[rescue] LCD monitor diagnosis

[Don Y]

Phil Stracchino wrote:
> For that matter, I've long thought it's time we had cars with ACCURATE
> fuel guages.  And it's easy to do:  Instead of mounting the tank on dumb
> bolts with a float and sender in the tank, chuck the sender and hang the
> tank on four loadcells.  Monitor the output from the load cells, say,

Yeah, but I think that would be a tricky mount.  The load cells
want to see their "loads" in specific orientations, etc.  I'm
not sure you could "measure" the tilt in order to be able to
accurately compensate for it.  E.g., even a small tank is
120+ pounds of fuel.  And you would need to be accurate to
within 4 or 5 oz (to get 1 mi at 25 MPG resolution).

> twice a second.  Average the results, correct for tilt of the vehicle,
> keep a 60-second buffer, and have the guage display a rolling average of
> that 60-second buffer.  If any cell reads differently from the other
> three by more than twice the expected margin or error for, say, twenty
> consecutive readings, drop it from the pool and flag a warning.
> 
> This ought to be able to keep track of how much fuel is in the tank to
> plus or minus a few ounces of fuel, not plus or minus two or three
> gallons (which seems to be the best the current eighty-year-old guage
> technology is capable of).

I cringe to think what my gas gauge's concept of reality might be.
It is incredibly nonlinear (aren;t they all?).  And, I never know
where bottom will be.  So, I use my odometer as a gas gauge.
<shrug>  For the little driving that I do, it works good enough
(and an empty fuel container in the trunk in case I need to
hoof it to a nearby gas station!  :> )

> As an added bonus, you could even track how much fuel the loadcells say
> you have against how fast the engine computer says you're using it.  If
> the fuel level in the tank is going down faster than you know you're
> using fuel, issue a fuel-leak warning.