[hifisponge]

I EQ'd my sub a while ago based on the frequency response graph alone before understanding the importance of using the waterfall plots. Now I'd like to get some advice on the matter because subjectively, the bass still sounds a bit hot in spots.

So here is what I was dealing with (FR plot of sub, no EQ, 70Hz x-over, 12dB slope, house curve for elevated LF response):

FR plot of sub with EQ:

With the EQ engaged, the 30Hz hump is much better balanced (I intentionally under-corrected the 30Hz hump based on listening preference), but there is something still going on between 30Hz and 70Hz that sounds peaky. Where? I'm not sure. That's what I'm trying to figure out.

Based on this this waterfall of the EQ'd response, which frequencies would you try to notch out?

41Hz and 69Hz?

If yes, it would seem that I would be creating an even deeper notch in the FR at 41Hz and a new notch in the FR at 69Hz. Is it advisable to sacrifice FR smoothness for lower decay in this case?

Here's the waterfall of the uncorrected sub:

Based on the above measurement, which frequencies would you filter and why? What is the goal? To get the decay below 150ms, 200ms, 300ms?

Oh, and is it OK to measure with the crossover engaged like I've done here?

Thanks in advance for your help.

- Tim

 

[brucek]

is it OK to measure with the crossover engaged like I've done here?

Actually, I think your waterfall looks fairly good. I would add the mains and take a look over the same bandwidth to see if the mains are creating something in the response that is causing the objectionable sound.

Is it advisable to sacrifice FR smoothness for lower decay in this case?

No, the filters have to be modified to reduce the decay. If you have a peak, and a filter is applied that is the correct center frequency and bandwidth, the decay will be reduced. If it's not exactly correct, it can easily reduce the amplitude of the peak and have minimal effect on the decay. It's always tricky when you have a peak and then a sharp dip in close proximate.

brucek

 

[hifisponge]

Actually, I think your waterfall looks fairly good. I would add the mains and take a look over the same bandwidth to see if the mains are creating something in the response that is causing the objectionable sound.

Thanks Bruce. :bigsmile:

I've listened to the mains running full range sans sub and the problem goes away. :mooooh:

No, the filters have to be modified to reduce the decay. If you have a peak, and a filter is applied that is the correct center frequency and bandwidth, the decay will be reduced. If it's not exactly correct, it can easily reduce the amplitude of the peak and have minimal effect on the decay. It's always tricky when you have a peak and then a sharp dip in close proximate.

brucek

OK, I'll see what I can do about fine tuning to the filter center frequency and width, but I'm still not sure which frequencies I should be aiming for based on the waterfall plots. Looking at the yellow waterfall plot, it looks like there are two closely spaced resonances on either side of 30Hz, another at 40ish, 50ish and 70ish. Do you agree? And what is the goal for decay times? What value am I aiming for?

 

[brucek]

If you have no ringing left above 45dB after 300msec, you're doing fine......

brucek

 

[hifisponge]

If you have no ringing left above 45dB after 300msec, you're doing fine......

brucek

Odd. Then that would suggest that the only issue left in my EQ'd waterfall plot is at 30Hz, which sounds fine to me. I swear that I hear a problem with certain bass notes above that point. Any thoughts?

 

[brucek]

the only issue left in my EQ'd waterfall plot is at 30Hz

And that problem at 30Hz may be external noise. If you see a signal extend out of the waterfall at a fairly constant rate that doesn't appear in the original time zero response, it can be a noise from cars, truck, air conditioner, etc......


brucek

 

[hifisponge]

And that problem at 30Hz may be external noise. If you see a signal extend out of the waterfall at a fairly constant rate that doesn't appear in the original time zero response, it can be a noise from cars, truck, air conditioner, etc......


brucek

Understood. So lets say this was your system, and you hear a problem in the subjective bass quality somewhere between 30 and 80Hz, based on the waterfall plot, what would you try?

 

[brucek]

Can we have a look at the EQ'd 15Hz-200hz graph (response and waterfall) with the mains added?

brucek

 

[hifisponge]

Can we have a look at the EQ'd 15Hz-200hz graph (response and waterfall) with the mains added?

brucek

Will do, but this may take me a while. The wife hates it when I run tests when she is home, so I'm going to have to see about taking a day off from work to do the next round of measurements. :waiting:

 

[Wayne A. Pflughaupt]

OK, I'll see what I can do about fine tuning to the filter center frequency and width…

REW can do that for you automatically. That’s what the program is designed to do.

So lets say this was your system, and you hear a problem in the subjective bass quality somewhere between 30 and 80Hz, based on the waterfall plot, what would you try?

It may be nothing more than response needing to track the Target better. As it is now, EQ’d response sags considerably below the Target between 40-80 Hz. Adding the mains may be exaggerating the 80 Hz area even worse. If that’s the case, with any luck it will be apparent with the extended-range w/ mains graph brucek requested.

