Floating point math

[Art]

Though this is fairly old, right now I’m writing C code for dspic33FJ64GP just to do floating point math
where I’m still more proficient with the rest of the program to use a 16F628A in the same circuit.
Two chips have to talk some serial with each other. Quite crazy since then the board requires
two regulators for two supply voltages and level conversion for connections between the 3.3 Volt dspic and any 5 Volt pic or peripherals.

The dspic is really only for some numbers go in, it does it’s sin, cos, tan, and some different numbers come out.

[pedja089]

Why you need FP for trig function?
You can use lookup table and linear interpolation to get 3-4 decimal places resolution.
Melanie already done that
http://www.picbasic.co.uk/forum/show...=2873#post2873
And just found this
http://www.picbasic.co.uk/forum/cont...h-PicBasic-Pro
So I think that you can do all that in one pic.
And you can use N bith math, or FP rutines, is long isn't enough... And it must be faster to do that in one PIC, than to sent and receive data thru UART.

[Dave]

Hey Art, Being you like to work with the 16F628's you might have a look at using one of the 16F1847's to replace it. 8K of flash, 1K of ram, 256 bytes of eeprom and 32Mhz. I have replaced a few 16F628's from old projects and they work a treet...

[Art]

I did give a a lot of thought of getting it into one chip, and it would definitely be the dspic
since I already have implemented the functions I want into another program for the same chip.
This is just the first time I’ve had to do the hardware. The last one was a kit web server.

Since I want the LCD to be continually driven, and am more apt with fancy graphics with an 8 bit pic,
that’s the reason to consider the second chip.
If I was happy with a fairly static LCD display, it would be easy with the dspic alone.
The algorithms are Sunrise/set, Moonrise/Set, and Moon Illumination for a clock.