Car door actuators would not need position sensing, they are just up or down, same as a solenoid.

I would go with door actuators unless you really need the extra power of the starter solenoids. at around 15A, and 20 solenoids you are going to need huge batteries, even if they don't all fire at once. You will also need a bucket of MOSFETS or 30A relays to drive the large solenoids which will either be expensive, bulky, or both. You could use much cheaper parts to drive the door actuators, which probably take 5A or less.

I have a few starter solenoids that you can check out at the next build day if you want. They weigh 800 to 860 grams and have about 1 cm of travel, which is typical of the units I found at PnP.

Another point for the door actuators is that you can buy them reasonably cheaply new from car places and Jaycar. The solenoinds will likely be more expensive unless they come from PnP.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Do you have any drawings of your designs?
_________________
So even the rain that falls isn't actually going to fill our dams and our river systems

I'm currently working on a cad... it's really tough to draw in a spherical environment...
_________________
( •_•)

( •_•)>⌐■-■

(⌐■_■)

YEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAH

You wouldn't be designing a robot ball by any chance?
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

nick i had the same thaught lol.

angus PWM is really sucky for transfering information digitally. (PWM is what the radios use.
If you want your own controll system you really should look at digital radio, even the ones from jaycar for like $10.

The big problem with sensors is basically they suck ;->

a "cheap" pezo rate gyro will drift about 1deg/s (so after 90 seconds where you think you are pointing and where you actually are pointing can be off by 90 degrees)
ok so thats bad.

what you need is a full 6dof (6 degrees of freedom) IMU (inertal measurement unit)

Basically it works thusly. you assume that on average gravity points down.
(over a period of say 20 seconds) so if you average the acceleration vector from the acceleromiters you work out which way is the direction of down.

In the short term however (assuming you want to roll the ball in a different direction) you need the rate gyros to tell you whats going on.

Its not rocket science but its pretty dern close lol.


Re controll chips.
I'd suggest PIC's rather than PIC axes.
the 16F877A runs at 5Mhz, has 40 I/O pins, 8x A/D inputs, 2x hardware PWM outputs, Hardware serial I/O, Hardware SPI, Timer interrupts and a bunch of other stuff.

Theres a little more setup to program them ( a $25 programmer) but the capabilites are so far infront of a PICaxe that on a large project it's really worth it.
JAL 2.0 is coming out soon which will make a big difference too, you'll be able to use the full capabilities of the PIC in a basic like language.
360 bytes of ram, 16kb worth of program storage and 256 bytes of eeprom.

You may want some external ram though to do your acceleration vector calculations but perhaps not.
_________________
Mechanical engineers build weapons, civil engineers build targets

Yep ball I just don't know how many chinese designers trawl this forum lol
_________________
( •_•)

( •_•)>⌐■-■

(⌐■_■)

YEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAH

It wasn't too hard to guess! I have two alternate ways to drive a ball (niether is new) that we can talk about off-line if you want. They both give you a clean skin on the ball as well. <Jake is nodding as he reads this, he knows what I am refering to>.

I think either design would be lower power consumption and at least one would be easier to build, although it doesn't steer as well.

EDIT: one of them is a great excuse to use brushless motors
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

now im not 100% positive im following it but i think if im even halfway there you could probably make it servo driven and control it by using a weight in another words the servos move the weight which changes the center of gravity and hence the ball rolls. a realy realy realy bad example is those ferret balls basically there is a weight connected to a single motor which spins throwing the weight around. if you were to slow that down to the point where you could hold it in position (servos) you could use one servo to point the other servo in the direction you want to drive in and the 2nd servo would move the weight forward or backward the biggest problem i see is getting the outer skin to to roll freely accross the inner frame work (possibly some ball castors or something similar) and then you have the part where you have to make the outter skin. i have no idea on how it would run on a guestimation probably something like how you crontrol the ball in super monkyball (bad example i know but get over it) you would definatly get free roll even as your adjusting the direction your going in (ie your rolling forwards dont expect to turn in a hurry so slowing down would be very very very essentual)
_________________
as a shadow is seen and not heard i am here
-Waddy the phoenix