[quote user="Doug"] I tried long double and got a 64 bit var.[/quote]
The important point is that sin is still being done with double types, which are 32-bits.
[quote user="Doug"]The sin() function returns a double which I believe is 64 bits ?[/quote]
It does return a double. But double is 32-bits on C2000.
[quote user="Doug"]Does it need to do the conversion for long double which in this case is 64 bits?[/quote]
Yes, but it does that conversion after computing sin.
[quote user="Doug"]It's small details but I was surprised to see the 16 bit MSP430 more precise than the 32 bit TMS320F28027[/quote]
On MSP430, provided you use the newer EABI, the type of double, and thus the type used to compute sin, is 64-bits.
Thanks and regards,
-George