Going to tackle adding an actual coastdown test to MPGuino

[t vago]

I am going to tackle adding an actual coastdown test program into MPGuino, not just allowing it to spit out datalogging numbers.

[MetroMPG]

Subscribed!

[AndrzejM]

Subscribed!

[t vago]

Notes from earlier, to help with research

[iveyjh]

Subscribed

[t vago]

Going to have to add an air density variable, which means that the user will have to figure that density out when he/she does a coastdown run set. Was tempted to include "locust density" as a nod to dcb...

Going to also have to include the viscosity component for the solver. This will be the equation I will use inside the solver ->

F(drag) = F(rr) + F(viscosity) + F(aero)

F(rr) = m * g * C(rr)

F(aero) = (rho) * C(d) * A * (v^2) / 2

F(viscosity) = C(viscosity) * v

F(viscosity) is not very large, but it is significant, and it will have an effect on calculating C(d) if it is ignored.

[t vago]

The solver will utilize matrix mathematics to solve for C(rr), C(v), and C(d).

A flat surface will still be required, and of course the vehicle to be tested will have to be put into neutral prior to starting the test. However, if my calculations are correct, then only one pass will be needed in order to determine values for the above coefficients, and that pass will require 1.5 seconds to collect all of the data it would need to calculate the above coefficients. It won't matter too much how fast (within reason) the vehicle will be travelling, either, as long as the road being used is reasonably flat and smooth and level. I think that a mid-range speed (around 30 MPH) should give the most accurate results.

It's not to say that only one pass will be needed. Several passes should be performed in order to get a good average for the three coefficients above.

[t vago]

Ugh...

In order to be able to provide the amount of precision necessary to generate a solution, given the raw data collected by MPGuino, I will state that those people who are running MPGuino hardware with the AtMega168, will very likely not be able to run this coastdown tester. This code may require using the AtMega328 or better, due to code size. We shall, see, though.

If I can modify my 64-bit bytecode processor (SWEET64) to operate on the matrices that are necessary for solving for the coefficients, then AtMega168 users may still be able to use this.

[t vago]

After dusting off this thread, I've been at work on it for the past day. I got it to the point where it will collect the raw data. Now, trying to figure out how to get SWEET64 to do the actual matrix math.