What am I doing wrong? First REW Measurements

[Kruppy]

First off this is my first attempt at an REW measurement and I have not had a lot of time to trouble shoot what's going on (I've already had a good couple hours into the set up).

I think I have the sound card calibrated correctly.

Why do things get so out of control at 200Hz and higher?

Why do I see output increase from about 7Hz and lower? That makes absolutely no sense to me.

I'm trying to evaluate my Martin Logan SL3s without a subwoofer.

What should I have my SPL meter set at fast or slow response for C weighted? I don't remember seeing that in the help file. The graph posted here is fast response, I did collect slow response also, but it's basically the same shape curve just a number of dBs lower.

I think I also had the levels set up correct, targeting 75 dBs.

When running a measurement, what kind of time setting should this this run for?

[brucek]

Why do things get so out of control at 200Hz and higher?

When plotting responses larger than a subwoofer standard of horizontal = 15Hz-200Hz, turn on smoothing. Smoothing is used for full range measurements (such as you are displaying), where reflections can cause comb filtering that makes it difficult to see the underlying trend of the response. It is rarely used for low frequency measurements though as it would obscure the true shape of the response and so not allow accurate correction filters to be determined.

Why do I see output increase from about 7Hz and lower? That makes absolutely no sense to me.

This is where your signal drops into the noise, but the meter compensation continues to raise the signal. Ignore it, it's not signal, it's noise. Limit your bottom horizontal axis to 15Hz.

What should I have my SPL meter set at fast or slow response for C weighted?

Slow and C-Weight.

The RadioShack meter can give a decent measure from 15Hz to 5000Hz.
Set your horizontal to that and always set your vertical to 45dB-105dB. If you want the sub only range, set the horizontal to 15Hz -200Hz. Don't use smoothing in the subwoofer range, but do so in the full range plots.

brucek

[Kruppy]

brucek wrote:

When plotting responses larger than a subwoofer standard of horizontal = 15Hz-200Hz, turn on smoothing. Smoothing is used for full range measurements (such as you are displaying), where reflections can cause comb filtering that makes it difficult to see the underlying trend of the response. It is rarely used for low frequency measurements though as it would obscure the true shape of the response and so not allow accurate correction filters to be determined.

This is where your signal drops into the noise, but the meter compensation continues to raise the signal. Ignore it, it's not signal, it's noise. Limit your bottom horizontal axis to 15Hz.

Slow and C-Weight.

The RadioShack meter can give a decent measure from 15Hz to 5000Hz.
Set your horizontal to that and always set your vertical to 45dB-105dB. If you want the sub only range, set the horizontal to 15Hz -200Hz. Don't use smoothing in the subwoofer range, but do so in the full range plots.

brucek

brucek,
Thanks for the responce. I'll give the Smoothing feature a try.

What would be a better way to evaluate above 5000Hz? How unaccurate is my RadioShack meter at those upper frequencies?

What about the time length for the frequency sweep?

[JohnM]

The soundcard cal trace looks wrong to me - I'd suspect some kind of effect or EQ was enabled causing the very uneven response with those peaks at 2, 4, 8 and 16kHz.

[Kruppy]

JohnM wrote:

The soundcard cal trace looks wrong to me - I'd suspect some kind of effect or EQ was enabled causing the very uneven response with those peaks at 2, 4, 8 and 16kHz.

Never thought of that. I'll check into it. I did think that it was odd that there were peaks at 2, 4, 8, and 16k.

Thanks
Bob

[Kruppy]

brucek wrote:

When plotting responses larger than a subwoofer standard of horizontal = 15Hz-200Hz, turn on smoothing. Smoothing is used for full range measurements (such as you are displaying), where reflections can cause comb filtering that makes it difficult to see the underlying trend of the response. It is rarely used for low frequency measurements though as it would obscure the true shape of the response and so not allow accurate correction filters to be determined.

This is where your signal drops into the noise, but the meter compensation continues to raise the signal. Ignore it, it's not signal, it's noise. Limit your bottom horizontal axis to 15Hz.

