On Nov 13, 4:47*pm, Tim Wescott wrote:
On Tue, 13 Nov 2012 13:14:02 -0800, rangerssuck wrote:
On Tuesday, November 13, 2012 2:28:05 PM UTC-5, Tim Wescott wrote:
OK, here it is again with more pictures and explanations. *Sorry for

starting a new thread, but it seemed reasonable given that I did such a

poor job of getting the design problem stated in the last one.

I'm building a gizmo, which, for lack of a better name I'm calling a
"fan

trainer". *It consists of an arm about 30" long with a small DC motor
and

propeller on one end, and a pivot and counterweight on the other. *The

thing is pivoted on a frame which has a controller circuit board. *The

controller monitors the motor current and the arm position, and drives

the motor voltage.

This isn't an executive toy that you sit back and watch move (that
would

get old very fast). *Rather, it is a platform to provide a series of

exercises for a student to tune the control system, and to do it in a
way

that you can see and feel it working. *I'd use it in conjunction with

seminars, and perhaps make it part of a "seminar at home" package.

So my larger problem is to manufacture this in small quantities (20 to

100 a year), and sell them at a reasonable price without losing my ass..

"Not losing my ass" translates to a total bill-of-materials cost of $50

or so, and if I could get it way lower then I'd just sell it for less
and

increase my potential market.

My immediate problem is that the potentiometer that senses the arm

position is getting punched off of the board in shipping. *The

potentiometer has a hole, through which you pass a 4mm shaft with a
flat

milled into it. *If you mill the flat so that you can easily pass the

shaft through the pot then the control is not smooth -- the arm hunts

across the rotational slop caused by the shaft rotating within the pot..

If you shim the flat for a snug fit (very very light press fit?) then

when you chuck the thing into a box and fly with it in checked baggage,

the pot gets punched off of the board, apparently by getting hit from

behind.

The shaft on which the arm pivots is restrained in the housing by a

couple of model airplane wheel collars. *On the trip out, both of these

collars, and the arm, loosened on the shaft, and the pot was punched

out. *On the return trip I loosened the arm and tied it to the back
plate

of the frame -- the wheel collars were fine then, but the pot was still

punched out.

I really like that pot: it costs less than $2.00 for onsies (less,

obviously, in higher quantities), it has undetectably small friction,
it

isn't noisy, and because it's board mounted it saves me from needing a

bunch of brackets which would just drive up my BOM cost. *So any

alternative that involves not using the pot has to compete with that

price, and being practical to do in small quantities in an environment

where labor is not free.

So I'm thinking at this point that perhaps I just need to be happy with

what I have, and to warn people not to drop the thing off of a table or

to ship it without disassembling it first. *But it would be nice if
there

was a way to make it more robust (by isolating the pot from the shaft

somehow).

The best suggestion I got from the other thread, assuming that I can do

it cheaply, is to put a slit in the end of the shaft, so I basically
have

a D-shaped shaft with some spring. *I would like that idea a lot if I

knew what it would cost to have a batch of 20 shafts made with the
slit,

vs. without, and if that cost wasn't much greater than just making the

shafts.

The second-best suggestion is to use a flat coupling. *This would
require

(essentially) two shafts and the coupling, which is clearly going to

drive up the BOM cost, but I'm still toying with how to make it cheap.

Several people suggested reinforcing the mounting of the pot to the

board: it probably doesn't show in the pictures, but it is clear from
the

construction of the thing that this would just result in _part_ of the

now-destroyed pot remaining on the board.

Here's a general arrangement shot:

https://docs.google.com/open?id=0B5l...0MxMEM3b1l2c0k

And the thing in action:

https://docs.google.com/file/d/0B5lS...0MxMEM3b1l2c0k

And, finally, a close-up of the potentiometer in question:

https://docs.google.com/open?id=0B5l...TBCc3VzWDYxZ0E

--

My liberal friends think I'm a conservative kook.

My conservative friends think I'm a liberal kook.

Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software

http://www.wescottdesign.com

I think you ought to focus on *why* the wheel collars loosened. I am
assuming that had they stayed secure on the shaft that the pot would
have survived. If that is a correct assumption, how about a dimple in
the shaft for the setscrew, or perhaps some loctite?

This is a pretty neat training gizmo for control loops. Sort of like the
ball balancer, but better.

I considered doing a ball balancer, but I figured this would cost less.

For that matter, I really wanted to do an inverted pendulum -- but I
couldn't see how to do the mechanism cheaply.

The wheel collars stayed tight on the flight home, yet the pot was still
punched off the board. *Not by as far -- but all leads broken is still
all leads broken. *I suspect that any sort of collar that rides close to
the board to limit travel in that direction will either rub or have too
much play to be safe -- but I could be wrong.

At this point what I see is a choice between some rotational slop between
shaft and pot (which messes up the educational value of the thing), the
current setup (which leaves it fragile, but possibly manageably so), some
sort of a flex coupling or spring (assuming I can figure out how to do
that well and cheaply), or some sort of a spring-loaded means of holding
the shaft-pot joint to be snug in rotation, but still low friction to
axial motion.

--
Tim Wescott
Control system and signal processing consultingwww.wescottdesign.com- Hide quoted text -

- Show quoted text -

How about a thermal thing for control loop education?

TEC, air fin heatsink, heater as variable load, maybe a couple of
sensors. It'd be nice to have variable thermal 'lengths' in the
system.
(More than $50 though I'd guess.)

George H.