Extraordinary behavior of High Shelf filter.
Hello,
I observe a very extraordinary behavior of the High Shelf filter on ADAU1467 and SigmaStudio 4.7 Build 0 Rev 1831 (the latest)
The project (attached) is primitive: WhiteNoise generator – the filter (1 kHz) – spdif out – USB spdif adapter – FFT. The final spectrum is far away from the expectation on any level excepting 0db. The video is attached.
Could it be a problem of the ADAU1467 instance?
Could you please advise what to do? Should I just try to bypass the use of the filter?
Is it possible to suggest any replacement or superposition of other algorithms instead of High Shelf?
Thank you,
SergeTestB.dspproj
Hello Serge,
Sorry for the delayed response, we didn't notice this since you posted in this forum.
Regarding your question, this is a bug with coefficient calculations module (this one calculates coefficients inside DSP). This bug is only with High shelf filter, it works fine with only 0dB boost (without any boost), if we go above 0dB or below 0dB, then the signal will get clipped. Some extra gain is being applied unexpectedly, so it suddenly clips. if we send the signal to the volume control set for -40dB before the filter, then setting the filter to +15dB boost seems fine.
But setting the volume control to -20dB and applying +2dB boost makes the signal to clip. Even going couple of dBs below 0dB makes the signal to clip. So, this is some undesired behavior with this block. The other filters HP, LP, parametric filters etc. are working fine under 'coeff calculation' section. So, we recommend you use other second order filters for High shelf.
The other filter sections are working fine with High shelf (when sigma studio calculates and safeload the coefficients)
Sorry for the trouble.
Regards,
Harish
Please log in to reply.