Project: Melissa Hands - Perhaps a little easier..

Discussions regarding building a walking robot at home. Most of the robots participating at Robo-One competitions are custom fabricated.
135 postsPage 3 of 91, 2, 3, 4, 5, 6 ... 9
135 postsPage 3 of 91, 2, 3, 4, 5, 6 ... 9

Post by PaulL » Sun Nov 20, 2011 5:56 am

Post by PaulL
Sun Nov 20, 2011 5:56 am

After a good bit of fiddling with CAD and a dial caliper earlier today, I'm about to go cut this on the mill and see how it fits. FINALLY! :)

Stay tuned, pics of actual metal to follow.. ;)

Image
After a good bit of fiddling with CAD and a dial caliper earlier today, I'm about to go cut this on the mill and see how it fits. FINALLY! :)

Stay tuned, pics of actual metal to follow.. ;)

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Sun Nov 20, 2011 10:03 am

Post by PaulL
Sun Nov 20, 2011 10:03 am

... and ...

Image

... and ...

Image

... and for scale ...

Image
... and ...

Image

... and ...

Image

... and for scale ...

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Sun Nov 20, 2011 11:02 am

Post by PaulL
Sun Nov 20, 2011 11:02 am

Has anyone been wondering how those funky brackets would go together? Wonder no more.. Some pics after bending the brackets:

Image

And...

Image

And...

Image

And...

Image
Has anyone been wondering how those funky brackets would go together? Wonder no more.. Some pics after bending the brackets:

Image

And...

Image

And...

Image

And...

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Sun Nov 20, 2011 1:07 pm

Post by PaulL
Sun Nov 20, 2011 1:07 pm

And, putting it all together, servos floating around inside, fingers attached, just to see it as it will be...

Image

Image

Image

Image
And, putting it all together, servos floating around inside, fingers attached, just to see it as it will be...

Image

Image

Image

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by billyzelsnack » Mon Nov 21, 2011 3:39 am

Post by billyzelsnack
Mon Nov 21, 2011 3:39 am

This is gorgeous.
This is gorgeous.
billyzelsnack offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 618
Joined: Sat Dec 30, 2006 1:00 am

Post by mog123 » Mon Nov 21, 2011 11:34 am

Post by mog123
Mon Nov 21, 2011 11:34 am

those are just fantastic
those are just fantastic
mog123 offline
Savvy Roboteer
Savvy Roboteer
Posts: 50
Joined: Sat Jan 30, 2010 2:19 pm

Post by Novice » Mon Nov 21, 2011 3:56 pm

Post by Novice
Mon Nov 21, 2011 3:56 pm

Hey Paul
This is a good work :D !! I'm looking for the similar things, who can produce something like this for my KHR-3 but a little beat cheeper against the original God's hands from Melissa course the cost of the last one is out of resonable sence to me :(
Did U think, how much may it cost already?What kind of servo did U take for? I have seen the photo.Is the servo's name HS-35D like here http://www.robotshop.com/world/hitec-hs ... motor.html or something else?
Finally, lets do not take offen :D but I think the fingers can not catch&hold something, even a pen, course the big finger is located a little beat wrong.Have to place it closer to another fingers or to change the angle between the big finger and supporting bracket.Any case your work is great but its your pilot project, hope U can fix all the problems and I wish U to win finally :D !
Take care
Oleg
Hey Paul
This is a good work :D !! I'm looking for the similar things, who can produce something like this for my KHR-3 but a little beat cheeper against the original God's hands from Melissa course the cost of the last one is out of resonable sence to me :(
Did U think, how much may it cost already?What kind of servo did U take for? I have seen the photo.Is the servo's name HS-35D like here http://www.robotshop.com/world/hitec-hs ... motor.html or something else?
Finally, lets do not take offen :D but I think the fingers can not catch&hold something, even a pen, course the big finger is located a little beat wrong.Have to place it closer to another fingers or to change the angle between the big finger and supporting bracket.Any case your work is great but its your pilot project, hope U can fix all the problems and I wish U to win finally :D !
Take care
Oleg
Novice offline
Savvy Roboteer
Savvy Roboteer
Posts: 26
Joined: Wed Dec 30, 2009 1:21 pm

