[Paul Spencer]
Definition: The rate of change of the total phase shift with respect to frequency
[What is group delay]
| Abbreviation: GD used on this page |
'
"No, really, what is group delay ... in English please!"'
If you look at a group delay plot, it is easier to understand. On the x axis frequency is plotted, group delay in milliseconds (ms) on the y axis. For a subwoofer, you will notice that GD increases as you go down lower in frequency. So it is a frequency-dependant delay of the signal. It is clearly audible, the point is at which point does it become audible. Without some understanding, you could be fooled into thinking that subwoofer A is better than B as it has lower GD. It may be that both are below the threshold of audibility.
In subjective terms, GD is thought to be a measure of "tightness." It does not necessarily follow that "less is more." There are believed to be thresholds of audibility.
How much is audible?
This is believed to be frequency dependant.
A study by Blauert & Laws found the following values as threshold levels where GD becomes audible.
| Frequency | GD |
| 8 kHz | 2 ms |
| 4 kHz | 1.5 ms |
| 2 kHz | 1 ms |
| 1 kHz | 2 ms |
| 500 Hz | 3.2 ms |
We can extrapolate this down to bass frequencies and derive the following thresholds:
| 100 Hz | 10 ms |
| 80 Hz | 12.5 ms |
| 60 Hz | 17 ms |
| 40 Hz | 25 ms |
| 20 Hz | 50 ms |
Caution: There is no research to back up these numbers that I know of, hence they should be used with some caution. They are based on
[John Murphy's discussion of group delay] where he made suggestions on how you may extrapolate the values in the Bauert and Laws study.
GD and sealed vs vented subwoofers
It is commonly thought that sealed subwoofers sound "tighter" due to their lower GD. However, if you eq a sealed subwoofer to achieve the same response, the GD will be the same. Hence by using a equalised sealed box what you have gained is merely a smaller subwoofer with 6db less output and higher distortion. It may not sound tighter when the response is the same. There are many who feel that sealed subs are inherently superior, and that their listening comparisons all reveal that they are more "musical." There are others who feel that this relates to the response and is not a weakness of vented subwoofers. I suggest you form your own opinion on this from both theory and experience, but before you form a conclusion, compare sealed vs vented where the frequency response of each is calibrated to match in-room response.
How to deal with GD
Lower GD is always better in theory, but in practice it may not be worthwhile reducing it if you have achieved a value below what you consider audible. You can reduce it in a number of ways:
- sealed box alignment without eq boost
- vented box tuned low such as an EBS
- digital delay
The third option is an interesting one. Since it is relative and not absolute delay that is important, you can get GD down with a digital delay unit. Units such as Behringer Ultracurve DEQ2496 and Ultradrive DCX2496 can do this. You use digital delay to match GD to that of the mains at the crossover point.
Example of digital correction: Suppose your sub runs from 20 - 40 Hz then crosses to the mains. Your sub has a GD of 50 ms at 40 Hz rising up to 80 ms at 20 Hz while your mains have a much lower GD of 10 ms at 40 Hz. The sub has 40 ms higher GD at the crossover point, and this is most likely audible. If you were the delay the mains by the difference (40 ms) then the GD of the subs is effectively now much lower. As a result, the GD is now matching the mains at the crossover point of 40 Hz, and 40 ms at 20 Hz. Now the GD is below the audible threshold for the entire range. These numbers suggest that a vented subwoofer can sound "tight" and that a sealed subwoofer is not necessarily needed to achieve this goal.
"But my friend designed a vented sub with a GD of only 15 ms!!!"
Did he factor in GD added by all the filters? The GD added by the filters can easily eclipse that of driver and vent! An interesting test of the audibility of GD would be to switch a rumble filter on/off.
Further reading
[AES paper - Audibility of group delay distortions - Blauert & Laws]
[Exerpt from online Ultimate AV magazine]
"While a detailed discussion of group delay is beyond the scope of this article, Bag End asserts that lower group delay translates into audibly better bass. And indeed, the 11 other subs in this survey had measured group delays two to six times higher than that of the S21E. But no reports of controlled double-blind listening tests of time-domain issues in the subwoofer range have appeared in the likeliest places (including the Journal of the Audio Engineering Society), and the few credible reports on the audibility of group delay have not considered frequencies below several hundred hertz. For many months, I ran level-matched comparisons using dozens of challenging tracks, from soundtrack spectaculars to more subtle classical and jazz numbers, without finding a qualitative difference that I could confidently ascribe to anything other than amplitude response, distortion, and power-compression factors.�KY"
[True Audio discussion of group delay]
[Whise paper includes discussion of group delay]
