Bizarre issue with Impulse Response

[nick72]

Happy New Years Everyone!

I already used REW and a BFD to equalize (as best as I could) my puny Klipsch KSW-10 subwoofer for my living room which is open to the dining and kitchen area.

Today though, I revisited REW in hopes in it assisting me in determining possible room acoustics options (if any) since I find the room sounds pretty terrible.

Reading the REW Help File, it seems the Impulse Response tab could be helpful in this regard.

So I ran 8, 1M full range frequency sweeps on my left Ascend Acoustics CBM-170 speaker at 75 db to get as much detail as possible (first graph below) being driven by a Yamaha HTR-5560 75 Watt x 5 receiving running in 2 channel mode.

To my great surprise, my impulse graph looks beyond bad when compared to the Help File examples and other Impulse Response plots posted in this forum.
I have a significant hill (distortion?) before the actual impulse. This is then followed by a significant delay before all the energy dissipates.

After seeing what may be distortion, I’m currently more concerned about the left side of the impulse. Perhaps it’s pointing to one of the reasons why my room sounds so bad?

I also tried only the subwoofer with the BFD filters disabled(2nd and 3rd graph).
Hmm…Seemed similar to with regard to the left of the impulse.
Finally, I ran a sweep with my BFD filters enabled. (4rth and 5th graph)This seemed better in terms of the left side of the impulse but I have no clue why?

I was wondering if anyone has any insight as to what I’m doing wrong (if anything).
Is that really all distortion on the left of the impulse? If so, could the receiver be introducing it? The speaker? At 75db? I tried the right speaker as well.. Same thing..

Any advice would be greatly appreciated…
Thanks
Nick

[JohnM]

The impulse plot looks entirely normal for a subwoofer, nothing to be concerned about. Distortion shows up as scaled-down images of the main response further to the left of the main peak, the longer the sweep the further to the left they are pushed.

[nick72]

JohnM wrote:

The impulse plot looks entirely normal for a subwoofer, nothing to be concerned about. Distortion shows up as scaled-down images of the main response further to the left of the main peak, the longer the sweep the further to the left they are pushed.

So, the fact that the first graph's full range sweep of the CBM-170 has a rather long rise toward the actual impulse (far more than the -125 ms left IR window) be of no concern?

If not, and it's not really distortion, any idea what it is? Another room issue?

Also, should I increase the left (and right) IR window to where the noise starts (-500 ms, and ~900 ms respectively) to capture this energy information?

Thanks for the help!!

Nick

[JohnM]

There can be pre-ringing on impulse responses from the soundcard filtering and/or the ADCs and DACs in the receiver. Widen the IR windows and see how the response looks. Generally speaking you will get good results with a single 256k sweep, it is rare that there would be a need for the higher S/N of multiple long sweeps unless you were acquiring an impulse response for subsequent post-processing.

[nick72]

JohnM wrote:

There can be pre-ringing on impulse responses from the soundcard filtering and/or the ADCs and DACs in the receiver.

I did a loopback test from my soundcards line out to the line in, removed all calibration files and C weighting. The result, no pre-ringing of any kind. Just the standard post impulse lobes like those displayed in the REW Help File (I can't say I understand why those post impulse lobes are there though)..

I guess the next step is to test the receiver. I'll need to figure out a way to test it. Any ideas?

JohnM wrote:

Widen the IR windows and see how the response looks. Generally speaking you will get good results with a single 256k sweep, it is rare that there would be a need for the higher S/N of multiple long sweeps unless you were acquiring an impulse response for subsequent post-processing.

Thanks John! I was hoping with the mutiple sweeps and higher samples would give me a better resolution on determining reflection points.
I will try widening the IR window and see what it gives me..

Thanks!

[JohnM]

nick72 wrote:

I did a loopback test from my soundcards line out to the line in, removed all calibration files and C weighting. The result, no pre-ringing of any kind. Just the standard post impulse lobes like those displayed in the REW Help File (I can't say I understand why those post impulse lobes are there though)..

They are there because the response of the soundcard does not go right down to DC. To have a perfect spike requires that the card be flat from DC to half the sampling frequency, about the only time that occurs is when an internal digital loopback is being used.