Post by PaulL » Tue Nov 22, 2011 12:29 pm

Post by PaulL
Tue Nov 22, 2011 12:29 pm

Thank You All for the kind words! :)

I'm happy with the results as a first run, but there are a few small tweaks, the main one is the drilled holes, they need a starter bit first. One bit in particular walked way too much and I didn't think it would turn out so well. Also, the v-cuts were MUCH deeper for the bends than I had intended (.005 inches is what I wanted, but didn't zero my tool right), but it made the bends nice and square. :)

This has been a very slow moving project, as I've spent most of my spare time in software. :)

The goal here isn't as much a "grasping" hand as it is an "expressive" hand for communication. People talk not only with voice (which is something else in my plans), but often communicate with their hands as well.

My goals for this project:

* Small Hand in similar size and shape to stock Robonova 1 hands when clenched, but matching look and feel of the Robonova 1.
* Aluminum bracketry with a hint of black plastic like the Robonova 1.
* Full finger articulation for all kinds of subtle and not-so-subtle finger motions.

Regarding Melissa's God Hands: Yes, they're way too expensive. And with fingers made of u-channeled aluminum, not easy to make. I went with Igus E-Chain, as it fits my application well and reduces the work of designing and building fingers by hand. The insert blocks (all same size / shape) are the only modifications for the e-chain.

These will cost me approximately $400 USD in materials when done for one pair of hands, but this is just for servos (most of the cost) / materials in the finished pieces. I have my own CNC machine, with a few thousand USD invested in that, along with "disposable" material costs for trial and error, etc.

It could hold a thick pen in the 4 fingers, but that isn't really why I'm building them. :) The thumb is more aesthetic and not so much to be functional.

The servos in these hands are MKS DS450 servos, high torque (3.1 kg/cm), metal gears, very small, very narrow for the torque they put out. I looked at many servos on the 'net, these were the best choice in the end. The other issue is travel, and too small means less travel for tendons. These servos are "just right" for a few reasons. Just realized, 15.5 kg/cm per hand... Wow.

I will be using Pololu servo controllers in these, and I have both USB and serial versions to test (it's the smallest size of each version). This will mean I have 3 wires to the hands, 1 data and two power (the Pololu boards don't need to talk back to the controller, a Roboard RB-100 in this case).

It's my intention to build two working sets for two equally equipped bots, RN-1's, 1 in Red and 1 in Blue. The hands will either be Anodized or Painted, and I may end up doing both before it's over (painting now, anodizing other sets later).

I debated over fingers a good bit before I started. If I built fingers by hand, it would certainly not be easy to make. And, going with E-chain guarantees smooth and stable motion (no play in joints), as well as low operational friction allowing me to use .020 inch teflon for tendons instead of nylon - less torque needed in the servos to make it work. And more, the e-chain fingers are light weight. I'm already adding a lot of weight with the servos, so better to save a little weight. On top of all this, the fingers have a "thick" look I haven't seen in alu Melissa Hands - the ones I've seen look thinner, more slender - I wanted fatter fingers. :) On top of all of this, the bends of this size e-chain very closely matches the profile of the stock Robonova 1 hands when clenched.

Drawbacks of e-chain? It's not metal - but it's tough, designed for innumerable cycles of use in their intended application, protection of wiring in a moving system. That said, they're not extremely strong. I think they're plenty strong for my application, but I haven't had to catch the force of a falling Robonova 1 with them yet. If I do snap one, replacing links of chain is easy - even the insert blocks can be removed, they just rest inside the chain, captured with their top channel cut.

When this is all together, will he be able to hang from his fingers? I don't think so! :) But, if his fingers get jammed, they'll either snap at the links, or worst, tear the teflon tendon - both easy and cheap to replace. The teflon is thin enough that it will probably tear at one of the mounting holes before the servo gears break. I do think he will be able to do cartwheels, as the strength then is in the e-chain when bent, not the teflon tendon.

