Phoenix - One Person Ride - Etek Brushless Motor Driven

Lynxmotion Legged Robots
1

Scaled Up Phoenix

I’d like to use the SSC-32 to power a scaled up Phoenix. I have 18 Briggs & Stratton Etek brushless motors in hand.

Here are the specifications:
Torque constant = 1.0 in-lbs/amp
Voltage constant = 84 RPM/volt
100A continuous, 200A/1 minute, 300A/30 seconds
6 HP continuous, 15 HP peak
88% efficient
24-36V design voltage, 48V absolute max
20 turn, fitted with hall effect sensors (+12V powered)
10 milliohm phase resistance
Output shaft 1.00" dia. by 2.5" long
Weight: 22 lbs.
I own a Phoenix that I could use as a template to scale up from but would prefer to buy it from Lynxmotion (if they would make it). Could be a nice promo for the Phoenix but I’ve sponsored the motors and would pay for the parts for this one just to have it for my own. Now I just need to figure out how far to scale up the parts and how to power 72 volts with an SSC32. Not a very big order for this forum. 8)

2

Nice motors! But you’re going to need a brushless motor driver. The SSC32 is made to drive R/C servo motors, not conventional servo motors. 72 volts is much too much for the SSC32!

Alan KM6VV

3

um you have 18 5KW 3-phase motors you want to make a hexapod robot from? what? you have 300lb in motors alone, and you have not accounted for gearheads, batteries, or electronic drives yet… the latter of which will most likely require active cooling of some sort. Aaand you are not talking about 0.090" aluminum brackets made on a break here, more like .250" to .375" steel plates and channel welded together.

Have you truly considered the scale of this project? O.o

4

Yeah right! I’m out! :open_mouth: :unamused: LOL

5

Oh c’mon Jim!! I know you can do it!!

(you might need some serious OT on it though) :slight_smile:

6

If you are serious about building something that big, then you probably should sell those motors and get smaller gear head ones. The below winch goes on sale at harbor freight in a week or so at their fall parking lot sale for $40. Remove the cable and weld an attach bracket to the spool. Till then you can study the bottom spider car.

harborfreight.com/cpi/ctaf/d … mber=92860

flixxy.com/spider-car.htm

7

I’d like to see this thing (from a safe distance), but I’ll admit it seems like at a minimum it’s going to be pricier than most cars. I’m not sure how serious you are, or what your budget is, but here goes :slight_smile:

It’s tough to judge what level of current draw you’ll actually need, but I’m assuming the safe continuous drive current. There’s a motor controller I ran across while searching around that should drive these, called the “Open Source Motor Controller”. It comes in a kit for about $600 / two channels.

So in principle that’s around $5400 for motor controllers, plus tax and shipping. That gets you more or less from the SSC-32 stage to driving the motors around in a circle; the feedback aspect would need some additional circuitry, but in principle isn’t that difficult.

The biggest issue I see is battery power. A 37V, 5Ah lithium polymer pack would run one of your motors for 3 minutes for about $600/ea. This is at 100A, which would be peak discharge for the pack (20C). So 18 of those ($11,000) would run the whole robot for 3 minutes.

Leaving aside electronics (which are pretty minor by comparison), the rest of the cost would be mechanical; you may have estimates of this.

I’m just throwing numbers around, but supposing you wanted to be able to run it for 15 minutes or so, I’d guess you’re looking at something in excess of $75,000 once you get taxes, etc. in, and assuming you have lots of spare mechanical parts lying around.

Man, this is starting to sound fun…

[Edit: Lithium ion may sound ridiculous, until you realize the weight of 3.2 kWh of batteries / 3 minutes of runtime in something like lead acid.]

8

One leg mockup for bench testing

Motors
3 Etek brushless motors
Total $0 (paid for, actually I have all eighteen)

BL1500 - 50A Brushless DC Motor Controller with Hall Sensor Inputs $275 or the BL1500BP - 2 x 60A $325
Total $825 to $975
roboteq.com/brushed-dc-motor-con … llers.html

One leg initial mockup will be done with PVC pipe.

I’m researching tubular design construction used in ultralight construction, bicycle, and go cart racing applications. Some people have been using old bicycle frame tubes from scrap.

Batteries are a problem. I’ve been researching options such as golf cart batteries, lithium batteries, and marine batteries. I’m working on an EV project that is currently using 6 marine batteries and will use three of them for my initial testing (one battery for each motor). Will probably end up with an umbilical cord and pull behind cart.

