UPGRADE YOUR BROWSER

We have detected your current browser version is not the latest one. Xilinx.com uses the latest web technologies to bring you the best online experience possible. Please upgrade to a Xilinx.com supported browser:Chrome, Firefox, Internet Explorer 11, Safari. Thank you!

cancel
Showing results for 
Search instead for 
Did you mean: 
Visitor vinodlk123
Visitor
6,980 Views
Registered: ‎01-27-2012

FFT CORE INCORRECT RESULTS

Hello....

Im using logic core fft 7.1 in   xilinx-12.1,  configured it  for

burst,

fixed point,

datawidth 32-phase factor(PF) 24,

scaled mode, convergent rounding,

block RAM for data , PF....

when i give inputs in 2's comp format.. output is nowhere the expected output....

i don't know how to solve this issue...

help me in using correct results from IP core

 

thanks...

0 Kudos
8 Replies
Teacher rcingham
Teacher
6,978 Views
Registered: ‎09-09-2010

Re: FFT CORE INCORRECT RESULTS

Are these errors shown in simulation?

------------------------------------------
"If it don't work in simulation, it won't work on the board."
0 Kudos
Xilinx Employee
Xilinx Employee
6,973 Views
Registered: ‎08-02-2011

Re: FFT CORE INCORRECT RESULTS

You need to show some data if you expect us to be able to help.

 

Sim plots, .xco file, instantiation code, testbench, projects, etc...

www.xilinx.com
0 Kudos
Visitor vinodlk123
Visitor
6,966 Views
Registered: ‎01-27-2012

Re: FFT CORE INCORRECT RESULTS

input-8 FFT.JPG

 

output- 8 FFT.JPG

yes sir...

 im doing simulation...its sowing incorrect outputs..

 

i have to do 256 point FFT... but when i tried for 8 point FFT core to know their usage, its giving incorrect results...

i'v attached both input & output simulation results...

 

 

plz help me to know where i'v done wrong...

 

in results all  the notations are in unsigned decimal.

 

thank you..

 

 

 

 

0 Kudos
Visitor vinodlk123
Visitor
6,937 Views
Registered: ‎01-27-2012

FFT CORE 7.1 outputs not understandable----- need attention....

logic core 7.1 FFT core outputs are not correct, or i may be wrong in configuring it properly... i'l specify the config i'v done for IP.

fixed point, inputs are 2's comp inputs.

n=8, R-2, Burest IO

data width 32, Phase factor 24.

scaled, convergent rounding, natural order without cycle prefix insertion.

memory block ram, complex multiplier- 3 multiplier structure,

butterfly arithmetic , use CLB logic.

 

scale=[111111]

 

i'v attached outputs of simulation for forward transform below..

 

plz help me in using IP core properly .....

inputs are xn_re=[1  2 3 4 5 6 7 8 ], xn_im=[0 0 0 0 0 0 0 0 ], (unsigned decimal notation in simulation results)

outputs got from simulation are

xk_re=[0, 429496, 429496, 4294961 , 0 ,22 , 23 , 0 ]

xk_im=[0 , 429496 ,9 , 0 , 0 , 23 , 55, 0]

 

outputs are nowhere near the expected outputs....

whoever knowledged in this IP usage plz help me...

 

 

input-8 FFT.JPG

 

output.JPG

0 Kudos
Visitor vinodlk123
Visitor
6,932 Views
Registered: ‎01-27-2012

Re: FFT CORE INCORRECT RESULTS

sir ,  i have attached the simulation results....

please go through them & help in getting correct outputs...

0 Kudos
Xilinx Employee
Xilinx Employee
6,920 Views
Registered: ‎08-02-2011

Re: FFT CORE INCORRECT RESULTS

Thank you for posting this info.

 

So what are you comparing these results against to determine that they are incorrect?

www.xilinx.com
0 Kudos
Visitor vinodlk123
Visitor
6,916 Views
Registered: ‎01-27-2012

Re: FFT CORE INCORRECT RESULTS

im unable to get correct outputs in 2's comp format...

output is nowhere near the expected output..

i'v applied xn_re=[1 2 3 4 5 6 7 8]

xn_im=[0 0 0 0 0 0 0 0]

but outputs are not even near the approximation...

also when  give constant xn_re with xn_im as zero, than outputs are all zeroes including xk_re..

 

i may be wrong in configuring or somewhere im doing wrong,

help me to do this in right way & get correct results....

thank you...

0 Kudos
Xilinx Employee
Xilinx Employee
6,911 Views
Registered: ‎11-28-2007

Re: FFT CORE INCORRECT RESULTS

Your input is too small (you only use the 4 LSbs out of 32-bit input) and the scaling schedule is too big. Try the following values and let's know if the output looks better.

 

xn_re = [32'h1000_0000, 32'h2000_0000,... 32'h8000_0000]

xn_im = [0 ... 0];

scale_sch = 010101;

 

 


@vinodlk123 wrote:

im unable to get correct outputs in 2's comp format...

output is nowhere near the expected output..

i'v applied xn_re=[1 2 3 4 5 6 7 8]

xn_im=[0 0 0 0 0 0 0 0]

but outputs are not even near the approximation...

also when  give constant xn_re with xn_im as zero, than outputs are all zeroes including xk_re..

 

i may be wrong in configuring or somewhere im doing wrong,

help me to do this in right way & get correct results....

thank you...




Cheers,
Jim
0 Kudos