cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Highlighted
Adventurer
Adventurer
11,239 Views
Registered: ‎10-18-2012

using FIR ip core as lowpass filter..?

hii group members, in my project i'm using fir(v6.3) ip core, for this ip core I generated coefficient file from matlab fda tool.In this .coe i fixed the stop band frequency limit to 2MHz, but when i'm giving 10MHz signal as the i/p to the fir, it processing the same 10MHz signal to o/p. I'm very much stuck in this error,please help me to fix this problem. 

Tags (1)
0 Kudos
13 Replies
Highlighted
Xilinx Employee
Xilinx Employee
11,234 Views
Registered: ‎08-02-2011

Are your clock/sampling rates correct on the core?

 

Can you post your .xco and/or .coe? It would also be helpful to see a screenshot or something of the observed behavior.

www.xilinx.com
0 Kudos
Highlighted
Professor
Professor
11,233 Views
Registered: ‎08-14-2007

A FIR low-pass filter will reduce the amplitude of frequencies greater than the cut-off, but

it won't filter them out completely.  Are you saying that your filter did not reduce the amplitude

of the 10 MHz input signal?

 

-- Gabor

-- Gabor
0 Kudos
Highlighted
Adventurer
Adventurer
11,216 Views
Registered: ‎10-18-2012

thank you for replying. yes the clock/sampling rates are same, here i'm posting my .coe file in fir ip core n fda tool screen shots. please let me know is there any solution, did u require any other info...?

fir ipfir frm fdatool

0 Kudos
Highlighted
Adventurer
Adventurer
11,203 Views
Registered: ‎10-18-2012

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?

0 Kudos
Highlighted
Professor
Professor
11,192 Views
Registered: ‎08-14-2007


@azay wrote:

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?


Not exactly.  If you look at the second image you posted, the amplitude seems to bounce between about

-66 dB and -80 dB in a comb-like pattern once you're above the cut-off.  In any case the amplitude should

be much less that the amplitude of the input signal, unless you also have gain in the filter.  You might want

to try a 1 MHz stimulus to see what amplitude comes through in the passband, and then compare that

to the 10 MHz stimulus.  If you don't see a difference between the two, then there's something wrong

with the filter.

 

-- Gabor

-- Gabor
0 Kudos
Highlighted
Adventurer
Adventurer
11,175 Views
Registered: ‎10-18-2012


@gszakacs wrote:

@azay wrote:

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?


Not exactly.  If you look at the second image you posted, the amplitude seems to bounce between about

-66 dB and -80 dB in a comb-like pattern once you're above the cut-off.  In any case the amplitude should

be much less that the amplitude of the input signal, unless you also have gain in the filter.  You might want

to try a 1 MHz stimulus to see what amplitude comes through in the passband, and then compare that

to the 10 MHz stimulus.  If you don't see a difference between the two, then there's something wrong

with the filter.

 

-- Gabor



@gszakacs wrote:

@azay wrote:

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?


Not exactly.  If you look at the second image you posted, the amplitude seems to bounce between about

-66 dB and -80 dB in a comb-like pattern once you're above the cut-off.  In any case the amplitude should

be much less that the amplitude of the input signal, unless you also have gain in the filter.  You might want

to try a 1 MHz stimulus to see what amplitude comes through in the passband, and then compare that

to the 10 MHz stimulus.  If you don't see a difference between the two, then there's something wrong

with the filter.

 

-- Gabor


once again thanking you, i tried for different combinations please have a look at them. here we go by different cases,

 

 

1)for 1MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>370mv

 

2)for 2MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

3)for 3MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

4)for 4MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

5)for 19MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.46V

o/p>>>400mv

 

6)for 19MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V 

o/p>>>260mv 

 

what's your analysis on these results? i have dought in the first result, i think o/p amplitude from 1Mhz signal must be greater than the 2mhz signal value. And i/p amplitude values for all frequencies are coming same so is there any error in measurement..? 

0 Kudos
Highlighted
Xilinx Employee
Xilinx Employee
11,157 Views
Registered: ‎11-28-2007

Are the results from simulation or hardward?

