Understanding waterfall plots?

**Kevin_Wadsworth** · **10-15-09, 11:32 AM**

I have a custom-built sub in me basement with a BFD 1124 doing the EQ. I had Eq'ed it with a desktop once after install, but after getting a Turtle Beach USB soundcard for my laptop to measure my parent's sub, I thought I'd play some with mine.

Here is the frequency response with Wayne's "hard knee" house curve:

I think the rise below 20 Hz is artificial/soundcard-induced because I didn't see that in previous measurments, but instead saw a 10 dB drop between 20 Hz and 10 Hz. I'm a little dissatisfied with the drop from the target in the 80-180Hz range, but I haven't figured out the filter combination to fix that yet.

I had never measured a waterfall of my system before, but I thought I'd take the opportunity with the new soundcard. The problem is, I have no idea how to interpret a waterfall plot! Any thoughts? How does this look?

**Wayne A. Pflughaupt** · **10-15-09, 12:07 PM**

Quote:

I'm a little dissatisfied with the drop from the target in the 80-180Hz range, but I haven't figured out the filter combination to fix that yet.

I usually don’t do much EQ in that range because the mains are overlapping. I’d suggest taking a reading with the mains added to see what you get. If you still think EQ is needed, a 1/6-octave filter (I think that’s something like 8-10/60 in “BFD-speak”) centered a bit north of 90 Hz should do the trick.

Quote:

I had never measured a waterfall of my system before, but I thought I'd take the opportunity with the new soundcard. The problem is, I have no idea how to interpret a waterfall plot! Any thoughts? How does this look?

The waterfall looks fine.

Regards,
Wayne A. Pflughaupt

**Low-Q** · **10-15-09, 02:44 PM**

The response below 20Hz is probably mechanical vibrations in the mic due to mechanical connection betwen the sub, via the floor, and to the mic. I allways get a quite different bass response when I am hanging the mic in elastic wires, and add some weight to the mic, or holding it in my hand during measuring.

Most probably you have a much less response in acoustical energy below 50 - 60Hz than measured with the mic.

Br.

Vidar

**glaufman** · **10-16-09, 11:24 AM**

Quote:

Wayne A. Pflughaupt wrote:

I usually don’t do much EQ in that range because the mains are overlapping. I’d suggest taking a reading with the mains added to see what you get. If you still think EQ is needed, a 1/6-octave filter (I think that’s something like 8-10/60 in “BFD-speak”) centered a bit north of 90 Hz should do the trick.

The waterfall looks fine.

Regards,
Wayne A. Pflughaupt

You're not concerned with 20-30 Hz?

**brucek** · **10-16-09, 12:02 PM**

Quote:

You're not concerned with 20-30 Hz?

If you're referring to the ringing at 20-30Hz, then no, it's not signal ringing - it's noise.

Notice how the signal very nicely decays at 20-30Hz from the first slice with fairly even spacing in time? Then all of a sudden a signal appears that rings and rings. It's noise -just as the signal at 120Hz is a harmonic of some 60 Hz hum he has somewhere (likely caused by the PC being connected to the system).

These type of things can be ignored - the waterfall is quite good...

brucek

**Kevin_Wadsworth** · **10-16-09, 02:06 PM**

Thanks for the comments, everyone. It's good to know the waterfall looks reasonable. I've been enjoying the performance of the sub for the past year.

I'd like ot figure out how to do the graphs that I've seen on other subwoofer reviews that show compression of the sub at higher volumes. When I tried increasing the volume with this setup, I got errors about somethign clipping, even though volumes were still relatively low.

Perhaps I just need to read the documentation on setup in REW again.

:

To take a full-range reading (and see if the >80 Hz dip is causing problems) do you just run the same sweep, but change the upper end of the frequency range to 2000 Hz instead of 200Hz? I tried that (I think), but got a graph that just looked like noise. FWIW, I'm using the old analog RS meter as a mic.

**brucek** · **10-16-09, 02:30 PM**

Quote:

I'd like to figure out how to do the graphs that I've seen on other subwoofer reviews that show compression of the sub at higher volumes. When I tried increasing the volume with this setup, I got errors about something clipping, even though volumes were still relatively low.

Each time you turn up the volume to take another sweep, run the Check Levels routine to setup your levels, then run the Calibrate SPL routine and set the REW meter to 75dB (even though it's actually higher each time you measure).

Then you can overlay all the results.

Quote:

To take a full-range reading (and see if the >80 Hz dip is causing problems) do you just run the same sweep, but change the upper end of the frequency range to 2000 Hz instead of 200Hz?

