low-cost programmable servo controller?

http://www.pololu.com/products/pololu/#servcon

Yes, that's the $20 serial servo controller that I mentioned having to
add to the Baby Orangutan to make a complete solution (i.e. one that's
programmable, and doesn't need a computer telling it what to do).

But can you suggest anything cheaper than this plus the Baby Orangutan?  
Maybe something that does it all in one board?

Thanks,
- Joe

You might check to see what the below chip can do.

http://www.kronosrobotics.com/xcart/customer/product.php?productid 428&cat24&page=1

Joe, what are your requirements ? as in max $$ to spend, and a brief description
of what you want the board to do? We
might be able to help more if you can scribble that out for us.

Having said that, I just bought a brand new complete serial version BoeBot kit
off of eBay for around $80 just to have a
cheap bargain basement controller with servo capability for a much larger robot
I am developing. The BOE has 4
connections for servos designed into the board, and offers some amount of
programmability for sensor inputs, etc. I
specifically went for the BoeBot because I am familiar with it from prior use
and knew it had the servo connections
broken out on the board, plus comes with 4 AA battery power pack, etc. Includes
2 continuous motion servos, also. The
thing I like about the Parallax products is that due to thier marketing exposure
I can resell them back on eBay when my
need for them is over. It could get you going until you find th100% ideal
solution.

Hope that helps a little bit!
JCDeen

Sorry, I thought I was clear about that: I want a programmable
controller that with four (or more, obviously) servo outputs.  As for
max $$, well my current best plan is a Pololu Baby Orangutan ($25) plus
a Pololu Micro Serial Servo Controller ($20) = $45.  So I'm asking
whether there's something less than $45 that will do the same job.


Hmm... I'll look into that.  Thanks for the tip!

Best,
- Joe

You were clear. My response to you was not so clear!  :-)

What I should have said is "what do you want the microcontroller to do =
besides control 4 servos?". But, having said that, I can see where I =
also might have asked the exact same question the way you did --- and =
probably have on this same forum. My apologies. And good luck!

Oh, I see!  Nothing much.  Maybe run through a demo sequence of moves,
or listen to wireless control signals of some sort, and fire off the
corresponding sequence of servo motions.  Nothing that would tax any
microcontroller, I imagine, but more than a nonprogrammable servo
controller can do by itself.

Thanks,
- Joe

off of eBay for around $80 just to have a

robot I am developing. The BOE has 4

programmability for sensor inputs, etc. I

and knew it had the servo connections

Includes 2 continuous motion servos, also. The

exposure I can resell them back on eBay when my

solution.


You will notice that the Basic Stamps are *not* servo controllers. They
can send a pulse out to a servo, but then they're dead for doing
anything else until that pulse finishes. If you look inside the cpu
during that time, you'll see it's doing nothing except waiting for the
pulse to end. By time you get to running 4 servos, with 20-msec refresh
rate, you've used almost 50% of the cpu duty cycle.

That being said, I love the Stamps, but they're not too good for any
kind of multitasking. They will start to bog down trying to do
everything else alongside the servo pulses.

If I use an Atmel controller (like the Baby Orangutan), then maybe I
could control the servos myself.  Starting and stopping pulses shouldn't
be a big deal -- I can just set up an event queue, and probably manage
lots of servo outputs that way.

But, unless I really complexify my code, or use something like the RTOS
described in the latest SERVO mag, it's going to be a pain in the neck
to do anything else while managing this event queue.  Or rather, if I
start doing something else, and that something takes too long, then the
pulse trains could get screwed up, with unhappy results.  So it'd really
be nice to just hand off the desired servo position and speed to a
separate chip, in a "set and forget" sort of way.

Best,
- Joe

Joe:

You are slowly reinventing the modular approach to doing
things.  A dedicated microcontroller to keep the pulses
going out to 4 servos does not cost very much.  I'll point
you to my solution:

   <http://gramlich.net/projects/rb2/servo4/index.html>

Enjoy,

-Wayne

Well, not if I can help it -- I'd much rather buy something someone else
has invented!


Yes, but like the Pololu Micro SSC, this isn't programmable by itself
either; I'd need to combine it with something else that does the
"thinking."  Which, I agree, is modular, but less than ideal for this
project (a very small robot with tight space and cost constraints).

