I don't think I understand group delay. Can someone give my the "Group Delay for Dummies" version of what it means and how it effects loudspeakers?
For example, I have attached the group delay plot from WinISD for the same driver (Dayton 7" Aluminum Cone) with 5 different volumes (the 5 different alignments available in WinISD).
What should I learn from the group delay of each alignment?
For example, I have attached the group delay plot from WinISD for the same driver (Dayton 7" Aluminum Cone) with 5 different volumes (the 5 different alignments available in WinISD).
What should I learn from the group delay of each alignment?
An externally hosted image should be here but it was not working when we last tested it.
Group delay is nothing more than the phase characteristic corrected to frequency.
In simple words, When a loudspeaker has a group delay of 2mS at 200 Hz and a group delay of 20 mS at 40 Hz (not uncommon for reflex enclosures) this means that when two tones of 40 and 200 Hz are played at the same time the 40HZ tone whil come out 20-2=18mS later than the 200Hz tone. in other words, the low frequencies are walking behind the low-mid frequencies in this example.
hope this helps.
In simple words, When a loudspeaker has a group delay of 2mS at 200 Hz and a group delay of 20 mS at 40 Hz (not uncommon for reflex enclosures) this means that when two tones of 40 and 200 Hz are played at the same time the 40HZ tone whil come out 20-2=18mS later than the 200Hz tone. in other words, the low frequencies are walking behind the low-mid frequencies in this example.
hope this helps.
Thanks Sjar-
In that case, using the sample of the graph I posted. Looking from 40 Hz on up, the Light Cyan curve (the real peaky one) would be "better" from a group delay standpoint?
How audioable is group delay? Is there a time lapse that starts becoming objectionable? For example, will the group delay difference at 40Hz be noticeable between the light cyan curve and the red curve?
In that case, using the sample of the graph I posted. Looking from 40 Hz on up, the Light Cyan curve (the real peaky one) would be "better" from a group delay standpoint?
How audioable is group delay? Is there a time lapse that starts becoming objectionable? For example, will the group delay difference at 40Hz be noticeable between the light cyan curve and the red curve?
Keep the group delay(GD) below 10ms in the 30-40hz+ range if possible.
Be aware that ported designs generally have worse GD than sealed at lower frequencies because of the 180degree phase shift at the tuning frequency.
Group delay is really only problem for the lowest frequencies, which one of the reasons why I much prefer sealed with a LT.
Be aware that ported designs generally have worse GD than sealed at lower frequencies because of the 180degree phase shift at the tuning frequency.
Group delay is really only problem for the lowest frequencies, which one of the reasons why I much prefer sealed with a LT.
I'm sorry, somehow I don't see your attacement.
There ar lot's of discussions of how audible the group delay is, but it's controversial. Every loudspeaker suffers from group delay. When you really want the least group delay just build the biggest enclosure you can and this problem is solved. But that means you won't get any bass from your system anymore. So it's all about compromise. The only choose the least worst alligment
There ar lot's of discussions of how audible the group delay is, but it's controversial. Every loudspeaker suffers from group delay. When you really want the least group delay just build the biggest enclosure you can and this problem is solved. But that means you won't get any bass from your system anymore. So it's all about compromise. The only choose the least worst alligment
ShinOBIWAN said:
Group delay is a problem over the entire spectrum, crossovers also have group delay.
The higher the slope the worse it gets, and also the crossover type.
For instance, 2nd order Bessel has much less group delay than 2nd order Butterworth.
It is clearly audible in the midband and treble.
carlosfm said:
Oh I agree its definitely audible around the XO points, during the time I spent tweaking my design I did a fair amount of goofing around with the digital delays, use the flexibility to your advantage though and it can be near ideal. Out of all the designs I've seen posted on here from the like of Tony, John et al. none whatsoever have anything approaching objectionable GD. So like I said you've got to mess up the design to do so.
Depends really what your format of choice is. Digital, I say keep it pristine, if you really want to alter a digital signal pass it through some DA conversions then through a bunch of passive analogue components.
For me its keep it bit perfect all the way upto the digital XO, use a high quality masterclock, then use proven studio grade FIR filtering(that's essentially been used throughout most of the digital recordings you've listened to BTW). Then and only then impose a single high quality DA conversion using an Apogee DA16X and then pass the analogue out to the amps which are directly coupled to the speakers(no passives here).
Using this formula, I'm closer to my perfect speaker than anything I've heard so far out the many speakers and solutions I've heard.
For digital is absolutely perfect in as far as a real world situation goes. The downside is if the recording is bad it really does sound terrible - no mp3's here
If you've got analogue sources, look elsewhere though.
Neither is a great speaker just tweaked out entirely by ear these days. The measurement software helps you achieve goals and direction and the results of which you then refine to your taste by ear.
ShinOBIWAN said:For digital is absolutely perfect in as far as a real world situation goes. The downside is if the recording is bad it really does sound terrible - no mp3's here
If the bad recording doesn't sound terrible, there's something wrong with the system.
Dacwise, I prefer NOS, with a good clock and (compensated) analog stage.
But with digital crossovers you can't help processing the signal... not what I would call bit perfect.
It also has delays, although less noticeable, but oversampling IS noticeable, it increases jitter.
Digital crossovers may be a good option, depending on the system.
Again, the less artifacts on the digital and analog domain, the better.
A good dac must have a clean, short, direct path, no need to multiplicate stages.
It is not necessarily better to use a digital crossover.
A good dac, preamp and crossovers (active or passive) can make it quite well.
If you really need steep slopes, that's where digital crossovers work better.
ShinOBIWAN said:
That's right, I do.
carlosfm said:
I agree, most of the digital solutions are lacking. What I use is completely transparent though. Like I said before the filters are FIR and the plugin is popular throughout studio's - Surely if you buy a CD mastered using these EQ filters and then went on to evaluate various speakers... well like I said they're very good with certainly much better figures than a passive network and especially phase.
The processing delay is completely mute with music. Tested the FIR filters with an impulse repsonse - very fast with no ringing. Oversampling is also mute for me thanks to locking it throughout all digital stages. If its 96Khz it stays that way through the digital domain, even through the filters, same goes for 192Khz, 44Khz etc.
A far less desirable solution would be something like the DCX2496 which did a real good job of mashing up the signal.[/QUOTE]
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.