Servos

[YAN-1]

Hi everyone. I'm attempting to build my first robotc arm and I need some help! I will be using servos for the joints and will control them using PIC microcontrollers. The arm I'm required to do has to have a shoulder, an elbow, a wrist, and a gripper. It should handle small loads like hold a cup of water and pour it. I've been looking at servos by Hitec and Futaba but I'm not sure what to choose! Please help me by directing me to torque specifications and prices. Thanks a lot.

[Admin]

this should be useful for determining required robot arm torques
http://www.societyofrobots.com/mechanics_statics.shtml

my personal recommendation is just get the most torque you can . . . but remember to also consider speed of the servo too
seen this yet?
http://www.societyofrobots.com/actuators_servos.shtml

lastly, have a look at the picture of the robot arm at the very bottom of this page
http://www.societyofrobots.com/materials_hdpe.shtml
i used a servo for the gripper, a servo for vertical motion, and a dc motor for rotation motion. the batteries are also used as counterweights.

one last thing good to mention, try to keep your weight as close to the base as possible. you want your arm as light as possible.

[YAN-1]

Thanks a lot. Well I've been looking around and I found some servos by Hitec and Futaba. The high-torque ones are rated at around 200 oz-in. There is one that says 161 oz-in, which is around 11.6 Kg.cm. This means that it will be able to provide a torque equivalent to a mass of 11.6 Kg hanging from an arm of 1 cm, right? I know this will need some more detailed calculations, but I guess this will be enough. I mean if it is used for the shoulder and it is to carry another one just like it (the servo weights 134g) at the elbow and one at the wrist and a small gripper, it should be fine, right? Assuming of course that the dimensions of the arm are small enough so as not to produce high torques from the weights of the components. I would appreciate your advice and directions.

P.S. I have to build the arm so that the shoulder is the first joint and it emerges from a wall so it won't be exactly like that picture. The shoulder motor will have to carry the whole arm and the load.

Nichola V. Abdo

[Admin]

Yeap that is all correct. If after your more complex calculations are done and you want to just verify them, just send over the info.

Also, dont be limited to just one servo per joint. The base joint can have two servo's (one on each side) rotating the arm. This way you can either double the torque, or have two smaller (and cheaper) servos actuating it.

Lastly, for the base joint, consider using a worm gear. Worm gears work only one way, so that your motor can rotate the arm, but the weight of the arm cannot rotate itself. You would still require the same amount of torque, but you only need power to move the arm and not hold its position. The other possible benefit of worm gears is the high torque possible. Something to look into . . .

[YAN-1]

Also, dont be limited to just one servo per joint. The base joint can have two servo's (one on each side) rotating the arm. This way you can either double the torque, or have two smaller (and cheaper) servos actuating it.

I just wanted to clarify one point. I'm not sure what you mean by 'base joint'. The arm won't be resting on a base like the one in the picture you showed me. It will resemble the shape of a human arm from the shoulder to the hand. So the 'base' is the shoulder joint which will be fixed to a vertical post or something and the arm will be hanging freely from there and will move to carry objects like the human arm. Do you think the shoulder will be able to carry the whole arm? Thanks a lot.

[Admin]

Yea, the shoulder would be the base joint. I am just saying do not be limited to just one servo per joint.

The reason I suggested a worm gear is because to hold an arm up requires constant power, but with a worm gear you only need power to move. It is very common to use worm gears in robot base (shoulder) joints.   

[YAN-1]

Hi. Well I am facing some trouble here. I found a servo motor that will give up to 24.7 Kg.cm of torque and it was rated under 'monster servo, the servo that will move just about anything'! I saw a couple of robotic arms on some sites that use servos that are weaker than this and that worked just fine. But when I attempted to do a simple calculation for the robotic arm I intend to build, I found that if I used this monster servo for the shoulder joint, it won't be enough to lift the arm at all! 

To make sure that I am not doing the math wrong, I sketched the arm and presented the equation I used for calculating the torque carried by the shoulder joint in the worst case. Here, L1, L2, and L3 are the links, S1, S2, and S3 are the servos, G is the gripper, and the load is a glass of water. The numbers I assumed are as follows:

Mass of each link = 300 g
Mass of each servo = 150 g
Mass of gripper = 300 g
Mass of load = 200 g

With these numbers, I get a torque of 5.5 Nm that will have to be provided by the shoulder servo. This corresponds to a torque of 56 Kg.cm, which obviously cannot be supplied by the monster servo! Am I doing it wrong? I don't think I can make the links any shorter in order to reduce the torques because then the arm won't be practical. Please help me out. Thanks a lot.