Yes, but remember you setup your levels using a band limited pink noise that had limits of 30Hz-80Hz, so you may have to fuss with the levels a bit. Test with the Check Levels in the Measurement screen to check if the levels are OK. If not, tweak the input level.

brucek

**glaufman** · **10-16-09, 04:38 PM**

Quote:

Kevin_Wadsworth wrote:

To take a full-range reading (and see if the >80 Hz dip is causing problems) do you just run the same sweep, but change the upper end of the frequency range to 2000 Hz instead of 200Hz? I tried that (I think), but got a graph that just looked like noise. FWIW, I'm using the old analog RS meter as a mic.

Quote:

brucek wrote:

Yes, but remember you setup your levels using a band limited pink noise that had limits of 30Hz-80Hz, so you may have to fuss with the levels a bit. Test with the Check Levels in the Measurement screen to check if the levels are OK. If not, tweak the input level.

If it was me, what Bruce said here would be more important than going up to 2kHz right now. Then, turning on the mains and doing the same sweep.

**Moonfly** · **10-17-09, 06:59 PM**

Quote:

brucek wrote:

Each time you turn up the volume to take another sweep, run the Check Levels routine to setup your levels, then run the Calibrate SPL routine and set the REW meter to 75dB (even though it's actually higher each time you measure).

Then you can overlay all the results.

brucek

I just leave the SPL as it is calibration wise (the radio shack one), and increase the range to my target area till the check levels comes back as fine. Might be the easier way, instead of calibrating the spl each time, I find it ok anyway.

I suppose, technically speaking, recalibrating is the right way to go, but I dont find it necessary myself.

As for waterfalls, IIRC ( and correct me if not), anything under 500ms is considered a good response. Thats what I work to anyway. As others have said, waterfall looks good to me.

EDIT:

I have a question for Bruce and wayne. I'm currently building a DIY sub, and want to test a compound push pull design (designed to remove even ordered harmonics and improve SQ) versus a Seaton Submerssive dual opposed design. Can I use the waterfalls to 'see' this difference if there is any?

**brucek** · **10-17-09, 08:02 PM**

Quote:

Can I use the waterfalls to 'see' this difference if there is any?

In what regard? What would you be looking for? Waterfalls are basically showing you the room...

brucek

**DrWho** · **10-18-09, 01:33 AM**

Quote:

Moonfly wrote:

Can I use the waterfalls to 'see' this difference if there is any?

"Can" you? If you recorded playing a single tone, I suppose you could always run it through the waterfall FFT to watch the distortion decay.

I think this would be much easier to see with the RTA function (the Spectrum tab in REW).

**Moonfly** · **10-18-09, 06:27 AM**

Quote:

brucek wrote:

In what regard? What would you be looking for? Waterfalls are basically showing you the room...

brucek

Well the push pull design is supposed to improve driver performance by removing distortion. I'm wandering if you can use REW specifically to measure this and actually see if there is any measurable difference in my room..

It may well just be a case of look at the waterfalls I guess. To measure the driver performance, would I technically speaking, have to put the SPL inside the cabinet, otherwise like you say, Im measureing the room rather than the drivers.

Sorry if I'm hijacking the thread a little.

**brucek** · **10-18-09, 08:21 AM**

Quote:

To measure the driver performance, would I technically speaking, have to put the SPL inside the cabinet

That wouldn't measure driver performance. If you want to measure a driver without the rooms influence you would take measurements outside.

brucek

**Moonfly** · **10-18-09, 01:16 PM**

Quote:

brucek wrote:

That wouldn't measure driver performance. If you want to measure a driver without the rooms influence you would take measurements outside.

brucek

Well yeah, but if there is one configuration distorting less than the other, I should still be able to see the same differences in room anyway, so no need to go outside I dont think.

I'm wandering if there is a way to use REW to see if there is any worthwhile removal of harmonic distortion. Waterfalls might not be the best way to see this directly, but I'm thinking that a sub distorting less should ring less. Not sure?

**brucek** · **10-18-09, 02:20 PM**

Quote:

but I'm thinking that a sub distorting less should ring less.

The sub isn't ringing though, the room is.........

brucek

**Moonfly** · **10-18-09, 05:49 PM**

Quote:

brucek wrote:

The sub isn't ringing though, the room is.........

brucek

Isnt there a relation between driver distortion and ringing. Sure the room has a bearing on this, but I thought higher distortion within a driver will provoke this ringing effect more?

