about fft ip core

Question

I'v read the document about the fft ip core(ug_fft.pdf), the document says "For single engine output FFT/IFFTs, there is an additional factor of Q and the shift must be

2-exponent – Q, where Q = 1 for an even power of 2 and Q = 2 for an odd

power of 2."

I don't know how to choose the value of Q for the single engine output FFT

or what's the meaning of "where Q = 1 for an even power of 2 and Q = 2 for an odd

power of 2"

For example ,

1)if I use 1024 point single engine FFT and the output of exponent scaling values is -11.

Should I choose Q=1? and the shift will be 2^10?

2)if I use 512 point single engine FFT and the output of exponent scaling values is -11.

Should I choose Q=2? and the shift will be 2^9?

altera_forum · Answer

Sounds like it should be this way...

512 points = 2^9 exponent points, therefore:

=> exponents = 9

=> Q = 2

Therefore, shift amount = 2^(-exponent-Q) = 2^-11

1024 points = 2^10 points, thherefore:

=> exponents = 10

=> Q = 1

Therefore shift amount is also 2^-11

altera_forum · Answer

Also another good place to check out the scaling scheme is AN Application Note# 404. In fact I was able to use that Application note to create a unity gain example. (The design, built in 6.1, should be up somewhere on the forum.)

altera_forum · Answer

Can somebody provide a simple example of how to use Altera's FFT IP core?

altera_forum · Answer

Here is a FFT example (with DSP Builder) that I did a while back (back in 6.1), to test out the block floating point equation. Hope this helps...

altera_forum · Answer

Thanks for the quick reply. But what am I supposed to do with it?

Forum Discussion

about fft ip core

8 Replies

Recent Discussions

How to Prevent Agilex 7 F-tile PMA Direct PHY TX Lane Skew

Agilex 7 R-Tile CXL Type-2 IP Hang with Incomplete CXL.cache Operations

Regarding the TX settings of MIPI CSI2 IP

Request for private case.

PCIe Enumeration Failure for CXL IP