Best,
- Joe

Joe:

I can not begin to guess what all of your constraints are.
If you tell us the entire project, we might be able to give
you better advice.  If you list all of your constraints,
we can give better suggestions.  By the way, the word
"animatronics" is used to talk about systems where a bunch
of servos are run through a set of preprogrammed moves
(i.e. not much feedback.)  If you search for "animatronics
servo" using your favorite search engine, something might
turn up.

-Wayne

Well, I think I've listed them all: cost under $45, programmable,
controls at least 3 servos, and takes up as little space as possible.  
That's about it.  Seems like one could make a board that had one chip
(plus support stuff) to act as the programmable microcontroller, and
another chip to drive a handful of servos; if these were all
surface-mount parts, it should be pretty small and cheap.  Kinda like
what Baby Orangutan would be if the motor driver were replaced with a
servo driver.  That's what I'm looking for -- but it seems as though it
doesn't exist.


Thanks, that's a good idea!  A quick search seems to indicate that the
animatronics community uses serial servo controllers, generally with a
PC as the master controller.  Makes sense for them -- they don't need
portability.

However, that led to this very interesting comparison table:

  <http://www.oricomtech.com/svc-comp.htm>

It's not quite complete (omits the Pololu MSSC for example), but still
pretty handy.

Best,
- Joe

You have not been very specific about how small you need to
get.  I can easily imagine using two PIC16F688 chips in
DIP14 sockets to get the job done.  This could be made with
point-to-point wiring on a small piece of perferated board.
SMT would be smaller, but you would need to create a PCB
for the chips.

[snip animatronic stuff]

-Wayne

There are any of a number of free trutorials on writing AVR and PIC
based serial servo controllers. I think WG has provided one, but a quick
Google should show lots more. For example, this was the first hit for
the query 'serial servo controller':

http://www.rentron.com/SerialServo.htm

Since you're building the controller out a PIC, if the design has extra
room (memory, CPU time), you might be able to do everything in one very
inexpensive chip. Even if you add the $$ for a low-cost PIC programmer,
you're still ahead compared to $45 for the Pololu pair. I doubt you will
get any cheaper than that, as they have some of the lowest prices
around.

Most people do not want an "all-in-one" pre-programmable servo
controller, which is why you aren't finding many commercially-available
options. There are a few; look at Pontech, for example.

-- Gordon

Actually, that's why I built my Bot-Cop controller board. It has a
20-channel servo controller, with velocity-control, on the same pcb as
a 24-pin Stamp/etc module. Gives you general programmability in HLL
without incurring the overhead of generating pulses for 20 servos. The
idea worked out really well for my hexapod walker, BTW. The pcb is a
little large for Joe's app, from the sounds of it.

http://www.oricomtech.com/prod2/bcop-svo.htm
http://www.oricomtech.com/projects/nico6prj.htm

Thanks Dan -- that's pretty neat.  The size doesn't look unreasonable
(which is to say, I can probably shoehorn it in if I'm sufficiently
motivated).

I really like the way you made it so it can run as a normal slave module
if no 24-pin controller is installed.  That offers a nice smooth upgrade
path -- start without it, then when you want autonomous operation, buy
one of those modules and plug it in.

Of course I suppose I get the same effect by starting with a Pololu MSSC
(which I already have) and then adding a controller like the Baby
Orangutan.  But then I have to rig up my own connection between the
boards, and my own DB-9 connector adapter, etc.  You've got a leg up in
terms of plug-and-playness.

It's certainly a neat option -- thanks for pointing it out!

Best,
- Joe

    Parallax has a new chip out called the "Propeller".    It  has eight 32
bit processors (COG's) on one chip.  With eight processors you can get some
true multitasking done.  Also,  the price is nice and like the Stamp chips
the development software is free.  I'm looking to use this as the controller
chip on a robot project.
    See link => http://www.parallax.com/propeller/index.asp

Steven

And very recently (last weekend) geeks.com had the full BoeBot serial kit
on sale for $59 each.  I picked up 4 of them for much the same reason you
describe, the parts alone are certainly worth more than that, but I wanted
the BoeBot as an intro into robotics for myself anyway.  It is my first
robot.

As I write this my BoeBot is wandering autonomously around the house,
looking for three specific IR sources (each source has a unique PRF), the
sources must be found in a certain order and a specific "button" pressed to
kill each source source.  The next source is not a valid target until the
source before it is detected, sought out, and shut down.

Right now the search is random, and I am making no attempt to cover
ground only one time, but I have a couple of ideas to try for that in the
next couple weeks or so.

C!