TOPIC: FIR measurement does not look normal

Hi there,

First of all thank you Eclipse Audio for the great support and the awesome software.

I am using PWR-ICE125 and have decided to use FIR filters on the loudspeakers I make. So I have found Fir Designer which I believe is the only wonderful fir software for OS X users…

Yesterday, I have applied my first fir filters to the speaker I made. I am on the way to improve the results so I have a few questions…

When I make the measurements very close (1 cm) to the drivers, I get a very nice and smooth results. But when I do a measurement from 1 meter, I get a really bad response.

Here is the measurement I made 1 centimetre close to the drivers. Fir filters are applied and everything looks normal. Link below
imgur.com/3s4L9xU

And here is the measurement I made 1 cm close to the drivers before fir filters applied. I have used these measurements to design the fir filter. Link below
imgur.com/2GbmraZ

Also notice that there is a crossover just to protect the drivers. I think that this approach (using a crossovered measurements to design fir filters) may be wrong maybe because I apply another crossover to the measurement on the Fir Designer… I feel like this may be causing a problem… So I would like to hear your comment about this issue…

So, here is the problematic measurement that I made from 1 meter. (link below) It looks terrible. Fir filters are applied here but they seem not to work… Both mag and phase look not normal. (I also did not apply any delay to any of the drivers from the PWR-ICE125 software. I just applied the fir filters.)
imgur.com/ArWV5Ej

And in addition, this is the fir filter design for the woofer
imgur.com/ck08iFA

And the fir filter applied measurement (from 1 cm) of the woofer
imgur.com/EnUpCLq

Fir filter design for tweeter
imgur.com/UZoZjSt

Fir filter applied tweeter (1 cm measurement)
imgur.com/SBSMFAt


So my first question is, why I get weird measurement from 1 meter which is closer to the listening position…

And my second question is, should I apply (or not) crossover to the first measurements which will be used to build the fir filters...

PS: here is the link that you can see all the pictures together
imgur.com/a/AhIR4

Thank you very much, I appreciate a lot...

Hi Emre,

Can you post a photo of your measurement setup at 1m, showing the speaker, microphone and anything else nearby?

Kind regards,
Michael

Thank you for fast response...

Here is my setup, I know not a good room to make measurements...
imgur.com/a/VWZ4n

I also made this measurement here at the same room too, it is the IIR filter that I measured and EQ'ed from REW...
imgur.com/vqcBJjh

Cheers...

PS: I am planning to make another measurement without using crossovers before building the fir filters...

Hi emre,

I'd suggest finding a more open space in which to take measurements and do some experimentation at different distances. Reflections and room modes/resonances can all affect measurements. (Even in an open space, there's still a floor reflection.) I often take PA speakers to a large hall - where I can push the measurement gating out enough to get a good response at low frequencies - or outside.

Take multiple measurements at slightly different locations and average them. Unless working in very controlled conditions - e.g. an anechoic chamber - with very controlled (in dispersion) drivers, there will always be some variation in/around the listening position. The "Measurement Averaging" tab in FIR Designer can help here.

Tuning a loudspeaker using measurements at one location can be problematic. It's easy to make the loudspeaker measure well at that position, but the speaker can sound (and measure) worse everywhere else.

Understanding what aspects of your environment are affecting the measurements is very important. For example a dip in a measurement could be a cancellation caused by a reflection off an adjacent object. An EQ boost to "fix" this dip will not work since there is an inherent cancellation occurring. I.e. any extra energy pushed into this frequency area, using EQ, will still cancel and the dip will remain.

Taking good, representative loudspeaker measurements is difficult and I'm sure there are many other folks on this forum better qualified to give guidance in this area. Hopefully they will weigh in...

Kind regards,
Michael