Amp + Filter sensor circuits

[Tim Vanderhoek <vanderh@ecf.utoronto.ca>]

[This message is mostly for Joyce].

On Tue, Jan 18, 2000 at 12:26:19AM -0500, Tim Vanderhoek wrote:
> 
>                        (we have to figure-out a more efficient way to
> use A/D converters for that to happen).

FWIW, I may have a solution to this.  Assume N IR sensors.  Rather
than build N complete IR amplification + filter units, build one
amplification + filter unit.  Connect this to an A/D and then to the
computer.  Total cost: A/D + several opamps.  For each sensor, one
opamp is still required.  Total cost: N * opamp.  The output from each
opamp connects to the single amp+filter unit.  The computer turns the
opamps on/off as necessary.  Total cost: 2*N*little-triangle-thingys.  

I like this solution for its simplicity.  We can quickly wire a dozen
sensors.

Cons: I don't know how much noise an opamp produces.  Completely
turning an opamp off may require something more complex than little
triangle thingys connected from the computer.

Pros: May be more reliable, since the same amp+filter circuit is being
used for each sensor, rather than different amp+filter circuits some
of which may be better, some of which may be worse than others.  Also,
being able to quickly wire sensors into the robot will be
advantageous, since some previous groups seems to indicate that this
is a time-consuming process.

We'll (Joyce and I) need to spend some time one night discussing
the feasibility of this.  The nice thing is: it's non-committal.
All of the parts required to implement this solution are required
to implement any solution (unlike multiplexers, which are only
required for certain solutions).


-- 
Signature withheld by request of author.