• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Are my subwoofers distorting? (Reading REW data)

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
Hello everyone,

I'm getting used to using REW with my UMIK-1, along with setting EQ curves for my 2 subwoofers using a MiniDSP 2x4HD. This is a pretty standard auto-EQ I did for my subs with a mini house curve. I didn't really do any individual tweaking yet (there are definitely a couple of places I was thinking of boosting/reducing a bit to make the response a little smoother). Before I did that though, I thought I would check the distortion, and I was surprised to see some rather high distortion around 30hz (up to about 7% THD). Am I reading this correct, and is this pretty bad distortion?

I'm a little surprised by this - I don't have the levels up in the AVR or anything and the gain on my subs is only around 10 o'clock (they're two HSU Research subs, a VTF-3 MK2 and a VTF-3 HO). The EQ did cut quite a bit out of the curve in some spots to try to flatten things out, but I don't think it boosted much yet from what I can tell. For what it's worth, I think one of the subs might have a rattle inside of it from old glue that I need to look into fixing, but I don't usually notice it. Besides that, both of them with individual measurements seem to have similar issues.

Please help me read these graphs to see if I have anything to be concerned about. I'm new to this stuff - was finally wrapping my head around reading the SPL graphs properly and understanding what curves should generally look at, but distortion is a whole new thing for me. These were generated with standard sweep measurements. I actually ran these just to see what the responses looked like and saved them a couple of days ago - it's only now that I'm looking at the distortion parts.

I uploaded screenshots of both the ALL SPL tab and the Distortion tab, but let me know if there's anything else that would be more helpful to see. Thank you for all of your help - I want to make sure this is looking OK distortion-wise before I start tweaking/boosting any portions of the response to flatten it out further.

Thank you!
 

Attachments

  • Subs 8-22 Distortion.jpg
    Subs 8-22 Distortion.jpg
    90.6 KB · Views: 227
  • Subs SPL 8-22.jpg
    Subs SPL 8-22.jpg
    91.1 KB · Views: 222

Everett T

Active Member
Forum Donor
Joined
Aug 2, 2020
Messages
201
Likes
123
7% for bass frequencies is not high by most standards FWIW. Most manufactures are fine with those numbers.
 

alex-z

Senior Member
Joined
Feb 19, 2021
Messages
312
Likes
469
Location
Canada
7% distortion at low frequencies is absolutely nothing. You could drive it up to 50% and still have trouble noticing.

Your response already looks pretty good. Rather than focusing on frequency response I recommend looking at your decay times, any problems like cabinet resonance are more likely to show up there.
 

hege

Active Member
Forum Donor
Joined
Dec 25, 2018
Messages
286
Likes
425
Location
Finland
Shouldn't see much distortion at 80dB, but might be perfectly normal since you are not measuring your subs alone. There's room floor noise, all sorts of resonances from things. You can try measuring with all speakers off to see the noise floor. Also you can use a much longer sweep/FFT length and try 5-10dB louder.

For easier reading, change y-axis to percent. I think you are mostly around 3% anyway with some resonance peaks. Which is fine.

1629779632636.png
 
Last edited:

ebslo

Active Member
Forum Donor
Joined
Jan 27, 2021
Messages
214
Likes
285
All distortion is not created equal; some harmonics are pretty benign and others can sound bad even at low levels. I can't tell from an unlabeled plot what type of distortion will sound ok vs what will sound horrible, but fortunately we have ears. I would suggest slowing the sweep so it takes at least 30 seconds to complete. Then just listen and see if any parts of the sweep sound bad (buzzy, crunchy, not smooth), or if it ever sounds like maybe more than one note is playing. If you can't hear the distortion playing a pure tone, you won't hear it playing music.
 
OP
R

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
All distortion is not created equal; some harmonics are pretty benign and others can sound bad even at low levels. I can't tell from an unlabeled plot what type of distortion will sound ok vs what will sound horrible, but fortunately we have ears. I would suggest slowing the sweep so it takes at least 30 seconds to complete. Then just listen and see if any parts of the sweep sound bad (buzzy, crunchy, not smooth), or if it ever sounds like maybe more than one note is playing. If you can't hear the distortion playing a pure tone, you won't hear it playing music.

This is a great idea - I’ll give this a try. Thanks!
 

JohnPM

Active Member
Technical Expert
Joined
Apr 9, 2018
Messages
204
Likes
591
Location
UK
I can't tell from an unlabeled plot what type of distortion will sound ok vs what will sound horrible
Harmonic traces use the resistor colour code, so 2nd is red, 3rd is orange etc. The white trace is THD. The steppy dark brown trace is the noise floor, the distortion is mostly in the noise for that measurement.
 
OP
R

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
Shouldn't see much distortion at 80dB, but might be perfectly normal since you are not measuring your subs alone. There's room floor noise, all sorts of resonances from things. You can try measuring with all speakers off to see the noise floor. Also you can use a much longer sweep/FFT length and try 5-10dB louder.

For easier reading, change y-axis to percent. I think you are mostly around 3% anyway with some resonance peaks. Which is fine.

View attachment 149195
Thanks for the heads up on this. Here is an updated graph organized by percentage and included the key this time (that was my first time saving an image from REW).

FYI - This is after a new EQ, so it might be slightly different than before. I do some distortion peaks above 10% in the latest graph, but they seem to be resonance peaks and shouldn't be a concern, right?