I have a few more prototypes to make to get dimensions and manufacturability straight, but there's not really that much to do on the design at this point.

Take Care,
Paul
Thank You All for the kind words! :)

I'm happy with the results as a first run, but there are a few small tweaks, the main one is the drilled holes, they need a starter bit first. One bit in particular walked way too much and I didn't think it would turn out so well. Also, the v-cuts were MUCH deeper for the bends than I had intended (.005 inches is what I wanted, but didn't zero my tool right), but it made the bends nice and square. :)

This has been a very slow moving project, as I've spent most of my spare time in software. :)

The goal here isn't as much a "grasping" hand as it is an "expressive" hand for communication. People talk not only with voice (which is something else in my plans), but often communicate with their hands as well.

My goals for this project:

* Small Hand in similar size and shape to stock Robonova 1 hands when clenched, but matching look and feel of the Robonova 1.
* Aluminum bracketry with a hint of black plastic like the Robonova 1.
* Full finger articulation for all kinds of subtle and not-so-subtle finger motions.

Regarding Melissa's God Hands: Yes, they're way too expensive. And with fingers made of u-channeled aluminum, not easy to make. I went with Igus E-Chain, as it fits my application well and reduces the work of designing and building fingers by hand. The insert blocks (all same size / shape) are the only modifications for the e-chain.

These will cost me approximately $400 USD in materials when done for one pair of hands, but this is just for servos (most of the cost) / materials in the finished pieces. I have my own CNC machine, with a few thousand USD invested in that, along with "disposable" material costs for trial and error, etc.

It could hold a thick pen in the 4 fingers, but that isn't really why I'm building them. :) The thumb is more aesthetic and not so much to be functional.

The servos in these hands are MKS DS450 servos, high torque (3.1 kg/cm), metal gears, very small, very narrow for the torque they put out. I looked at many servos on the 'net, these were the best choice in the end. The other issue is travel, and too small means less travel for tendons. These servos are "just right" for a few reasons. Just realized, 15.5 kg/cm per hand... Wow.

I will be using Pololu servo controllers in these, and I have both USB and serial versions to test (it's the smallest size of each version). This will mean I have 3 wires to the hands, 1 data and two power (the Pololu boards don't need to talk back to the controller, a Roboard RB-100 in this case).

It's my intention to build two working sets for two equally equipped bots, RN-1's, 1 in Red and 1 in Blue. The hands will either be Anodized or Painted, and I may end up doing both before it's over (painting now, anodizing other sets later).

I debated over fingers a good bit before I started. If I built fingers by hand, it would certainly not be easy to make. And, going with E-chain guarantees smooth and stable motion (no play in joints), as well as low operational friction allowing me to use .020 inch teflon for tendons instead of nylon - less torque needed in the servos to make it work. And more, the e-chain fingers are light weight. I'm already adding a lot of weight with the servos, so better to save a little weight. On top of all this, the fingers have a "thick" look I haven't seen in alu Melissa Hands - the ones I've seen look thinner, more slender - I wanted fatter fingers. :) On top of all of this, the bends of this size e-chain very closely matches the profile of the stock Robonova 1 hands when clenched.

Drawbacks of e-chain? It's not metal - but it's tough, designed for innumerable cycles of use in their intended application, protection of wiring in a moving system. That said, they're not extremely strong. I think they're plenty strong for my application, but I haven't had to catch the force of a falling Robonova 1 with them yet. If I do snap one, replacing links of chain is easy - even the insert blocks can be removed, they just rest inside the chain, captured with their top channel cut.

When this is all together, will he be able to hang from his fingers? I don't think so! :) But, if his fingers get jammed, they'll either snap at the links, or worst, tear the teflon tendon - both easy and cheap to replace. The teflon is thin enough that it will probably tear at one of the mounting holes before the servo gears break. I do think he will be able to do cartwheels, as the strength then is in the e-chain when bent, not the teflon tendon.

