## Re: [stella] TIA Audio Polynomials

 Subject: Re: [stella] TIA Audio Polynomials From: Adam Wozniak Date: Fri, 21 Nov 2003 13:26:36 -0800 (PST)
```Ok, here's some clues.

Mode 3:
:00000000001100011011110111111000000000011000011010000111110
000000001110001111010111111110000011111000001111010111111110
000110000000011100010111111100001100000001111101110111111100
001000011111111001000111100000001000111111100011011111100000
011001111000001110011111100000010001110011111000111111100011
110001100110000011111111100010000011101100111111111110000010
000011101001111111111110000010011111001011111111111100000010
0111110110111111111100000011100110000100111111
[465 465]

The tia manual says "5 bit poly --> 4 bit poly"
note that 465 == 31*15
There's probably some way to chain the two 4 and 5 bit polynomials
together to get that same bit pattern.  Suggestions are welcome.

Mode 15:
:11111111110000011100000001111000000000011111100011111100001
1111111100000011111000000111100000
[68 68 67 66 65 63 62 37 36 35 34 32 31 6 5 4 3 1]

note the pattern in the coefficients:
(0 is always assumed to be a coefficient)
0+31=31   31+31=62
1+31=32   32+31=63
3+31=34   34+31=65
4+31=35   35+31=66
5+31=36   36+31=67
6+31=37   37+31=68

Again, suggestions on how to simplify this are welcome.

Taking a step back and looking at it, I'm probably the only one who
understands this.  Probably my responsibility to turn it all into C code
then. hehehe.

I could live with the 68 bit pattern in an FPGA implementation, but
the 465 bit pattern seems like a bit much.  There's gotta be a way to
simplify that one.  Suggestions?

--