Slow and C-Weight.

The RadioShack meter can give a decent measure from 15Hz to 5000Hz.
Set your horizontal to that and always set your vertical to 45dB-105dB. If you want the sub only range, set the horizontal to 15Hz -200Hz. Don't use smoothing in the subwoofer range, but do so in the full range plots.

brucek

Here's the smoothed full freq response and the unsmoothed 15-200Hz.

I need to check into the soundcard eq issues as mentioned above.

[brucek]

The soundcard cal trace looks wrong to me

Yeah, I missed that in my rush to get through a bunch of answers this morning. The soundcard calibration should be redone. It should be quite smooth. Then you should check it by doing a sweep measurement of the loopback cable with the soundcard cal file stored. The result should be dead flat. Don't carry on until you do this.........

What would be a better way to evaluate above 5000Hz?

Better microphone... i.e. ECM8000 with preamp or a Galaxy CM140 SPL meter. We have cal files for both on the site. The RS meter is fine for subwoofer checks though, and up to maybe 5K.

and the unsmoothed 15-200Hz

When we test the subwoofer, it's usually best to check it first without the mains turned on. Then you can see exactly how the sub itself reacts to the room. Then you add the mains and see the effects that are caused at the crossover. For the 15Hz to 200Hz sweep use the specific subwoofer test signal in the REW pulldown.

I see you have a dip at about 80Hz. This would likely be caused by the interaction of the mains and sub at the crossover. Sub phase control and sub receiver distance trim can solve that usually.

What about the time length for the frequency sweep?

Longer sweeps provide higher signal-to-noise ratio (S/N) in the measurements, each doubling of the sweep length improves S/N by almost 3dB.
You can also set the number of sweeps to use. This then applies synchronous pre-averaging, capturing the selected number of sweeps per measurement and averaging the results to reduce the effects of noise and interference. Again, the pre-averaging improves S/N by almost 3dB for each doubling of the number of sweeps. This feature helps if you are experiencing noise in the room.

brucek

[Kruppy]

Alrighty,
Yup, you guys were right, the sound card eq was active and the 2, 4, 8, and 16 peaks corresponded to eq settings.

Well, I set the eq to flat and re-cal'd the sound card (notice how flat it is now) and then took another meausurement, see below:

No huge change that I can see.

Bruce,
No sub. These are full range speakers. I'm hoping to add a sub in the near future.

[Kruppy]

Here's another (potentially dumb) question: How directional are the RadioShack meters? All my measurements have the business end of the meter pointing in-between my mains. Should it be pointing up?

[brucek]

No sub. These are full range speakers

Really good low end extension to them....

Should it be pointing up?

Well, the accepted position is pointed at and between the two mains at the listening position with about a 45 degree angle toward the ceiling. I don't know that it will make that much difference, but why not.

Try testing at 75dB. You're testing a bit low. Test only to 5K with the RS meter.

brucek

[Kruppy]

brucek wrote:

Really good low end extension to them....


Well, the accepted position is pointed at and between the two mains at the listening position with about a 45 degree angle toward the ceiling. I don't know that it will make that much difference, but why not.

Try testing at 75dB. You're testing a bit low. Test only to 5K with the RS meter.

brucek

I'm glad I asked. I'll take another measurement at 45deg.

As far as I understand, I am testing at 75 dB. Every level check I do, I'm right at it. I was concerned at first that something was not right.....

Btw, thanks for all the help.

[brucek]

Every level check I do, I'm right at it.

Probably the big peak from 6K to 9K is adding the extra energy to the level setting and so the overall is below 75dB.... no big deal.

brucek

[clubfoot]

Do you have your MLs toed in, out or straight ahead?

[Kruppy]

Toed in slightly. I've played with location quite a bit. I've had to compromise between low end output and soundstage.

[clubfoot]

Kruppy wrote:

Toed in slightly. I've played with location quite a bit. I've had to compromise between low end output and soundstage.

Try them straight ahead and not too close to side walls see if the 80Hz dip reduces.

I have to say very good extension low down.