Nichola V. Abdo

[Admin]

So your equation is correct. You forgot to give me lengths, so I just assumed 10cm each. I assume they are shorter?

Plugging in your numbers, I get:

(.3*10/2+.3*(10/2+10)+.3*(10/2+10+10)+.15*10+.15*20+.3*30+.2*30)*9.81 = 323.73 kg*cm required

And this is what you get if you dont have the gripper and water weight:

(.3*10/2+.3*(10/2+10)+.3*(10/2+10+10)+.15*10+.15*20)*9.81 =
176.58 kg*cm required

So you calculations all look correct to me. Perhaps why your calculation looks so large is that you have two really heavy loads at the very end of your arm, making the torque nearly twice as high.

Possible solutions:
- Make your shoulder joint the translating one (or at least not a vertical motion). That way you can make it rigid, and all your torque will then end up at S2, so it will be much less. Can you grab the glass from the top instead of the side?

- Use two servos at your base (your calculations say 2 servos will be more than enough)

- Go to using a larger DC motor

In all cases I still highly recommend using a worm gear . . .

[YAN-1]

Hi. Well thanks a lot. Aren't the numbers you got supposed to be in N.cm (you multiplied by 9.81)? I'm sorry I forgot to give you the lengths of the links. L1 and L2 are 20 cm and L3 is 5 cm. What I don't understand is how does an arm like this one http://www.crustcrawler.com/products/arm6.php?prod=11 operate with these servos. They are smaller than the one I was talking about. Can't I use a gear set at the shoulder with the same servo? And how do I mount 2 servos together if I am to use more than one at the same joint? Thanks a lot.

Nichola Victor Abdo

[Admin]

oops sorry you are right, they should be in N*cm

so I recalculated with your numbers and got what you got:
(.3*20/2+.3*(20/2+20)+.3*(5/2+20+20)+.15*20+.15*40+.3*45+.2*45) =
56.25 kg*cm or 5.52N*m

56.25 (kg cm) = 781.17 inch ounces

But that robot arm uses a 343 inch ounces capable arm and claims to lift 403.41g plus its on body weight . . .

Looking at the arm, based on given weight and length, and an estimated center of mass at 1/2 distance (probably more), it would require at least
2.34 lbs * 16 oz/lbs * (17.2"/2) = 321 oz inches
just to lift itself . . .

Based on the servo specs, it can only lift another 22 (343-321 = 22) inch ounces (or 1.58 kg cm) at full extension. Your glass would require 200g*17.2" = 8.74kg cm if their arm were to lift it. So their robot arm would not be able to lift your load.

So the reason why their robot arm claims to lift so much is that the rating is NOT at full extension ("max lift capability" just means how much it can lift, not at what distance). If the robot arm was retracted half way, the torque required would also be halved (since the distance is half). Theoritically, the arm can lift its maximum amount when it is completely vertical because the distance is near zero.

Attached are two pics. The first shows two servos moving one arm. The other shows an idea I got of basically a servo using a string to move the arm up and down. The reason it would work is because of mechanical advantage, but would involve a list of other problems so its just food for thought. You may just have to redesign your arm a different way? Perhaps try to reduce arm weight by using different materials or putting holes in it?

[YAN-1]

Well I am considering finding a stronger servo for the shoulder and use this monster servo for the elbow. Do you know of any type of servo that can provide this torque? I would appreciate that a lot. Thank you very much.

Nichola V. Abdo

[Admin]

Again, your only option is to do gearing. There are no servos any stronger.

Robotzone has this item that can multiply torque by x5 on the HS-805BB servo:
http://www.robotzone.com/customer/product.php?productid=186&cat=51&page=1
But personally its way over priced for a gear and some plastic hinges. Better if you copy and make it yourself.

But I still say worm gears are best. Just google worm gear box and find one to attach your servo to.
http://www.google.com/search?sourceid=navclient&ie=UTF-8&rls=GGLC,GGLC:1969-53,GGLC:en&q=%22worm+gear+box%22

[Admin]

FYI, Hitec has come out with new 7.4V higher torque digital servos.

I am not sure if they are selling them yet, but the HSR-5995TG has 416.61 oz*in (30 kgcm)

[YAN-1]

Again, your only option is to do gearing. There are no servos any stronger.

Robotzone has this item that can multiply torque by x5 on the HS-805BB servo:
http://www.robotzone.com/customer/product.php?productid=186&cat=51&page=1
But personally its way over priced for a gear and some plastic hinges. Better if you copy and make it yourself.