I have a few more prototypes to make to get dimensions and manufacturability straight, but there's not really that much to do on the design at this point.

Take Care,
Paul
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Sat Nov 26, 2011 11:41 pm

Post by PaulL
Sat Nov 26, 2011 11:41 pm

An update:

I did quite a few modifications to the CAD drawing, and re-ran one set of brackets the other day. It didn't turn out too well! The main problem I had was that with a very light marking cut at the bends, manual bending caused splitting of the aluminum in the corners, worse than when I did a .5mm deep v-cut in the bend. Basically, the aluminum in all of the corners is cracked and ready to give. I'll use the exercise to tweak some measurements, but 6061-T6 does NOT like sharp 90 degree bends.

In my frustration, I ordered a Grizzly Pan and Box brake (G0556) for bending, along with 10 sheets of 5052-H32 aluminum (which is supposed to bend better, but not machine as well).

We'll see how it goes. Now I get to wait for stuff to get here (already sourced some Robonova screws since I have nothing better / more interesting to do).

I was pretty frustrated, the brackets looked great before I bent them, and after bending, the bends are split and just will not work. I thought about the design, wondering if there was a way to do this without bending, but there really is no better option.

I've read I could anneal the 6061, but at that point, using 6061 makes no sense, even if I could find success with that. Better to go for 5052 (which Tyberius uses in Giger), seems others have had good luck with it as well.

So for now, manufacturing the hands has to wait, pending 5052 aluminum and the pan and box brake delivery / setup / testing (to see what the radius is like and how it changes the CAD drawing). The uknown at this point is what kind of radius I can get out of the brake and .040 inch thick 5052. A press brake won't work for me, and it has to be a bending brake (with a lever instead of a ram), and a pan and box brake is required based on the design. Funny, I wanted a bending brake a few years ago, should've bought one then. ;)

Take Care,
Paul
An update:

I did quite a few modifications to the CAD drawing, and re-ran one set of brackets the other day. It didn't turn out too well! The main problem I had was that with a very light marking cut at the bends, manual bending caused splitting of the aluminum in the corners, worse than when I did a .5mm deep v-cut in the bend. Basically, the aluminum in all of the corners is cracked and ready to give. I'll use the exercise to tweak some measurements, but 6061-T6 does NOT like sharp 90 degree bends.

In my frustration, I ordered a Grizzly Pan and Box brake (G0556) for bending, along with 10 sheets of 5052-H32 aluminum (which is supposed to bend better, but not machine as well).

We'll see how it goes. Now I get to wait for stuff to get here (already sourced some Robonova screws since I have nothing better / more interesting to do).

I was pretty frustrated, the brackets looked great before I bent them, and after bending, the bends are split and just will not work. I thought about the design, wondering if there was a way to do this without bending, but there really is no better option.

I've read I could anneal the 6061, but at that point, using 6061 makes no sense, even if I could find success with that. Better to go for 5052 (which Tyberius uses in Giger), seems others have had good luck with it as well.

So for now, manufacturing the hands has to wait, pending 5052 aluminum and the pan and box brake delivery / setup / testing (to see what the radius is like and how it changes the CAD drawing). The uknown at this point is what kind of radius I can get out of the brake and .040 inch thick 5052. A press brake won't work for me, and it has to be a bending brake (with a lever instead of a ram), and a pan and box brake is required based on the design. Funny, I wanted a bending brake a few years ago, should've bought one then. ;)

Take Care,
Paul
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Fri Dec 02, 2011 12:07 pm

Post by PaulL
Fri Dec 02, 2011 12:07 pm

I got the bending brake yesterday. For anyone that is interested, the fingers on the bending brake don't have a radius at all - they're very sharp at the bending edge.

This brake has plenty of adjustment capability. I still need to tear it down and clean it off (gets packed with a protective grease that is more like glue), but I did test bend a piece of 6061 - and it DID NOT CRACK with a tight radius bend. I was amazed. I need to try some more bends in 6061 to see if any problems crop up, but the one bend I've done so far looks great. All the same, I did buy 5052, and it showed up, so I'm going to try that, too.

