Hmmm....I always get that a low-pass is what causes most phase shift....and by far..
Guess what? They both do, just look at the group delay
It only "shifts" the other way around.
So one will add a shift from +180 to 0 degrees, while to other one does from 0 degrees to -180.
At Fb, the port output lags the driver by 270 degrees, 90 degrees short of 360 degrees.
This can clearly seen in STV's "hard burst" charts, and both his ported and sealed tests show the negative phase "blip" caused by the driver's response to what amounts to a square wave when the sine wave starts:
The driver's response is similar to that of a moving-coil (dynamic) microphone, it's electrical waveform lags by 90 degrees and distorts the amplitude of the initial transient:
Wired according to the Pin 2 positive standard, a condenser (capacitor) microphone's output electrical polarity and phase response is the same as the acoustical waveform.
Assuming both STV's test mics and cords were wired with the same polarity, the driver's polarity makes no difference in the test results.
Back to the bass reflex response, above Fb, the port output lags the driver's by 360 degrees, since the port output is one full cycle late, but "in phase", it reinforces the driver's output.
Art
Guess what? They both do, just look at the group delay
It only "shifts" the other way around.
So one will add a shift from +180 to 0 degrees, while to other one does from 0 degrees to -180.
What I'm trying to point out, is compare a high-pass filter's loopback delay, vs a low-pass filters loopback delay....at the same freq.
The low-pass delay will be greater for any freq within is the first 5 octaves...which is the range of the discussion in this thread.
Personally, I think group delay is one of the most BS and misunderstood specs in audio.
The only time I see it as valid, is between level regions in frequency response....where then all it represents are fixed time alignment differences between driver sections. (I think nicely explained in REW's help on minimum phase)
Otherwise I see group delay as envelope distortion, with no meaningful fixed time properties.
Whatever you decide to call it. But it's still real delay.
Which can mess a crossover up or it fix.
For example when there a delay between a tweeter and a woofer.
Group delay isn't a spec, it's again a inherent property of a system.
The higher the order, the more there will be.
That it's grossly pulled out of proportion is something I agree with.
But that's with the majority of things in audio.
I personally find group delay not that bad.
There are other things that are way worse when it comes down to mythical stories.
No ....sigh, it's not a real delay. And what havoc the idea that it does, has wreaked.
I'll say again, group delay has no validity other than between areas of flat level response, where it then validates fixed time delays. Aka ToF.
If group delay has ANY curve , shape, slope , to it....it has zero meaning in terms of constant time.
Fine for you, but for the rest of the world, schools, universities and engineers it's delay.
Not the voodoo kind, but an actual kind
It has a lot more practical use and or issues. Especially when very high order variants are being used.
In signal theory, antennas, pure math...yes group delay can cary a constant fixed time, as well as variable.
The giant mistake re group delay in audio, imho yet very strong opinion, is yet again....thinking that GD has any significant meaning other than differences in between flat regions of freq/phase response, can give any meaningful insight into fixed time response.
Check out Heyser's work on this..
Yes, GD creates envelope distortion.....but if GD varies vs freq, has curve, it is completely invalid in portraying fixed time.
And hey, the rest of the world, universities, and engineers will agree when question is appropriately framed..
Group delay as typically used by the audio world is indeed voodoo, hoodoo, BS ...near nutz.......imh&so.
well in audio it could help a bit with like a delay difference between a woofer and tweeter, by selecting an higher order filter for the tweeter, which will add some additional delay.
Besides that, if you follow all the research, papers and literature, group delay "distortion" is extremely debatable at best.
Although within certain limits that is.
But you have to use quite some wild filters to be able to create those.
Here’s a good quick summary on the physics and effects of temporal distortion:
https://www.acculution.com/single-post/2017/08/23/009-time-alignment-in-loudspeaker
@weltersys If the port’s output is 360 degrees delayed it would reinforce the drivers output as long as the signal is steady state, but what happens to a step?
https://www.acculution.com/single-post/2017/08/23/009-time-alignment-in-loudspeaker
@weltersys If the port’s output is 360 degrees delayed it would reinforce the drivers output as long as the signal is steady state, but what happens to a step?
