- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi,
I have a fir compiler with the following settings: Input Bit Width: 16 Coefficient Scaling: Auto with 16 Bits Filter Type: LP with 37 Coefficients Sample Rate: 50 MHz Cut Off Frequency: 10 MHz Passband Ripple: near 0 dB A NCO core with a 16 Bit output supplies the filter with a 1 MHz sinus signal. I analyzed the full resolution output (38 Bits) of the filter in Signal Tap II. 6 MSB Bits were identical so I truncated 5 MSB Bits. I also truncated 17 LSB Bits to get 16 Bits output. But the filter outputs a sinus signal that is attenuated by about -4 dB. In the Frequency Response Tab in fir compiler parameterize GUI it should be near 0 dB. Does anyone have an Idea what the problem could be?Link Copied
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
since you are truncating 17 bits,in effect you are dividing the sum of internal products by 2^17. Thus if you scale your coeffs as:
coeff = coeff/sum(coeff) *2^17 then you should get unity dc gain. but what actually happens at 1MHz may be due to your response- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
thanks kaz! I scaled the coefficients as you said and now I get the right output signal.
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page