I am unclear on the mechanics of why my manual bends cracked, but I'd guess the scoring (no matter what the depth of the v-groove) provided a point of failure for the bend.

All the same, the result of an un-scored bend on this brake is very nice, much better than I could do manually. I'm glad I bought it, another piece of valued equipment. :)

Next, I need to recalculate the measurements applied when bending to the CAD drawing, run the .DWG through Vectric, and cut out some 5052 and see how it goes.
I got the bending brake yesterday. For anyone that is interested, the fingers on the bending brake don't have a radius at all - they're very sharp at the bending edge.

This brake has plenty of adjustment capability. I still need to tear it down and clean it off (gets packed with a protective grease that is more like glue), but I did test bend a piece of 6061 - and it DID NOT CRACK with a tight radius bend. I was amazed. I need to try some more bends in 6061 to see if any problems crop up, but the one bend I've done so far looks great. All the same, I did buy 5052, and it showed up, so I'm going to try that, too.

I am unclear on the mechanics of why my manual bends cracked, but I'd guess the scoring (no matter what the depth of the v-groove) provided a point of failure for the bend.

All the same, the result of an un-scored bend on this brake is very nice, much better than I could do manually. I'm glad I bought it, another piece of valued equipment. :)

Next, I need to recalculate the measurements applied when bending to the CAD drawing, run the .DWG through Vectric, and cut out some 5052 and see how it goes.
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by Joe » Sat Dec 03, 2011 3:45 am

Post by Joe
Sat Dec 03, 2011 3:45 am

What make/model brake did you get? Sounds like you're happy with it, and it's a tool I've been thinking about investing in myself.

Thanks,
- Joe
What make/model brake did you get? Sounds like you're happy with it, and it's a tool I've been thinking about investing in myself.

Thanks,
- Joe
Joe offline
Savvy Roboteer
Savvy Roboteer
User avatar
Posts: 204
Joined: Tue Sep 05, 2006 1:00 am

Post by PaulL » Wed Dec 21, 2011 12:23 pm

Post by PaulL
Wed Dec 21, 2011 12:23 pm

It's a G0556 from www.grizzlyindustrial.com. :) I should mention, if you haven't bought any of the chinese import machines (Mill, Lathe, etc), the quality may let you down. After having taken the machine apart, I see the hidden flaws - gouged mounting plates for the fingers, fingers themselves have "splinters", some parts misaligned aeshtetically - basically it's a typical chinese import machine: a rough piece of equipment that does what it should, but isn't going to be great in all respects.

I tore down, cleaned, and reassembled the bending brake, CNC'd a dozen test pieces of 5052, and here are the results of bending:

* First two pieces were to test 5052 bending, no measurements made, bent fine.

* Set up a perpendicular guide, set up the machine for thickness and bend offset, bent some long scrap strips of .038 inch thick 6061, verified perpendicular bends. Results look great.

* 5052 is .040 inches thick, not .038 like the 6061 (yes, splitting hairs here).

* Measured all 10 remaining CNC'd test pieces, range from 0.999 to 1.001 (target size, 1.0 inches). Good repeatability by the CNC machine.

* Bent a few pieces, dialing in the location of the bend with a known offset. Distance seems to be around 2.010 inches from the front flange to the intended center of the bend.

* Bent remaining pieces at an offset of 1.510 inches, resulting in a middle bend of the 1 inch length test piece.

RESULTS:

I encountered NO splitting of alu using 5052 OR 6061 on what is a fairly tight radius.

But, the more interesting part is the numbers applying to the bend itself: as with manual bending, the resulting average length of each side of a bent piece of 1 inch aluminum is .5333 (.533) inches over 10 pieces. I WAS AMAZED!!! Virtually identical change in distance, using two different alloys, same width / length pieces, manual versus machine bending. The difference? No cracks as with manual bending (not even on 6061 scrap bent in the brake).

