Welcome, Guest
Username: Password: Remember me

TOPIC:

Differences when loading Averager results into REW 3 years 10 months ago #45067

  • musan
  • musan's Avatar
  • Offline
  • Junior Member
  • Junior Member
  • Posts: 25
  • Thank you received: 0
Yes, the measurements are taken at different locations with a timing reference using a loopback and that I did to preserve the phase relationships between all the measurements. Shouldn't removing measurement delays and vector averaging the results, approximates the power averaging response? I just did that and the vector average of all measurements (with measurement delays removed) still results in lower mid to high frequency range. See below;

The intent is to use the average magnitude response with the "average" phase response at the design axis for DSP based EQ. In other words, I want to EQ both the "average" magnitude and phase response at the design axis using FIR filtering ; just not quite convinced the vector magnitude response is the right one to use for the car environment where I am carrying out these measurements.

In any case, this is where Averager comes in, in that I can quickly evaluate these options and decide which one I want to work with further. Is it not to say it cannot be done in REW, I just find it easier to do with Averager hence I use both tools interchangeably.

Please Log in or Create an account to join the conversation.

Differences when loading Averager results into REW 3 years 10 months ago #45069

  • JohnPM
  • JohnPM's Avatar
  • Offline
  • Senior Member
  • Senior Member
  • Posts: 40
  • Thank you received: 15
It doesn't make sense to preserve relative timing when averaging measurements made at different positions. REW's vector average is averaging magnitude and phase rather than power, it is a coherent average. I will look at providing an option to average power and phase instead, more akin to incoherent averaging.

Please Log in or Create an account to join the conversation.

Differences when loading Averager results into REW 3 years 10 months ago #45072

  • musan
  • musan's Avatar
  • Offline
  • Junior Member
  • Junior Member
  • Posts: 25
  • Thank you received: 0
The idea is to look for magnitude and phase similarities while cross correlating these measurements around the one made at the design axis or where the head would be located. Once the delays are removed, cross correlation is performed followed by averaging.

Thanks for looking at the incoherent averaging capability. :)

Please Log in or Create an account to join the conversation.

Moderators: devteam