low frequency response from linear sweep wav file

Rider

Registered
Thread Starter
Joined
Feb 16, 2018
Posts
2
Dear All,
I wanted to check the frequency response of a linear sweep audio file (wav) with REW. The file is an output file of the "benchmark" tool from the "Equalizer-APO" software.
I imported the file as audio file into REW and expected to see a totally flat response (since no filters active in Equalizer APO). However in the low frequencies, the response cureve was wavy. In order to doublecheck, I also used the REW Generator to generate a linear sweep from 10Hz-22kHz, but get the same result (see picture). Can someone explain me, what I'm doing wrong, or why the FR for low frequencies is not displayed flat as I expected? Thanks in advance!
upload_2018-5-31_20-17-18.png
 

John Mulcahy

REW Author
Joined
Apr 3, 2017
Posts
8,030
Check the windowed view on the Impulse graph and make sure the window is not truncating the sweep. Beyond that you will need to post the file, there is only so much that can be gleaned from an image.
 

Rider

Registered
Thread Starter
Joined
Feb 16, 2018
Posts
2
Hi John,
Thanks for your quick response! In the attachment the wav file which I have generated by the REW Generator. I have loaded that file via File->Import->audio data back into REW. Ripple at low frequencies with max. of +1,6dB @ 30Hz...
Have checked the windowed view of the impulse, but for me all seems o.k., although I have to admit that I'm just starting with REW and therefore have not much compedence on its usage.
 

Attachments

  • LinearSweep_10_22050_44k_16-bit.zip
    6.7 MB · Views: 36

John Mulcahy

REW Author
Joined
Apr 3, 2017
Posts
8,030
Thanks. The ripple at the start of the response is because the sweep starts abruptly at its initial frequency (10 Hz in this case) without being slowly ramped up. That abrupt start creates ripples in the frequency response. It is a property of the signal, generally speaking a sudden change in the time domain (like the start of the sweep) will cause ripples in the frequency domain, and a sudden change in the frequency domain would have corresponding ripples in the time domain.

The ripple you are seeing for that particular file is larger than it should be due to a bug in the generator which is not resetting the sweep starting phase to zero each time you generate a new sweep (I've fixed that for the next beta release), giving an even more abrupt start, but even without that (e.g. for the first sweep generated after starting up) the first peak of that ripple is about 0.8 dB above the main level.
 
Top Bottom