Some tips:

* Eyeballing a scored mark in any relationship to the fingers, or the center of the bending plates, does not work, not reliable at all.

* Distance from the front plate back WILL get consistent bends, within + / - about .003 inches.

* When bending .038 / .040 alu sheet on a brake (depending on setup), expect the inside distance to decrease on each side of the bend by about .007 inches (about .014 inches off total length). Expect the outside length including sheet thickness to be about .033 inches longer on both sides of the bend. This means that with thickness subtracted, expect .007 inches to be lost in the bend on the outside as well. This is for a rather tight bend.

* Whatever you do, DO NOT SCORE THE METAL at the bend, it will provide a place for the metal to crack when put under the stress of bending. Even scoring on inside bends causes trouble, it forces the outside of the bend to stretch too much.

* DO do relief cuts for the bend sides. You can't mill flat sheet metal perpendicular to the cutting tool with a true 90 (all cutting tools for a mill cut in circles), and even if you could, a bend nearby would cause tearing. For .038 / .040 sheet, I'd recommend at LEAST a .033 inch relief, double that is even better (I'm using .063 inch, about half the dia of a 1/8 (.125) inch end mill).

I don't think this type of brake can bend this thickness of sheet much tighter than this without damage to the fingers.

5052 is a bit uglier to machine, but not so much as to be problematic. I'm using a VERY slow feed rate and shallow cuts to be on the safe side.

Next up, cut the latest rev of the drawing, then cut teflon hub / tendon pieces, then design thumb tendon guide, then test and see if guides are needed at the servo hubs and make / cut those if needed. I also need to design a bot-side wrist mount bracket, and cut the wrist rotation hub on the lathe. Getting closer, but I'd be kidding myself if I didn't acknowledge there is a fair amount of work left.

Other than hands, I designed and cut some alu pulleys for hip / wrist to offset servos, they look good! I need to make a mandrel for the pulleys on the lathe to cut the pulley v-groove, though - there isn't much metal on the metal HSR-5498SG hubs, and my lathe chuck jaws aren't flush at the end - they have a bevel, which means I need a mandrel to position the pulley out and away from the chuck jaws.

Take Care,
Paul
It's a G0556 from www.grizzlyindustrial.com. :) I should mention, if you haven't bought any of the chinese import machines (Mill, Lathe, etc), the quality may let you down. After having taken the machine apart, I see the hidden flaws - gouged mounting plates for the fingers, fingers themselves have "splinters", some parts misaligned aeshtetically - basically it's a typical chinese import machine: a rough piece of equipment that does what it should, but isn't going to be great in all respects.

I tore down, cleaned, and reassembled the bending brake, CNC'd a dozen test pieces of 5052, and here are the results of bending:

* First two pieces were to test 5052 bending, no measurements made, bent fine.

* Set up a perpendicular guide, set up the machine for thickness and bend offset, bent some long scrap strips of .038 inch thick 6061, verified perpendicular bends. Results look great.

* 5052 is .040 inches thick, not .038 like the 6061 (yes, splitting hairs here).

* Measured all 10 remaining CNC'd test pieces, range from 0.999 to 1.001 (target size, 1.0 inches). Good repeatability by the CNC machine.

* Bent a few pieces, dialing in the location of the bend with a known offset. Distance seems to be around 2.010 inches from the front flange to the intended center of the bend.

* Bent remaining pieces at an offset of 1.510 inches, resulting in a middle bend of the 1 inch length test piece.

RESULTS:

I encountered NO splitting of alu using 5052 OR 6061 on what is a fairly tight radius.

But, the more interesting part is the numbers applying to the bend itself: as with manual bending, the resulting average length of each side of a bent piece of 1 inch aluminum is .5333 (.533) inches over 10 pieces. I WAS AMAZED!!! Virtually identical change in distance, using two different alloys, same width / length pieces, manual versus machine bending. The difference? No cracks as with manual bending (not even on 6061 scrap bent in the brake).

