I am using winisd to adjust different box volumes and port tunings and i am noticing that the graphs for phase plot and group delay change as well as the freq response graphs and SPL graph...what do the phase plot graphs and group delay graphs show me? what do they mean?
Group delay means the time in milliseconds (usually) the signal at that frequency is delayed versus the rest of the frequency spectrum.
so is it usually a good idea to try and get that graph to be relatively flat, seeing as having certian frequencies delay might interfere with good sound?
how about phase plot? is that important?
how about phase plot? is that important?
Hi,
from The Unofficial Speaker Workshop Manual 2.0
Group Delay = - (Phase @f2- Phase @f1)/(f2-f1).
Group Delay is derived from the slope of the phase response in the frequency domain. Group delay really refers to the relationship of phase of a small group of frequencies in a narrow range. It reports the data based upon a central frequency in this range but is looking at the change in phase vs. change in frequency in this range.
In a perfect world, there should be no delay from one frequency to the next so the slope would be zero. In general, the group delay decreases with increasing frequency. A flat Group Delay curve assures that all frequencies are arriving simultaneously. Group Delay provides some information about the Q of the filter action and hence the ringing of the loudspeaker at Fs. Up to Q values of about 0.6, there is no peak and therefore no ringing is noted (A Qtc of about 0.5 would be considered Transient Perfect; note that lower Qtc’s will have a better transient response but will sound more “thin” as a general rule. Increasing the Q will lead to more peaking with a result of more ringing. It may sound more “full bodied” than the original signal source (which some people like though it is not accurate reproduction). Speakers with a high peak in the Group Delay (at low frequencies) tend to sound “boomy”.
If Group Delay is large enough it will be audible Blauert and Laws [May 1978] report threshold levels in the range of 2 msec @8kHz; 1 msec @ 2kHz; and 3.2 msec @ 500 Hz). John Murphy has taken this information one step further. He “normalizes” Group Delay into cycles of delay calculated by dividing the group delay by the period of the frequency at each point (or the frequency times the number of seconds of delay); he feels that this may be a more accurate comparator. He notes that below 2000 Hz, 1 - 2 periods of Group Delay (or group delay * frequency products of less than 1000 to 2000) are likely to be inaudible or barely audible. This would equate to 1 ms at 2000 Hz or (through extrapolation) 10 ms at 200 Hz. The audible threshold reported by John Dunlavy [1997] is about 30 degrees of phase shift per octave of Group Delay (the product of Group Delay and Frequency would be about 167 or a cycle of delay of .167 periods). Dan Wiggins (Adire Audio) recommends targeting a Group Delay and Frequency product of less than 400 (or a “cycle of delay” of 0.4). Other authors recommend that the Group Delay should be down to about 10 ms by 100 Hz. For ported systems, the maximum Group delay occurs around the box tuning frequency. For a given woofer and box size, lowering the Fb will increase the peak value of the group delay but this will also reduce the Group Delay at higher frequencies.
The perception of Group Delay is related to frequency and you cannot compare the group delay of two separate systems if they have different frequency responses, distortion characteristics, and out of band influences as these response differences will have much more of an audible impact than will the Group Delay. Group Delay from crossovers will be more of an issue as your ears are much more sensitive in the mid frequency ranges to it. A group delay of .2 milliseconds at 2000 Hz will be as audible as one of 2 milliseconds at 200 Hz.
Hope that helps,
Jay
from The Unofficial Speaker Workshop Manual 2.0
Group Delay = - (Phase @f2- Phase @f1)/(f2-f1).
Group Delay is derived from the slope of the phase response in the frequency domain. Group delay really refers to the relationship of phase of a small group of frequencies in a narrow range. It reports the data based upon a central frequency in this range but is looking at the change in phase vs. change in frequency in this range.
In a perfect world, there should be no delay from one frequency to the next so the slope would be zero. In general, the group delay decreases with increasing frequency. A flat Group Delay curve assures that all frequencies are arriving simultaneously. Group Delay provides some information about the Q of the filter action and hence the ringing of the loudspeaker at Fs. Up to Q values of about 0.6, there is no peak and therefore no ringing is noted (A Qtc of about 0.5 would be considered Transient Perfect; note that lower Qtc’s will have a better transient response but will sound more “thin” as a general rule. Increasing the Q will lead to more peaking with a result of more ringing. It may sound more “full bodied” than the original signal source (which some people like though it is not accurate reproduction). Speakers with a high peak in the Group Delay (at low frequencies) tend to sound “boomy”.
If Group Delay is large enough it will be audible Blauert and Laws [May 1978] report threshold levels in the range of 2 msec @8kHz; 1 msec @ 2kHz; and 3.2 msec @ 500 Hz). John Murphy has taken this information one step further. He “normalizes” Group Delay into cycles of delay calculated by dividing the group delay by the period of the frequency at each point (or the frequency times the number of seconds of delay); he feels that this may be a more accurate comparator. He notes that below 2000 Hz, 1 - 2 periods of Group Delay (or group delay * frequency products of less than 1000 to 2000) are likely to be inaudible or barely audible. This would equate to 1 ms at 2000 Hz or (through extrapolation) 10 ms at 200 Hz. The audible threshold reported by John Dunlavy [1997] is about 30 degrees of phase shift per octave of Group Delay (the product of Group Delay and Frequency would be about 167 or a cycle of delay of .167 periods). Dan Wiggins (Adire Audio) recommends targeting a Group Delay and Frequency product of less than 400 (or a “cycle of delay” of 0.4). Other authors recommend that the Group Delay should be down to about 10 ms by 100 Hz. For ported systems, the maximum Group delay occurs around the box tuning frequency. For a given woofer and box size, lowering the Fb will increase the peak value of the group delay but this will also reduce the Group Delay at higher frequencies.
The perception of Group Delay is related to frequency and you cannot compare the group delay of two separate systems if they have different frequency responses, distortion characteristics, and out of band influences as these response differences will have much more of an audible impact than will the Group Delay. Group Delay from crossovers will be more of an issue as your ears are much more sensitive in the mid frequency ranges to it. A group delay of .2 milliseconds at 2000 Hz will be as audible as one of 2 milliseconds at 200 Hz.
Hope that helps,
Jay
- Status
- Not open for further replies.