REW & RS Digital Meter - strange results in the high freqs

[cyberbri]

I was wondering how accurate or inaccurate the newer digital RS meters are in the higher frequencies. I know they are pretty close in the bass region.

But I ask because this is the result I got with REW from my Ascend 340SE center channel (I had the receiver in DPL II Movie) + sub, with the meter on a tripod in the LP aiming diagonally up, as well as aimed forward. (click links to see graphs)




In my room I do have absorber panels on the left/right sides, and I have a large 3~4" thick floor cushion that I put on top of the large square coffee table.

Any idea why the frequency response looks like this in the highs?

[curtis]

I have been told that the RS meter is not good for treble frequencies.

Also, it could still be reflections eventhough you have panels, right? The panels do not absorb everything do they?

[cyberbri]

I have been told that the RS meter is not good for treble frequencies.

Also, it could still be reflections eventhough you have panels, right? The panels do not absorb everything do they?

Thanks.
The panels are just at 1st ref points, with a big cushion on the coffee table in the middle.

If it's just the meter, I wonder if there's a .cal file that works with REW to compensate...

I also just read this:
http://www.avsforum.com/avs-vb/showt...&&#post9542081

So I guess I need to do different tests that focus on just the high range. Another thing is, REW often (always?) tells me the signal coming from the SPL meter is clipped. I guess I'll just run some different tests.

[davef]

Hi Cyber,

Looks normal to me for an in-room response

I am not familiar with Room EQ Wizard but what you are seeing in the HF range are nothing more than HF reflections. The higher the frequency, the shorter the wavelength and the easier it will reflect off various surfaces.

It looks particularly bad because the graph is not smoothed. For in-room response, it is common to use 1/3 octave smoothing. This means 3 measurements for each octave. Personally, I prefer 1/12 octave smoothing for in-room measurements.

Additionally, it is important to understand the nature of the graph. The response graph is logarithmic, not linear. At higher frequency ranges there is much more data than at lower frequenicies fit into the same visual scale.

For example, look at the range from 100Hz to 200Hz. That is a range of 100Hz fit within a visual scale of about 3/4 inch. Now look at the range from 1kHz to 2kHz. This is a 1000Hz range of data also being fit within a visual scale of about 3/4 of an inch. 10 times the data forced to fit within the same scale. Look at the range from 10kHz to 20kHz, that is 1/2 of all the response data fit into a visual scale of only 3/4 of an inch.

Smooth the response to 1/3 or 1/6 octave and you will get a graph that looks more familiar to you.

[cyberbri]

Thank you! I've been experimenting more, and trying the different smoothing as well.

I also tried moving th speakers back a bit, and at a distance of .73 x distance from LP to middle (of T), as prescribed in the manual. Tighter center imaging, seemingly wider soundstage, at least on a few tracks I've listened to. After I did that, I've been tinkering with REW to re-do the sub filters and stuff...