Well thanks a lot. I think I'm gonna use the HS-805BB for the shoulder with that gear box using a gear ratio of 4:1. This will give a tourque of about 97 Kg.cm, which is very safe. As for the elbow, I think I'll use the same combination but with a gear ratio of 2:1. The previous calculations, when done on the elbow joint, show that it needs to carry around 25Kg.cm in the horizontal position. This motor/gear combination will give it around 49 Kg.cm. But as for this gear box, can its input be easily attached to the servo and can its output connect to the arm easily as well?

And as for the servo itself, it says here: http://www.servocity.com/html/hs-805bb_mega_power.html that 'Operating Angle: 45 Deg. one side pulse traveling 400usec'. What does that mean? It only moves forwards and backwards within a range of 45 degrees? That's not enough! And won't this range be divided by the gear ration if I use the gearbox?! Also, what does '360 Modifiable: Yes' mean and how do I adjust it to rotate for 360 degrees?

Thanks a lot. I really appreciate your help.

Nichola V. Abdo

[Admin]

You might be interested in this too:
http://www.servocity.com/html/hsr-5995tg_ultra_torque.html

I have never used that gear box, I couldnt tell ya if it is easy. The pic looks simple tho . . .

All servos will travel about 180 degrees, unless otherwise specified. Remember, if you do a 4:1 gearing, your full motion will be 180/4 = 45 degrees.

I am not entirely sure what Operating Angle means, I couldnt find the definition anywhere. My GUESS is that it is how many degrees your servo will travel with a single pulse of 400us.

[YAN-1]

Hi. Does no one know how I can find out the range of motion of the 805BB servo or any other servo? It would be really bad if it were just 45 or 90 degrees since I intend to gear it up. Thanks a lot.

[Admin]

I assure you it is 180 degrees.

[YAN-1]

Hi. Well I sent an e-mail to servocity and turns out it is 90 degrees. The thing that is puzzling me though is the HS-805BB 5:1 Gearbox http://www.servocity.com/html/robotzone_servos.html . It looks too good to be true since they claim it can raise the torque without affecting the operating angle by bypassing the internal potentiometer of the servo motor. Is this possible?! I thought the Servo was mechanically restricted by its internal potentiometer, not just by its signal. Does this mean that I'll have to open it up and change things around? Please Please help. Thanks a lot.

Nichola V. Abdo

[Admin]

Hmmm I guess I was wrong then . . . 

I am shocked they replied to your email, they always ignore me . . .

So I didnt read the description until now, and I am as confused as you are about it. I dont see how it is possible without modifying the servo in some way . . . call them up and let us know 

[YAN-1]

Well they barely replied! They answered in short sentences and didn't asnwer all the questions. I also asked Hitec and their reply was worse! Anyways, I sent Servocity asking about the gearbox. I hope that'll do me any good. The thing is I live in Jordan and I'm not sure whether there is a holiday in the states now because of Easter or not so I'm not sure if they'll reply before it's too late. As for calling them up, I fear that I won't be able to communicate well with them or get a clear answer for my inquiry over the phone. 

[Admin]

Ok so I called them for you.

Apparently what you do is remove the mechanical stop and rewire the pot to their pot on the servo. He said it comes with an 'easy to follow manual with pictures' to tell you how to do it.

The transit time is increased by the ratio you choose. If say you use 5:1, it will take 5 times longer to do a full rotation.

He also said they sell 'quite a few of these.' O and that they have aluminum gears as optional if your plastic might break or wear or something. I think you might have to special order the aluminum gears tho if you want them, I didnt ask . . .

Hope this helps.

[YAN-1]

THANK YOU SOOOOO MUCH     You really didn't have to go through this much trouble. I appreciate it. Well guess I'll just have to deal with their sales department now and figure out a way to ship this thing to Jordan. I just hope it's not too heavy in weight to be located at a joint in the arm but with the torque increased, that shouldn't be a big problem. I also cannot understand why neither they nor Hitec answered me regarding the frequency of the control signal that I should send the servo.

Sorry for troubling you but do you also know where I can find a simple electrical gripper? Something not too expensive.

Thanks a million. This forum rocks! I appreciate the help.

Nichola V. Abdo

[Admin]

http://www.societyofrobots.com/robot_parts_list.shtml

stuff I found just from my partlist:

http://www.robotstore.com/store/product.asp?pid=267&catid=11
http://www.lynxmotion.com/Search.aspx?txtSearch=gripper

and of course froogle:
http://froogle.google.com/froogle?sourceid=navclient&ie=UTF-8&rls=GGLD,GGLD:2004-37,GGLD:en&q=%22robot+gripper%22