Subwoofer response - what's important?
Brian Steele - 22nd July 1997, Updated 14th February 1998

NOTE: Before embarking on a subwoofer design with a frequency  response that's accurate to 0.1dB within its passband, you might want to consider the following information!

In order of importance, I'd say that the following factors play a major role in the response of a subwoofer:

• Your Listening Room
  Your room is going to play a MAJOR role in the final response of your subwoofer, more so than slight variations in cabinet size, port length or other factors directly affecting the design. If your room displays horrible resonance problems in the bass region, then no amount of tweaking the subwoofer's output will help you - you need to fix the room! Further on down this page, I will illustrate this fact by actual response measurements done with a listening room with definite resonance problems.
• Positioning
  The position of your subwoofer in your listening room is going to have a major impact on the the results you get at your normal listening position. Experiment by using different positions for your subwoofer and you'll see what I mean. You'll probably find that the best positions for your subwoofer are either (a) in a corner of the room, or (b) right next to your seat.
• Output Level / Power Handling
  Better subwoofers are capable of higher, undistorted levels than inferior ones.
• Distortion
  We are less sensitive to distortion in the bass frequencies than elsewhere in the audio spectrum, but it is still important to minimize distortion as much as possible in subwoofer design as our ears are less sensitive to the lower bass frequencies. For example, at a fundamental frequency of 15 Hz, 20% third harmonic distortion will probably be more audible than the fundamental itself.  I suspect that a good number of claims made about the audibility of transient response (see below) in a subwoofer can actually attributed to the subwoofer's distortion characteristics.
• Frequency Response
  Important of course, but as far as I'm concerned, not as important as the previous characteristics. A subwoofer that supposedly has an excellent frequency response will sound crappy in a room that has bad resonance problems - there is no way around this. 1dB variations in response will be completely masked by the effects of the room.
• Transient Response
  Probably the least important of the characteristics that define the response of a subwoofer. The average person is unable to discern differences in transient response in the bass region, up to about 18ms, and that's assuming a perfect listening environment. In a normal environment, the other characteristics measured above will take precedence.

Room Response
To show the major effect that your listening room would have on the final result you get from your subwoofer, I've illustrated below some measurements of my "El Uglito" 4th order bandpass system in my present listening room, which displays horrible resonance problems at 40 Hz and other frequencies. Thankfully, I'll be moving to a different house in a few months, but then again I'll have another listening room to deal with!

The picture above shows a recording I made of a 40 Hz signal at various playback stages in the following setup:

1. Dell XPS P120c - Pentium 120 with 48 MB of RAM
2. SoundBlaster AWE32 with CoolEdit 96 (signal generator)
3. Realistic Analog Sound Level Meter (used for recording)
4. Denon PMA-250II Integrated Amplifier (25Wx2)
5. Home-built 4th order bandpass system ( two 6.5" drivers, F3~53 Hz))
6. System placed near to one corner of my listening room.

The actual WAV file is located that the following address: measure.wav

The first part of the signal is the original signal, a single 40 Hz sine wave pulse created using CoolEdit.

The second part of the signal is the result I got when I hooked the line-in and line-out connections for the SoundBlaster AWE32 together, and then played back and recorded the signal using two copies of CoolEdit. The resulting waveform shows the effect of the input and output sections of the AWE32 on the original 40 Hz signal.

The third part of the signal is the result I got when I inserted the amplifier's line-in and tape-out connections into the loop, and demonstrates the effects of the amplifier's preamplification section on the signal. There is very little difference between this and the second part of the recording - as it should be with all good amplifiers.

The fourth part of the signal is the result I got when I hooked the 4th order bandpass system to the amplifier, and recorded the results using the Sound Level Meter hooked into the line-in connection for the SoundBlaster card. The close-miked method was used for the recording to minimize room effects (the microphone was placed directly in front of the port). The resulting signal is quite different to the input, but guess what - the difference isn't very audible (check by listening to the actual recording, using the link given above). In interesting thing to note here is that the signal decays at about 52 Hz - the resonance frequency of the drivers used in this design.

The final part of the signal is the result I got using the same configuration above, but this time the Sound Level Meter was placed out in the room away from the 4th order bandpass system. The room in question has very bad standing wave problems, and the results show up in the recorded signal.

It is this final signal that's the most important here, as it basically shows that the effect of the room on the original signal is far greater than the effect of the sound generator, amplifier, or speaker on the signal. Food for thought when you're designing your sub to meet a required frequency response to the nearest 0.1 or even 1 dB - the room is going to have a far greater effect on what you hear than a 1dB variation in the frequency response of the subwoofer.

14th February 1998
The picture below is an FFT analysis of the room response (the last section of the picture above). Each horizontal line represents a 12dB interval.

Note the huge peaks at 26 Hz, 43 Hz and 66 Hz? Those are caused by room modes. The 66 Hz mode is caused by the ceiling in the room, which is 8.5 feet high. The other modes are caused by the room's length and width. There is also some funny stuff happening in the 90-120 Hz region, but the effects on the overall response are considerably less than what's being caused low down in the audio spectrum. In fact, any speaker that goes much below 70 Hz can be expected to excite these room modes, with perhaps horrendous results.

As part of my next audio project, I will be attempting to tame these room modes, by using bass traps in each corner of the room. I expect to get much better bass response once the room modes are tamed somewhat.

Brian Steele
14th February 1998

Visit the Subwoofer DIY Page at www.diysubwoofers.org