 


@azay wrote:

@gszakacs wrote:

@azay wrote:

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?


Not exactly.  If you look at the second image you posted, the amplitude seems to bounce between about

-66 dB and -80 dB in a comb-like pattern once you're above the cut-off.  In any case the amplitude should

be much less that the amplitude of the input signal, unless you also have gain in the filter.  You might want

to try a 1 MHz stimulus to see what amplitude comes through in the passband, and then compare that

to the 10 MHz stimulus.  If you don't see a difference between the two, then there's something wrong

with the filter.

 

-- Gabor



@gszakacs wrote:

@azay wrote:

thanks for your reply, i'm trying it. So the amlitude for all frequencies greater than cut off should have same amplitude,    isn't it?


Not exactly.  If you look at the second image you posted, the amplitude seems to bounce between about

-66 dB and -80 dB in a comb-like pattern once you're above the cut-off.  In any case the amplitude should

be much less that the amplitude of the input signal, unless you also have gain in the filter.  You might want

to try a 1 MHz stimulus to see what amplitude comes through in the passband, and then compare that

to the 10 MHz stimulus.  If you don't see a difference between the two, then there's something wrong

with the filter.

 

-- Gabor


once again thanking you, i tried for different combinations please have a look at them. here we go by different cases,

 

 

1)for 1MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>370mv

 

2)for 2MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

3)for 3MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

4)for 4MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V

o/p>>>400mv

 

5)for 19MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.46V

o/p>>>400mv

 

6)for 19MHz i/p signal with 2Mhz cutoff freq. the i/p n o/p amplitudesof FIR LOWPASS filter are

i/p>>>1.02V 

o/p>>>260mv 

 

what's your analysis on these results? i have dought in the first result, i think o/p amplitude from 1Mhz signal must be greater than the 2mhz signal value. And i/p amplitude values for all frequencies are coming same so is there any error in measurement..? 




Cheers,
Jim
0 Kudos
Highlighted
Adventurer
Adventurer
11,150 Views
Registered: ‎10-18-2012

the results are from hardware

0 Kudos
Highlighted
Professor
Professor
11,140 Views
Registered: ‎08-14-2007


@azay wrote:

the results are from hardware


Then the first thing I'd check is that you are not saturating the A/D and or the D/A converters.  ChipScope

can show you a lot in this case.  Did you simulate the code before you tried it in hardware?

 

-- Gabor

-- Gabor
0 Kudos
Highlighted
Adventurer
Adventurer
5,514 Views
Registered: ‎10-18-2012

i'm not simulated this code, i will try it and post u the result.,,thank you very much

0 Kudos
Highlighted
Adventurer
Adventurer
5,506 Views
Registered: ‎10-18-2012

hii ,while i'm trying to generate chiscope(.cdc) file, the code was showing error in mapping. 

0 Kudos
Highlighted
Adventurer
Adventurer
5,458 Views
Registered: ‎10-18-2012

 

 

This is the error chipscope showing

---------------------------------------------------------------------------------------------------------------------------------------

Writing output files...
ERROR:PhysDesignRules:1683 - Unsupported programming for BSCAN block and
JTAG_CHAIN attribute value 1. The BSCAN component
icon_inst/U0/U_ICON/I_YES_BSCAN.U_BS/I_K7.ISYN.I_USE_SOFTBSCAN_EQ0.U_BS has
placement which requires the JTAG_CHAIN attribute to be set to the value 2.
The JTAG_CHAIN attribute must be changed or the programming for the BSCAN
block must be moved to a location corresponding to the JTAG_CHAIN attribute
setting.
ERROR:Pack:1642 - Errors in physical DRC.

-----------------------------------------------------------------------------------------------------------------------------------------------------

0 Kudos
Highlighted
Professor
Professor
5,448 Views
Registered: ‎08-14-2007

It looks like you already have another core (microblaze with debug?) that uses the boundary

scan logic.  You need to go back and re-customize the ICON to use JTAG chain 2, as indicated

in the error message.

-- Gabor
0 Kudos