**brucek** · **10-18-09, 06:39 PM**

Quote:

Isnt there a relation between driver distortion and ringing. Sure the room has a bearing on this, but I thought higher distortion within a driver will provoke this ringing effect more?

No, I don't believe so. The window that is used on the impulse response to produce the frequency response and the waterfall slices specifically does not include the harmonic distortion images, so they won't show up in the waterfall plots. You can examine the impulse response itself though (before time zero) and see the distortion harmonics. The first harmonic before time zero is the second harmonic and the next is the third, etc, as time gets more negative.

brucek

**Wayne A. Pflughaupt** · **10-18-09, 07:16 PM**

Quote:

Moonfly wrote:

Isnt there a relation between driver distortion and ringing. Sure the room has a bearing on this, but I thought higher distortion within a driver will provoke this ringing effect more?

All the waterfall is showing is how long it takes for the signal to fade away (or drop) about 40-45 dB - i.e., from whatever volume level it was taken at (typically something like 80-85 dB at the peaks) to the 45 dB floor of the graph (of course, the graph's floor can be lowered, if desired). Logically, the driver itself has nothing to do with signal decay time, and certainly nothing related to distortion will affect the signal decay time. The only thing that will affect the decay time is bass frequency treatments (like bass traps).

Regards,
Wayne

**Moonfly** · **10-19-09, 05:53 AM**

Cheers guys,

Thats cleared that up. So other than using my ears, there no real way for joe public to measure harmonics variations of a sub?

**Wayne A. Pflughaupt** · **10-19-09, 11:09 AM**

You'd probably get better information by posting your question for the guys at our DIY Subwoofer Forum. We're mainly equalizer and room tuning geeks here.

Regards,
Wayne

**Kevin_Wadsworth** · **10-19-09, 12:53 PM**

Okay, can anyone explain this one to me? I was measuring the response of the subwoofer compared to subwoofer + speakers, and for kicks I did both the left and right channel with the speaker unplugged. (i.e. soundcard from REW was plugged into the L analog in for one test and the R analog in for the other, but in both tests the speaker amps were not on)

On the left channel, there appears to be a huge null at 50 Hz. Why would these two graphs be different when it should be the same signal to the same subwoofer in both cases?

**Kevin_Wadsworth** · **10-19-09, 01:09 PM**

Okay, and here is one more graph/question. Thanks for the advice on recalibrating between each measurement when trying to do a subwoofer compression test. I've attached my results.

I'm curious if I found a limitation with my sub or the pre-pro (Emotiva LMC-1). I ran the first test with the volume at "55" and increased by 5dB per test. When I ran the calibration, each time the SPL increased by ~5 dB, except for the last two tests (75 and 80, which is the top end of the LMC's dial), when the volume barely increased during the calibration. Correspondingly, the volume of these two sweeps barely increased (taupe and teal lines).

However, I conducted a couple more sweeps, increasing the output level from REW until it clipped during the last (purple) sweep. Which seems to indicate that the sub had some more headroom left. Any thoughts on what this means?

**Wayne A. Pflughaupt** · **10-19-09, 02:21 PM**

I'm not sure that REW is the right tool for testing headroom or compression...

Regards,
Wayne

**brucek** · **10-19-09, 02:22 PM**

Quote:

Why would these two graphs be different when it should be the same signal to the same subwoofer in both cases?

The two graphs should be the same. I have no idea why they wouldn't be other than some hitch in the sweep during the test.

Quote:

Any thoughts on what this means?

Sorry, I'm not really sure what your question is.

Your sub appears fairly well behaved at higher SPL readings.

brucek

**Wayne A. Pflughaupt** · **10-19-09, 03:43 PM**

Quote:

brucek wrote:

The two graphs should be the same. I have no idea why they wouldn't be other than some hitch in the sweep during the test.

Yeah, and other thought that occurred to me is that there might some EQ filters in place on the L channel that shouldn't be there...

Quote:

Kevin_Wadsworth wrote:

Thanks for the advice on recalibrating between each measurement when trying to do a subwoofer compression test.

However, I conducted a couple more sweeps, increasing the output level from REW until it clipped during the last (purple) sweep. Which seems to indicate that the sub had some more headroom left. Any thoughts on what this means?

I think a better headroom test would be to simply use a broadband pink noise signal and a SPL meter. Just increase the volume until the SPL no longer rises, or the sub starts making rude noises, or whatever the obvious indicator is that you're out of headroom. If there is a compressor or limiter involved, at some point the sub level will no longer increase with additional volume control travel. That's where the compression is kicking in.

Regards,
Wayne

Advertisement