If all else fails, here’s a little trick you can do: Dial in a filter on the equalizer about 1/3 - 1/4-octave bandwidth, boost 5-6 dB, and move the frequency setting slowly between 30-80 Hz. When you hit the problem frequency it will “jump out” big time. That will let you know where the problem is. At that point you can cut the gain and tweak the bandwidth as needed to tame the problem.

Will do, but this may take me a while. The wife hates it when I run tests when she is home, so I'm going to have to see about taking a day off from work to do the next round of measurements.

No need to do that; just send her to the mall with your credit card for an hour or two.

Regards,
Wayne

 

[hifisponge]

REW can do that for you automatically. That’s what the program is designed to do.

I was playing around with that feature last night and it didn't seem to hit all the problem frequencies. It only chose two filters. It left one of the peaks around 30Hz which resulted in an uncorrected 10dB peak in that range.

It may be nothing more than response needing to track the Target better. As it is now, EQ’d response sags considerably below the Target between 40-80 Hz. Adding the mains may be exaggerating the 80 Hz area even worse. If that’s the case, with any luck it will be apparent with the extended-range w/ mains graph brucek requested.

Good point, there is a broad dip in the range that you point out now that I look at it. So it could be the imbalance that is making 80Hz sound outa whack. Seems probable.

If all else fails, here’s a little trick you can do: Dial in a filter on the equalizer about 1/3 - 1/4-octave bandwidth, boost 5-6 dB, and move the frequency setting slowly between 30-80 Hz. When you hit the problem frequency it will “jump out” big time. That will let you know where the problem is. At that point you can cut the gain and tweak the bandwidth as needed to tame the problem.

That's great advice! Question: My EQ sets the filter width by a Q value. Do you know the Q value equivalent for 1/3 Octave? I'm definitely going to work on getting a more respectable plot first, of course.

Note: Wayne answered with this: 1/3-octave is Q-4.3, 1/4-octave is Q-5.8" but it got deleted somehow.

No need to do that; just send her to the mall with your credit card for an hour or two.

Regards,
Wayne

Too funny, but also a great idea. :clap:

 

[brucek]

That all depends on the equalizer, since the definition of bandwidth (i.e. 3db down or the half gain points) and the Q formula can be different for each type.

For the DSP1124P, it defines bandwidth as:

Bandwidth (Hz) = centre frequency*(BW/60)*sqrt(2)

So, the Q formula becomes:

Q = 60/[(BW/60)*sqrt(2)]

For the FBQ2496 the bandwidth control adjusts in 1/60 of an octave steps from 1/60 to 5/60 of an octave, then goes through 1/10, 1/9, 1/8, 1/7, 1/6, 1/5, 1/4, 1/3, 1/2, 3/4, 1, 1.5, 2, 3, 4, 5, 6, 7, 8, 10 octaves, and so the relationship between Q and Bandwidth becomes:

Q = sqrt(2)/BW

so the bandwidth range of 1/60 to 10 octaves gives a Q range from 84.85 to 0.14.

and so on, and so on.

You'll have to look at your own equalizer manual to find out.

Anyway, Wayne only meant your filter to be within a ballpark. Use this quick calculator (with BW spec shown) as a rough guide.

brucek

 

[hifisponge]

Good news, I'm taking tomorrow and Friday off, so I can take another crack at getting my subwoofer measured and corrected. I also did a little more by-ear testing yesterday by listening to sweep tone tracks that I downloaded. To my ear, it sounds like the problem area is in the 65-70Hz range.

 

[JohnM]

REW can be helpful to track down problem frequencies, just set the Signal Generator in Sine mode and check the "frequency tracks cursor" box, you can then move the cursor about on the graph and the signal generator will follow it.

 

[hifisponge]

REW can be helpful to track down problem frequencies, just set the Signal Generator in Sine mode and check the "frequency tracks cursor" box, you can then move the cursor about on the graph and the signal generator will follow it.

Very cool. Thanks for the tip!

 

[hifisponge]

OK, so I took another stab at getting my sub to sound smoother by setting the filters to address long decay times instead of just going for flat response and I think I've gotten closer.

Here's what I started with (no EQ)

Here is what I got when I tried to flatten the FR

And here is what I got when I went after extended decay

You'll notice that there is a dip at around 120Hz in the new graph, but this is the nature of the beast when it comes to measuring bass. Move the mic and inch and you get a different FR plot.

I'm fairly certain that the range that I was hearing problems was due to the ridge at 69Hz in the green plot. I confirmed this by using JohM's recommendation to use the Sine Wave Generator and clicking on that part of the graph and then comparing it by ear to some of the neighboring frequencies.

I only have 5 EQ filters in my prepro, so I had to hit the high priority problems and leave a hump at 25Hz (no complaints by me), and a ridge at 89Hz (which is in the crossover region anyway and of little consequence).

Feedback welcome.

I'll do some music listening tomorrow to see if I've truly addressed the mid/upper bass problem that I was hearing.

Cheers,

- Tim