lost both the waist lean and artneck side sets almost at the same time.

If you guys have done it successfully with the HiTec servos without issue, then I must have done something wrong, but not sure what it could have been.

The second servo was gutted and prepped for 360 rotation, just two wires between the motors.  No pot, and no electronics in the slave servo.

Nothing was tight, and I could manually rotate the neck without issue.

Both sets oscillated, overheated then burned out in a very short period of time.

I have these HobbyKing HK15328Ds lying around, so I'll install them, retrace all my steps, and see how it works out for the Articulating Neck.

If they work well, they'd be a nice $8 alternative to the Hitecs.

Alan

[Robert Doerr]

What are you powering them with?

[RobotGuy6]

Alan, all you are doing is driving 2 servos with 1 servo control board... So you remove the pot (for example waist rotation) in one servo... The second servo you remove the pot and control board, everything except the motor and gear box... Now here is the easy part. You solder wires to the motor contacts of the first servo (the one with the control board still in). You then solder the other ends of those 2 wires (paying attention to the positive an negative contacts) wires to the motor contacts of the second servo... Like an umbilical cord so the speak..  Now what you have is one servo board driving 2 servos motors... If you think about it it's like a diver sharing his oxygen tank with another dive using 2 mouth pieces... That's all there is to it...

On Friday, February 26, 2016 at 11:29:29 AM UTC-5, Alan Timm wrote:

[Alan Timm]

Yep, that's exactly what I did.  Of the 3 sets of daisy chained servos, two of them blew, while the third (waist rotate) did not.

The HiTecs have nice fat 3pin power mosfets on the board, so I wasn't worried about about blowing them.  Honestly I thought the HobbyKing pair would have been the ones to blow from double the drive output with two motors.

Not sure what I did wrong, but it must have been something since you guys are having good luck with them.

I most likely need to add some filtering caps into the circuit anyways between the servo supply and also the microcontroller supply.  I wonder if that contributed to the meltdown.

That controller handles all of the head and torso, and is powered by a parallel wired pair of 6v lead acid cells.

I was able to get the HobbyKing horns designed and printed this morning, so with any luck I'll have the neck reassembled and a youtube clip posted tonight.

Alan

[Robert Doerr]

Hello Alan,

I'm getting ready to print out these upgraded parts.  It sounds like the shoulder pieces are good to go.  Have the Bicept pieces been tested yet?

Robert

[Alan Timm]

Hey Robert,.

The shoulder pieces are good to go, I've already printed out the entire set and they fit together very well.

Using #8 screws is best for the entire shoulder assembly.  Maybe #6 is ok.

Hang tight on the bicep remix, it's not 100% printed and verified.  I missed two recesses necessary for the cover panels, and I'll be uploading the final parts later this week.

The good news is that everything else in the bicep remix seems to be fitting very well.  

Thanks!

Alan

[Peloche]

Hi Alan,

Good job, i've printed your shoulder parts and it's ok, biceps should be of the same quality ;-))

[Alan Timm]

Everything looks good for the bicep remix, so I'll release it now.

The only thing I haven't checked 100% is the recesses that I've added for the front cover.  They look right, and I'll get them on the printer tomorrow.

Everything you see below has been uploaded to the thingiverse remix here:

http://www.thingiverse.com/thing:134711

This derivative beefs up the bicep frame by adding the following features:

• Frame pieces are now 10mm wide, recesses to work with all existing components not replaced here
  
• Frame pieces now can be anchored into RotMit using screws (recommend #8 or maybe #6)
  
• Recesses added to thicker frame to allow all existing pieces to fit without modification, including covers
  
• Servo holder and horn reshaped to fit
  
• Servo wiring now routed down and out the back
  
• Reinforcer lengthened and hole added for second tab
• Fillets everywhere!  rounded out every edge and corner I could find.

I was careful to make sure that all remaining bicep parts should still fit, with the exception of the parts posted here. This replaces the equivalent parts from the original bicep design.

Over the next few days I'll upload a version for the original round potentiometers, and ArmSide broken down into two pieces for smaller printers.

Alan

[Peloche]

Thanks Alan, uploaded and i'll start to print them ;-))

Very good job ;-))

Serge,

[gael langevin]

Very good job Alan!!

I see you managed also a cavitie for the servo horn to properly move up and down without hitting the sides of the bicep.

Are the side reinforcer still fitting a 12x12 footprint printer? They wouldn't be long enough  on my orginal design only for that reason...

Gael Langevin

Creator of InMoov

InMoov Robot

@inmoov

--
You received this message because you are subscribed to the Google Groups "InMoov" group.
To unsubscribe from this group and stop receiving emails from it, send an email to inmoov+un...@googlegroups.com.
To post to this group, send email to inm...@googlegroups.com.
Visit this group at https://groups.google.com/group/inmoov.
For more options, visit https://groups.google.com/d/optout.

[Robert Doerr]

Really nice!  Warming up the printer now.....

[Alan Timm]

Hey Gael,

Thanks!  I just checked the reinforcer remix, it's 137mm long, and the entire part fits diagonally into a 120x120mm print area.  