Some tips:

* Eyeballing a scored mark in any relationship to the fingers, or the center of the bending plates, does not work, not reliable at all.

* Distance from the front plate back WILL get consistent bends, within + / - about .003 inches.

* When bending .038 / .040 alu sheet on a brake (depending on setup), expect the inside distance to decrease on each side of the bend by about .007 inches (about .014 inches off total length). Expect the outside length including sheet thickness to be about .033 inches longer on both sides of the bend. This means that with thickness subtracted, expect .007 inches to be lost in the bend on the outside as well. This is for a rather tight bend.

* Whatever you do, DO NOT SCORE THE METAL at the bend, it will provide a place for the metal to crack when put under the stress of bending. Even scoring on inside bends causes trouble, it forces the outside of the bend to stretch too much.

* DO do relief cuts for the bend sides. You can't mill flat sheet metal perpendicular to the cutting tool with a true 90 (all cutting tools for a mill cut in circles), and even if you could, a bend nearby would cause tearing. For .038 / .040 sheet, I'd recommend at LEAST a .033 inch relief, double that is even better (I'm using .063 inch, about half the dia of a 1/8 (.125) inch end mill).

I don't think this type of brake can bend this thickness of sheet much tighter than this without damage to the fingers.

5052 is a bit uglier to machine, but not so much as to be problematic. I'm using a VERY slow feed rate and shallow cuts to be on the safe side.

Next up, cut the latest rev of the drawing, then cut teflon hub / tendon pieces, then design thumb tendon guide, then test and see if guides are needed at the servo hubs and make / cut those if needed. I also need to design a bot-side wrist mount bracket, and cut the wrist rotation hub on the lathe. Getting closer, but I'd be kidding myself if I didn't acknowledge there is a fair amount of work left.

Other than hands, I designed and cut some alu pulleys for hip / wrist to offset servos, they look good! I need to make a mandrel for the pulleys on the lathe to cut the pulley v-groove, though - there isn't much metal on the metal HSR-5498SG hubs, and my lathe chuck jaws aren't flush at the end - they have a bevel, which means I need a mandrel to position the pulley out and away from the chuck jaws.

Take Care,
Paul
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Tue Jun 04, 2013 11:34 am

Post by PaulL
Tue Jun 04, 2013 11:34 am

Wow, it's been a really, really long time since I've posted any kind of update on the hands.

As I have my electronics (main board, sub board, control methodology) and software stuff mostly sorted, my focus has shifted back to hardware.

I have been debating on whether I should post updates to what I'm doing, or if I should just post when I'm done, as it's taken a long time to make progress - life happens. Well, here's a small update.

Since the last update, I've done two more bracket revisions. The latest revision is ready for the bending brake, and I cut test parts Sunday to set the backstop I added to the brake a few weeks ago. My CNC machine got a cabinet, a project that took about a month, and included addition of a spindle controller, something I did without in the past. Now, I can let the long part of the CNC job run unattended (the 1/8" end mill), and chips are contained. The cabinet is a really nice change.

The bracket design really hasn't changed much, just some improvements to make it easier to build. One of the changes is in the front bracket where the fingers mount, as the original hand-bent version in the pics above isn't something that can be done on a typical pan and box brake like the one I have. The thumb piece is now separate, and the tabs that wrap around the base of the hand are gone, tabs now come off the base instead.

The backstop on the bending brake will be fixed at 2 inches, that will work for all of the parts I need to bend - torso, wrists, hips, all of the bracket work. I'll use spacers cut on the mill to position the part in the proper location for the bend. I cut the spacers I'll need to bend the hand brackets when I cut the test bending strips.

I think things will move a bit more quickly from here on out. :)

In the pics of the mill cabinet and mill, the zip lock bag contains the hand parts I need to bend after I set the brake backstop.

Image

Image

Image

Image
Wow, it's been a really, really long time since I've posted any kind of update on the hands.

As I have my electronics (main board, sub board, control methodology) and software stuff mostly sorted, my focus has shifted back to hardware.