3 Optima BlueTop batteries $169 - 44.7lbs each
$509 (should be able to get some close out ones cheaper)

One Leg?
$1,500

One Quadrapod?
$6,000

One Hexapod?
$9,000

What is the average cost of a NASCAR stock car?
•Drivers’ salaries: Not Listed
•Team salaries: $2.5-3.5 million, or an average salary of ~$30,000/year for ~100 employees
•Travel: $1 million per team
•Tires: $1 million per team ($20,000 per race weekend plus testing)
•In-house engine program: $3.5 million+

•Cars: $1-3 million per team

What would optima pay to advertise on this at the fair or a car show? Official sponsorships of NASCAR generally cost at least $1 million a year in rights fees. I’d settle for much less. :laughing:

9

Yeah, absolutely; if you get the weight of the batteries off of the legs, that opens up a lot of possibilities… Lithium ion is about 1/3 of the weight of lead acid, I think, per the same capacity (and somewhat more than 3 times the price). Maybe NIMH would be a good middle ground?

That controller does look like a great option too, assuming you can slave both channels together without a lot of rewiring.

The mechanical design is still pretty daunting to me, but I guess the biggest question is how much time, money, and help you have :slight_smile:

One question that might be worth figuring out is the actual power to weight ratio of your motors and the full scale legs/body vs. something on the scale with the Phoenix. Certainly things are going to end up being pretty heavy, but knowing whether you can actually lift a chromoly vs. aluminum vs. carbon fibre chassis would give you an idea of what you need to work with. Bike parts perhaps?

10

This article shows a diagram of how to hook the SSC-32 to the Roboteq motor controller. It even links back to the Lynxmotion site! The Mini SSC in the picture is the SSC-32.

http://www.roboteq.com/images/animatronicshardware.gif

The article doesn’t go into great detail but it does seem to indicate the servo connection pins (from the SSC-32) going into the RS232 port on the motor controller.

roboteq.com/robotics-motion- … props.html

11

Maybe you should concider a gas generator instead.

hondapowerequipment.com/prod … egory=play

12

At first I laughed, but you know if you connect several large power supplies to the output of the generator you have yourself one bad ass Hybrid. :wink:

13

I’ve seriously looked into these for my electric vehicle project. Would be a nice option so I don’t get completely stranded if I run out of juice. You can program them to kick in if the batteries drain below a certain level.

14

It looks like the motor controller can run in R/C mode. It comes with a premade cable for hooking a R/C wire to the RS232 on the motor controller (lower right in picture). Can it really be that easy? Plug the R/C cable from the SSC-32 into this and I now have a gigantic servo? Sure looks to be. 8)

http://www.roboteq.com/images/orderkit.gif

15

Could work, but if I was going to drive that big a motor, I’d go to a true DC Servo motor, and a servo motor controller.

geckodrive.com/

With the addition of encoders, you’ve got a full closed-loop system to control your motors precisely (and very fast!).

The AX3500 or whatever can use RS-232, and encoders, as well as closed loop. That’s a good way to go!

Alan KM6VV

16

The motors were purchased for robotic EV projects and I have 24 of them. Then I thought that’s more than enough to build a Phoenix! That got me researching to see if I could actually turn them into servo motors. On one part of Roboteq’s site they say the BL1500BP motor controller can be hooked up to an optical encoder (it definitely can be for the brushed motor controllers they sell) and run closed-loop. In another part of the site they say the BL1500BP can’t. I may have to check with them to see for sure. If the brushless controller could be hooked up to the encoder that would give me that level of accuracy. I thought I might try to just send the commands and see how it does on it’s own without an encoder. I’m going to purchase one to test with so I’ll be able to find out, if not it goes into the EV project so nothing lost if it doesn’t work. I might also be able to place a sensor on the bottom of each foot to bring it down the rest of the way and that might be good enough.

Right now I’m borrowing everything from the EV projects so I can do a proof of concept on the Phoenix. Nothing beats working with the real thing! If the motors all end up in EV’s eventually, I might take the funds from the EV projects and go your route. Thanks for the info.

17

If the motors are brushless, then you’ll definitely need a brushless controller. Brushless motors aren’t used that often in servo systems that I know of.

But if you can live with just running a joint on a leadscrew to it’s two limits, then you can get by a little cheaper. brushless motors often use the hall-effect (encoder) to determine when it’s time to switch the fields around.

Of course it’s better if you have closed loop control on the “actuators” for each joint. Could be in the form of a long linear pot ($$). With true servos and encoders, you’d put the encoders on the motor shafts, of course.

Alan KM6VV

18

You know, you could put a three-axis accelerometer and/or gyros on each joint. The ADXL345 connects via I2C, so you could probably do it with 4 logic wires per leg. Twelve bit sampling plus tap sensing and free-fall detection (which might let you detect ground contact).

An XMOS chip or something similar could easily handle finding orientation from the leg kinematics and the tilt, since you could delegate threads to as many I2C lines as needed, and still have lots left over to get 50 MIPS per joint :wink:

It’d suck to code though.

But in general, with this kind of price range and size, you can do things you just can’t at the small scale, like wacky motion control sensors.

19

I’d definitely put a controller on each leg, which would allow all the sensors you suggest!

Alan KM6VV

20

Just to give everyone an idea of the size of the project I took a picture of one of the Etek motors next to my Phoenix.

http://www.otherrobots.com/lynxmotion/etekphoenix.jpg