If anyone needs it, I could always remove the far tab and shorten it down a bit more.

Alan

[Alan Timm]

The Neck Remix is done, and it's looking really good.  I was holding off on publicizing the mod because I was having problems with the servos.

So it turns out that it was 100% operator error, and everything's ready to go.

Turns out I had two separate issues.  

1.  I hadn't doublechecked my max and min limits, keeping in mind that there's 3-5mm of play in the pistons because of the clearances between piston and piston base

2.  The crossover wiring between the servo and the slave servo had a loose crimp, which could have also cause the burnout.

So I'm sure that HiTec servos work just fine.

These HobbyKing HK15328D digital servos are working out really well also, and they're only $8 each.

https://www.youtube.com/watch?v=FPV6k8QfDY8

[Bob H]

Wow! Nice work, boy that is smooth movement. Glad to hear you got your servo issues worked out.

[Peloche]

Hi Alan,

I've started to print out right bicep using your Strongs parts, it's very nice.

The main problem is that when printing out Strong_ArmSideV2+square, it is using all the bed surface of my Creator Pro (the width of the bed is 232 mm and Strong_ArmSideV2+square width is around 220 mm), that implies the bed to be very well levelled, when using small surface around the bed center it's almost good each time, but with your nice big part it was a "disaster"…;-))

I made at least ten tries before to get something admissible, the main reason IMHO is that the bed when heated at 110°C is cambered with a top point in the center and lower points on the sides and the differences in between center and sides are enough to mess the print out beginning.

For the very large parts as yours i start with a small one and once it's hot i go ahead with yours, but this is to be confirmed daily …

Anyway thanks again for your Strongs,

Serge,

Le mardi 1 mars 2016 06:39:00 UTC+1, Alan Timm a écrit :

[Alan Timm]

Hey Peloche,

That part looks great.  I'll publish the armsidehigh and armsidelow variants tonight.  They'll fit in the same 120mm area that the original parts fit into.

Would you be interested in running those?

Alan

[Robert Doerr]

When you leveled the bed did you level it after preheating it first?  That can make a difference.  Also if it looks like the bed itself isn't perfectly flat you can try printing on glass instead.  It just sits on your existing bed.  Just need to print a spacer for the rear limit switch the thickness of the glass and print corner clips.  That is how I print ABS parts.  For PLA I still use Blue tape on the bed.

[Alan Timm]

Ok, I've updated the Bicep remix, and most importantly I added in ArmHigh and ArmLow parts for normal sized printers.

http://www.thingiverse.com/thing:1347113

I did also need to make a small change to armside for how the frontcover fits, it's a bit loose in the V2 version.

Also, I slightly moved the side cover mount holes to reduce the tension just a bit.  They still fit tightly, but not so tight as to risk snapping the covers.

I also pushed a few small updates, mostly minor, although the new servoholder now routes all the wiring out the back instead of down through servobase.

Changes:

2016-03-03:

ArmHighV3 & ArmLowV3 added armhigh and armlow variants for normal sized printers;

ArmSideV3 made a few minor changes to armside, changed front cover attachment point a bit

ReinforcerV3 shaved a bit off to reduce interference with rotmit

ServoHolderV3 cable routing now goes out the back of the servoholder; added opening to bottom so you don't have to resolder the pot every time you need to take off the servo.

ServoBaseV3 cable routing now goes out the back of servoholder

Please let me know if you have any problems with the parts, or need the rount potentiometer variation.

Thanks!

Alan

[Peloche]

Hi Robert,

Yeap always, you're right this can make a difference since the adjustment is about 1/10 mm, for the time being it is well set but i do prefer to print out small parts ;-))

Thanks for your advices,

Serge,

[Peloche]

Hi Alan,

Thanks a lot for your mods, it's very helpful ;-)).

I'll try to keep StrongArmSideV3+square that i've already printed out and will go with StrongArmSideHighV3-square and StrongArmSideLowV3-square to complete the right bicep, for the left one i'll use High and Low parts for both "-" and "+".

Concerning ServoHolder the opening is cleaver but for ServoBase i don't see any difference with the former V3, am i right ?

Best regards,

Serge,

[Robert Doerr]

Hello Alan,

These new updated parts that you've posted are printing very well!  It's nice that you made a singe larger side piece as well as the smaller 2-piece version.  I am printing the large ones that just fit on the bed of the printer.  Have a few more parts to print and then I can move onto the updated neck design.  Excellent work.

Robert

[Alan Timm]

Excellent! 

I'm so glad they're working out for you.

As an fyi, I'll be pushing a final armside set either tonight or tomorrow.  I'm making a small tweak just to the recess for the top cover mount.  I think I can get it to fit even better.

If you wanted to hang tight for another day until I publish the V4 armsides you'll print the final version.

Thanks!

Alan

[Peloche]

Hi Alan,

I've got some neck problems so i'll print your ArtNeck ... jewelry parts, are you a Jeweller ? Your work looks like, what a team ! Gaël, Bob Houston and yourself you're the Dream Team !

Thanks a lot for sharing Alan ;-))

Serge,