Thank you everyone again for your help!
Latest Distortion Graph 9-2-2021.jpg
 

abdo123

Major Contributor
Forum Donor
Joined
Nov 15, 2020
Messages
4,852
Likes
4,435
Location
Brussels, Belgium
Oh btw, if your subwoofers are ported seal the ports and the majority of the distortion will disappear.
 
OP
R

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
They’re ported - they’re variable tune subs from HSU. Each of the two subs have two ports. I’m plugging one port on each, leaving one port open.
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
1,932
Likes
1,810
Location
.de
Your latest measurement looks like more noise than anything else. Recording level seems to be way too low. Get the mic substantially closer to the sub.

Also note that the UMIK-1 is not a world-beater for noise and, typically for condenser mics of all kinds, its noise floor has a substantial 1/f component. From here:
The A-wheighted noise of the UMIK-1 (700-0756) was 29.1 dB(A) and 43.1dB(unwheighted), which is both absolutely OK (a lower noise floor would limit the max. SPL). Please find attached the measured noise-spectrum.


Also note that REW on Windows is limited to 16-bit I/O when using Java, and cobbling something together using ASIO4All may be required to make use of the full input dynamic range.
 
OP
R

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
Your latest measurement looks like more noise than anything else. Recording level seems to be way too low. Get the mic substantially closer to the sub.

Also note that the UMIK-1 is not a world-beater for noise and, typically for condenser mics of all kinds, its noise floor has a substantial 1/f component. From here:


Also note that REW on Windows is limited to 16-bit I/O when using Java, and cobbling something together using ASIO4All may be required to make use of the full input dynamic range.

Thanks - this makes a lot of sense. I’ll try redoing the measurement closer and/or perhaps at higher volume and see what it shows. I’ll also keep the noise level generally speaking in mind. Much appreciated.
 

Jukka

Active Member
Joined
Apr 24, 2019
Messages
117
Likes
60
Your measurements are good. As pointed out, it's the noise floor that rises and that's always the case towards 0 Hz. Noise floor below 20 Hz is inaudible and any distortion below noise floor is inaudible. Anyway you're doing less than 10 % THD at 20 Hz, which is great, but don't be surprised if this number goes up rapidly when increasing volume. You should test at 90 dB and 96 db, and if you dare, 100 dB assuming your subs have built-in safety mechanics. That's where the fun begins

After you've done that and looked at the results, the distortion / frequency / volume tables that subwoofer manufacturers and testers sometimes publish are much more meaningful to you!

PS. You should scale the graph differently to make it easier to read.
 
OP
R

ronnzi

Member
Joined
Aug 20, 2021
Messages
27
Likes
9
Your measurements are good. As pointed out, it's the noise floor that rises and that's always the case towards 0 Hz. Noise floor below 20 Hz is inaudible and any distortion below noise floor is inaudible. Anyway you're doing less than 10 % THD at 20 Hz, which is great, but don't be surprised if this number goes up rapidly when increasing volume. You should test at 90 dB and 96 db, and if you dare, 100 dB assuming your subs have built-in safety mechanics. That's where the fun begins

After you've done that and looked at the results, the distortion / frequency / volume tables that subwoofer manufacturers and testers sometimes publish are much more meaningful to you!

PS. You should scale the graph differently to make it easier to read.
Thank you for the additional info! I'm not worried about the distortion at this point, as everything sounds great and judging from everyone's comments, my graphs seem to be sharing normal/good results.

For what it's worth, here are two graphs from a run I just did (they're both from the same test, just with different scales to show the SPL I was running it at). I ran this at closer to 90-95db this time, and the distortion levels seem to be about the same, so I think everything looks pretty good?

Is the percentage scale on this graph a little better? I just chose fit to data.

Thanks again!
 

Attachments

  • Latest Distortion Graphs 1.jpg
    Latest Distortion Graphs 1.jpg
    114.5 KB · Views: 26
  • Latest Distortion Graphs 2.jpg
    Latest Distortion Graphs 2.jpg
    116.7 KB · Views: 27

Jukka

Active Member
Joined
Apr 24, 2019
Messages
117
Likes
60
Thank you for the additional info! I'm not worried about the distortion at this point, as everything sounds great and judging from everyone's comments, my graphs seem to be sharing normal/good results.

For what it's worth, here are two graphs from a run I just did (they're both from the same test, just with different scales to show the SPL I was running it at). I ran this at closer to 90-95db this time, and the distortion levels seem to be about the same, so I think everything looks pretty good?

Is the percentage scale on this graph a little better? I just chose fit to data.

Thanks again!
Much better now, it's nice to see 50 or 60 dB below fundamental. Percentage graph scaling is perfect! Also the last orders of harmonics don't usually offer much information, because they are very low. I usually limit my graphs to max 4th harmonic.

Percentage scaled graph can sometimes lie to you, if you don't have dB scale for comparison. One such situation is when you have a null (sharp dip in frequency response), the distance between fundamental and distortion is much shorter so there is equally sharp rise in percentual distortion, but the absolute level of the distortion is in check with the rest of the frequency graph. You don't seem to have any of those in the graph.

You do have a strange rise of 2nd harmonic at 54 Hz, which is very high. I have no idea where that's coming from, maybe a resonance in your room and furniture or just an anomaly during measuring, like a passing car or train? Same story with the THD hump, although it's not as high in the first graph. Also THD is often followed closely by another contour, which I'm not able to see. Or is there even higher order distortion going that loud? Highly unlikely.
 
Top Bottom