I have been debating on whether I should post updates to what I'm doing, or if I should just post when I'm done, as it's taken a long time to make progress - life happens. Well, here's a small update.

Since the last update, I've done two more bracket revisions. The latest revision is ready for the bending brake, and I cut test parts Sunday to set the backstop I added to the brake a few weeks ago. My CNC machine got a cabinet, a project that took about a month, and included addition of a spindle controller, something I did without in the past. Now, I can let the long part of the CNC job run unattended (the 1/8" end mill), and chips are contained. The cabinet is a really nice change.

The bracket design really hasn't changed much, just some improvements to make it easier to build. One of the changes is in the front bracket where the fingers mount, as the original hand-bent version in the pics above isn't something that can be done on a typical pan and box brake like the one I have. The thumb piece is now separate, and the tabs that wrap around the base of the hand are gone, tabs now come off the base instead.

The backstop on the bending brake will be fixed at 2 inches, that will work for all of the parts I need to bend - torso, wrists, hips, all of the bracket work. I'll use spacers cut on the mill to position the part in the proper location for the bend. I cut the spacers I'll need to bend the hand brackets when I cut the test bending strips.

I think things will move a bit more quickly from here on out. :)

In the pics of the mill cabinet and mill, the zip lock bag contains the hand parts I need to bend after I set the brake backstop.

Image

Image

Image

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Post by PaulL » Tue Jun 04, 2013 11:38 am

Post by PaulL
Tue Jun 04, 2013 11:38 am

I should add, these pics are a couple weeks old, there were a few more things done since then.
I should add, these pics are a couple weeks old, there were a few more things done since then.
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

Hand Update...

Post by PaulL » Sat Jun 08, 2013 9:41 pm

Post by PaulL
Sat Jun 08, 2013 9:41 pm

Got the bending brake straight, bent the latest bracket version. Still not quite right. I'm a bit frustrated at the bends on the main bracket, and I basically added new problems by adding the tabs for the top plate. For the tabs on the top to be right, two bends must be really accurate. Also, there's a lot of fiddling for the various depths required for bending.

I'm going to take what I have and swap out the bends for separate small brackets, using the same small bracket over and over in the design instead of these multiple odd depth bends. I think the look will be OK, and I should be able to work this into the design pretty easily. I'll make separate pieces of the sides, base, and top, and join them with uniform brackets. If I screw up a small bracket, it's no big deal, bend another one. If I screw up the bends on the designs shown here, the part is scrap.

Pics below: yes, some pieces are snapped off, I was testing the strength of manual bends - they aren't strong at all, cracked, etc. Bending with the brake (newest rev) is great, very strong. Oldest rev on left / back, most recent on right / front:

Image

Image
Got the bending brake straight, bent the latest bracket version. Still not quite right. I'm a bit frustrated at the bends on the main bracket, and I basically added new problems by adding the tabs for the top plate. For the tabs on the top to be right, two bends must be really accurate. Also, there's a lot of fiddling for the various depths required for bending.

I'm going to take what I have and swap out the bends for separate small brackets, using the same small bracket over and over in the design instead of these multiple odd depth bends. I think the look will be OK, and I should be able to work this into the design pretty easily. I'll make separate pieces of the sides, base, and top, and join them with uniform brackets. If I screw up a small bracket, it's no big deal, bend another one. If I screw up the bends on the designs shown here, the part is scrap.

Pics below: yes, some pieces are snapped off, I was testing the strength of manual bends - they aren't strong at all, cracked, etc. Bending with the brake (newest rev) is great, very strong. Oldest rev on left / back, most recent on right / front:

Image

Image
PaulL offline
Savvy Roboteer
Savvy Roboteer
Posts: 423
Joined: Sat Sep 15, 2007 12:52 am

PreviousNext
PreviousNext
135 postsPage 3 of 91, 2, 3, 4, 5, 6 ... 9
135 postsPage 3 of 91, 2, 3, 4, 5, 6 ... 9