Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

- Community Forums
- :
- Forums
- :
- Hardware Development
- :
- AI Engine, DSP IP and Tools
- :
- Re: Why the output of my FFT_IFFT model is incorre...

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page

swain168

Visitor

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-22-2009 05:42 AM - edited 07-22-2009 05:55 AM

9,173 Views

Registered:
11-18-2008

input signal is cos(2*pi/192)(real component) and sin(2*pi/192)(image component)

Simple time is 3,

Simulink system period is 1s,

Down sampling rate is 3,

FFT bolck is FFT v4_1

Message Edited by swain168 on 07-22-2009 05:55 AM

1 Solution

Accepted Solutions

aarunya

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

05-09-2010 06:41 PM

6,084 Views

Registered:
04-13-2010

um hi,

i have actually switched updated my xilinx sofware ISE 11 to get Virtex6 support.

in that i have used the FFT v7.0. i have made a few corrections and both FFT ant IFFT have started working

1. i was using the run time configurable transform length.

-the nfft_we should be high for only one clock cycle during which the transform length Nfft is being configured. if Nfft_we is tied to high, it keeps on trying to configure the block and hence the processing never starts.

-the input data to the FFT block has to be sent with one clock delay after the dynamic transform length is set.

2. though, the dynamic transform length is being used, the transform length that can be chosen in block parameters from the drop down list has to be set at the maximum that u may be using. the dynamic length can be for setting any size smaller than the transform length in the block parameters only.

with these 2 modifications, the FFT started to work.

3. for the IFFT, i have noticed that the output of FFT seems to saturate the IFFT and hence was not giving me results. so i did the 1/N scaling after the FFT and havethen passed the data to IFFT. due to this scaling, the IFFT also started working

Aarunya

13 Replies

swain168

Visitor

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-22-2009 06:01 AM

9,175 Views

Registered:
11-18-2008

The output of 1024 point FFT

swain168

Visitor

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

07-22-2009 06:08 AM

9,165 Views

Registered:
11-18-2008

The output of 1024 points IFFT

Could you help me ?

pjayav

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

02-07-2010 07:45 PM

8,538 Views

Registered:
11-11-2009

hi,

me too having the same problem. Did u got solution for your problem. If so, please share with me

Thanks & regards

selvam

aarunya

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-26-2010 08:37 PM

8,323 Views

Registered:
04-13-2010

I am having that problem too.

I have also noticed that my FFT output from Matlab and output from Sysgen do not match properly.

aarunya

syoma

Adventurer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-27-2010 01:39 AM

8,316 Views

Registered:
08-20-2007

ywu

Xilinx Employee

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-27-2010 08:17 PM

8,304 Views

Registered:
11-28-2007

If you are asking about IFFT, please take a look at this blog: http://myfpgablog.blogspot.com/2009/11/ifft-in-system-generator.html

aarunya wrote:I am having that problem too.

I have also noticed that my FFT output from Matlab and output from Sysgen do not match properly.

aarunya

Cheers,

Jim

Jim

aarunya

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-29-2010 08:14 PM

8,278 Views

Registered:
04-13-2010

thank you syoma. The FFT is solved. but my IFFT is not solved.

I'll look into the blog which Jim is refering to and if i can't solve, I'll get back.

thank you.

aarunya

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-29-2010 08:15 PM

8,277 Views

Registered:
04-13-2010

Thank you Jim.

I'll check that

farshade

Visitor

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

05-02-2010 09:48 PM

8,248 Views

Registered:
12-20-2009

@aarunya wrote:

thank you syoma. The FFT is solved. but my IFFT is not solved.

I'll look into the blog which Jim is refering to and if i can't solve, I'll get back.

thank you.

Can you please tell me how you solved the mismatching outputs for FFT between MATLAB and SysGen?

I appreciate if you attach the working .mdl file.

Thanks a lot.

aarunya

Observer

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

05-09-2010 06:41 PM

6,085 Views

Registered:
04-13-2010

um hi,

i have actually switched updated my xilinx sofware ISE 11 to get Virtex6 support.

in that i have used the FFT v7.0. i have made a few corrections and both FFT ant IFFT have started working

1. i was using the run time configurable transform length.

-the nfft_we should be high for only one clock cycle during which the transform length Nfft is being configured. if Nfft_we is tied to high, it keeps on trying to configure the block and hence the processing never starts.

-the input data to the FFT block has to be sent with one clock delay after the dynamic transform length is set.

2. though, the dynamic transform length is being used, the transform length that can be chosen in block parameters from the drop down list has to be set at the maximum that u may be using. the dynamic length can be for setting any size smaller than the transform length in the block parameters only.

with these 2 modifications, the FFT started to work.

3. for the IFFT, i have noticed that the output of FFT seems to saturate the IFFT and hence was not giving me results. so i did the 1/N scaling after the FFT and havethen passed the data to IFFT. due to this scaling, the IFFT also started working

Aarunya

akilan

Newbie

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

05-18-2010 09:38 AM

5,030 Views

Registered:
04-15-2010

Thanks a lot for the solution

king1

Visitor

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

12-11-2011 08:15 PM - edited 12-11-2011 08:16 PM

4,417 Views

Registered:
12-10-2011

Hi Aarunya,

Can you please mail me the solution for FFT_IFFT model. I require it urgently. My email id is vikrantmahajan.iitkgp@gmail.com.

Thanx in advance.

ywu

Xilinx Employee

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

12-12-2011 04:30 AM

4,410 Views

Registered:
11-28-2007

Check http://forums.xilinx.com/t5/DSP-Tools/FFT-IFFT-famous-problem/m-p/193978#M4856

@king1 wrote:

Hi Aarunya,

Can you please mail me the solution for FFT_IFFT model. I require it urgently. My email id is vikrantmahajan.iitkgp@gmail.com.

Thanx in advance.

Cheers,

Jim

Jim