Feedback in FM synthesis

[Martin Czech]

:::and what about this: 
:::
:::(I implemented it this way) 
:::
:::e=Ac*sin((Fc+sin(Fm*t)*Am)*t) 
:::
:::One thing I noticed is that it creates very dull sounds, not bright.
:::There are no sharp peaks. 

Formally:
The given formula is not correct (physically).
Fc*t gives 1/s*s, no unit, which is ok.
but there remains t*sin() which is of the unit of time (s) ?!


Next: for real FM with non static Fm we would have to compute

e=Ac*sin[Fc(t)*t+Am(t)*INT{sin(Fm(t)*t)}dt]

For discrete computation the integral INT{sin(Fm(t)*t)}dt
will translate into accumulate/multiply, you know, the way
we learned integration by filling areas with little squares...

also Fc*t etc. will preferably translate into accumulate/add,
if the time step delta t is constant. So we don't need to multiply.

That is the way I stumbled into this FM/PM thing, when
I made a program error that interchanged some accumulate/add
steps I suddendly had FM instead PM!

m.c.