Last edited:
Y’all got it I referred to equal loudness contours (Fletcher Munson for the old-fashioned) which describe that the average Joe hears low frequencies less loud than higher ones? The difference between 30Hz and 90Hz can be translated to about 25dB. So masking of harmonics isn’t the issue, rather masking of the base frequencies if H3 distortion levels reach 10% at 30Hz. Which sadly isn’t that uncommon due to nonlinearities coming from high cone excursion.
Distortion audibility is easy to do with REW.
First do a slow sweep with mic on the floor at rather high spl. Study distortion view
Peaks of high D3 are often mechanical - play that F as sine wave with REWs generator and try to find it's source and try to fix it - it might be just a bookshelf door resonating!
Then play eg. 30Hz with variating spl and stop when you think it has distortion. Then take a new sweep with that spl and look at distortion plot. you can also play that F only as sine and see distortion components clearly as either % or dB.
Takes some time and a quiet house... but we love diy audio and hifi as a hobby!
First do a slow sweep with mic on the floor at rather high spl. Study distortion view
Peaks of high D3 are often mechanical - play that F as sine wave with REWs generator and try to find it's source and try to fix it - it might be just a bookshelf door resonating!
Then play eg. 30Hz with variating spl and stop when you think it has distortion. Then take a new sweep with that spl and look at distortion plot. you can also play that F only as sine and see distortion components clearly as either % or dB.
Takes some time and a quiet house... but we love diy audio and hifi as a hobby!
Why more significant for the ported system?
The woofer of a sealed box system needs to move significantly more, so produces significantly more, excursion related non-linearities (BL(x), Kms(x), HD, IMD etc).
Of course in the case of a correctly designed ported system, where below the tuning frequeny, the cone excursion is limited by adequate filtering or when the frequencies below the tuning region doesn't get excited too much.
Last edited:
I'm confused to be honest. Does a port lag the driver by 90 or 270 degrees at port tuning freq? I'm sure you agree this should be a universal for any bass reflex regardless of size or tuning. I thought the port is 90 degrees lagging the driver front radiation as you said earlier in the thread. This matches perfectly with my own measurements.
STV's measurements don't look as I had expected, I can't reconcile my measurements showing the port lag as 90 degrees and his 270 degrees. Firstly we can agree the phase of the driver is inverted, but it actually looks to me like one of the two mics is inverted compared to the other. Alternatively the waveform alignment was adjusted, because there seems to be a blip of sound from the port mic before the first sound from the driver. As you said it looks like the port lags 270 degrees, but if the driver mic (or port mic) only was inverted we would see the same 90 degree port lag we get on my measurements and that would make everything click for me.
Or... I suppose it could be my setup that has one mic inverted. It's not always obvious because the pre-amp in the mic can invert. You know what I'll check right now!
I was talking about a shift in BL and what it does to the system response.
The so called excursion dip, created by the port, is only small.
But it's true that for a well tuned system, the overall excursion will be smaller.
Although you'll get port issues in return.
I think it was @markbakk wondering how that practically will look like?
Because the BL(x) and Cms(x) will be very small at a certain frequency all of a sudden.
Very good question actually, never thought about it or the potential implications (or not).
Okay, solved! One of MY mics was inverted! The port is indeed 270 degrees behind the driver, not 90.
Here is the correct image. I've also included the summation of the two waves done in software.
I think the interesting part is that we do not actually see a 'delay' from the port output except for the propagation time. We simply have a phase shift.
Here is the correct image. I've also included the summation of the two waves done in software.
I think the interesting part is that we do not actually see a 'delay' from the port output except for the propagation time. We simply have a phase shift.
Attachments
Last edited:
Well, there is a delay in the sense that the port takes time to increase its amplitude - the rise time, if you like - which correlates with the energy input to, and stored in this resonant system. It's the 'area under the graph' of the driving input compared with the resonator output. This energy is then released as the resonant tail seen in a step response test, the decay time being considerably longer than the rise time, as expected.
Yes, you might say there is a delay if you look for when the maximum amplitude peak occurs, but not if you look for when sound starts to be generated. Again, this is all defined by the minimum-phase behaviour of the combined system so not really a function of 'port or no port'.