|
Message
From: Tom Hawkins<tom@c...>
Date: Fri Dec 10 04:37:28 CET 2004
Subject: [oc] Enquiries on Tom's CF FIR
wanderwonderus@y... wrote: > dear all, > > i've read thru Tom's CF FIR design using the Confluence software. > Anyone knows whether the VHDL coding is suitable for any fixed point > and sign input or it's just limited to the unsign binary? > > Besides,can i have further details on the vhdl coding for example the > procedure of the coding?I will be much appreciated if you could provide > further reading material on the design.
Hi Esther,
The core was generated with the following Confluence design. The syntax is a bit out of date, but at only 20 lines of code, it would take all of 5 minutes it bring it inline with the latest version of Confluence. James Gilb has a good table in the documentation comparing pre and post 0.9 syntax.
Looking at the MultiplierBank, it appears I used unsigned multipliers. Drop in '*+' if you need signed. Or better yet, make the multiplier operation an input to the FirFilter, so it can be parameterized.
For the latest Confluence compiler:
http://www.confluent.org/
-Tom
component FirFilter +FilterReset +Coefficients +FilterInput -FilterOutput with DelayTaps MultResults is {Reset FilterReset {DelayBank (length Coefficients) FilterInput DelayTaps}} {MultiplierBank Coefficients DelayTaps MultResults} {AdderTree MultResults FilterOutput} end
component DelayBank +NumberOfTaps +BankInput -BankOutputs is if NumberOfTaps == 0 BankOutputs <- [] else with NextTap is {Delay 1 BankInput NextTap} BankOutputs <- NextTap :: {DelayBank (NumberOfTaps - 1) NextTap $} end end
component MultiplierBank +SignalsA +SignalsB -Results is {Map2 (comp +A +B -X is {Delay 1 (A '*' B) X} end) SignalsA SignalsB Results} end
component AdderTree +Signals -Result with BinOp UnOp is component BinOp +A +B -X is {Delay 1 (('msb' A '++' A) '+' ('msb' B '++' B)) X} end component UnOp +A -X is {Delay 1 ('msb' A '++' A) X} end {Tree BinOp UnOp Signals Result} end
> > Thanks, > Esther > _______________________________________________ > http://www.opencores.org/mailman/listinfo/cores > >
|
 |