I agree wholeheartedly. In an ideal situation, the mic should be at the listener's distance. The designer will also have to decide on the axis, whether on tweeter or between the tweeter and the woofer and the summation of the the two sounds as in the JBL.
I have generated the wavelet in accordance to your settings. I was trying to get a wavelet like this before but didn't play around with the setting enough. Takes for the tip.
This doesn't look too bad, can we compare it to the same settings used on your prior crossover without the delay and tweeter inversion?
It does show the STEP shape wasn't a coincidence. While the sum might not be perfect (a little lower SPL at the crossover frequency itself) it's time response at the measurement position looks quite nice. This does not tell us anything about the off-axis response yet. Off-axis response would determine most of the reflected energy and might influence tonal balance.
Due to the design axis at tweeter level, you won't get that same sum at that same tweeter level at larger distances. Which is why I would propose to measure (at the same distance as before) between woofer and tweeter height to determine the actual crossover. By doing that we can move further from the speaker and still keep the hand-over just about the same. (as in the example from JBL)
It does show the STEP shape wasn't a coincidence. While the sum might not be perfect (a little lower SPL at the crossover frequency itself) it's time response at the measurement position looks quite nice. This does not tell us anything about the off-axis response yet. Off-axis response would determine most of the reflected energy and might influence tonal balance.
Due to the design axis at tweeter level, you won't get that same sum at that same tweeter level at larger distances. Which is why I would propose to measure (at the same distance as before) between woofer and tweeter height to determine the actual crossover. By doing that we can move further from the speaker and still keep the hand-over just about the same. (as in the example from JBL)
Last edited:
Maybe this exercise shows why the Synergy horn from Tom Danley is such a cool concept.
By using a horn with band pass ports to get the mid energy in there, he puts the driver spacing very close together. Meaning this strange contraption can keep it's output in phase under all usable angles of that waveguide. Very clever!
I look at it as a MTM or WMTMW with optimal qualities due to that close driver spacing.
With the bigger spacing we have here between woofer and tweeter we will have a smaller area vertically where the tweeter and woofer will remain in sync.
A WAW or FAST can keep driver spacing relatively closer together due to the lower crossover point.
All of these are things to ponder about
.
By using a horn with band pass ports to get the mid energy in there, he puts the driver spacing very close together. Meaning this strange contraption can keep it's output in phase under all usable angles of that waveguide. Very clever!
I look at it as a MTM or WMTMW with optimal qualities due to that close driver spacing.
With the bigger spacing we have here between woofer and tweeter we will have a smaller area vertically where the tweeter and woofer will remain in sync.
A WAW or FAST can keep driver spacing relatively closer together due to the lower crossover point.
All of these are things to ponder about
.
Hi wesayso, you may be right, wiggles in mag and phase surely show up in the impulse response.
But IME, the wiggle at 10kHz isn't near big enough to cause the original wonky impulse. I agree there is still room to improve the handover ..all the measurements are saying that I think...but the box does look pretty good
Hi Michael, nice work. You know, your focusing in on time/phase alignment, may just be showing the design tradeoffs JBL chose..... perhaps flatter mag response over slight transient response improvement. I'm assuming your original traces are as per factory?
I saw you said your lab is cramped... a trick I just recently learned is to lay a speaker on the floor on it's side and put the mic on the floor too. I use a mic distance 1-2m away and block the back of the speaker up so it tilts on axis to the mic. Gives a somewhat decent surrogate "free space" measurement. Then compare to outdoor meas when you get a chance, ... saves alot of time and effort doing fundamental tuning IMO.
Yes, a smooth curve. In these graphs we are only looking at timing.
You could turn off "Normalise to peak at each frequency" and you will see a slightly different picture.
It will show where the sound is softer compared to the loudest parts.
Doing that for both graphs could be interesting too. It will show the dip at 10 kHz but will also show differences at the crossover.
All of this data is derived from that IR, it is the same data, just viewed differently. Making use of this data to look what happens helps to get a feel for it. Use the time (span after peak) to see what changes when you increase it. It should show reflections turning up.
Even the waterfall is still the same data, viewed with other sets of variables.
This makes it possible for us to examine what we really have.
The frequency resolution was set to smooth things out, you still have one step left to show a less smoothed version too.
Smoothing makes for pretty graphs, sometimes we want to see the warts and all.
You could turn off "Normalise to peak at each frequency" and you will see a slightly different picture.
It will show where the sound is softer compared to the loudest parts.
Doing that for both graphs could be interesting too. It will show the dip at 10 kHz but will also show differences at the crossover.
All of this data is derived from that IR, it is the same data, just viewed differently. Making use of this data to look what happens helps to get a feel for it. Use the time (span after peak) to see what changes when you increase it. It should show reflections turning up.
Even the waterfall is still the same data, viewed with other sets of variables.
This makes it possible for us to examine what we really have.
The frequency resolution was set to smooth things out, you still have one step left to show a less smoothed version too.
Smoothing makes for pretty graphs, sometimes we want to see the warts and all
.
Smoothing makes for pretty graphs, sometimes we want to see the warts and all
LOL!
Yes.
We need to.
Hi Mark, if Michael is willing to share the tweeter measurement and upload it somehow, I might be able to prove my theories about that 10 kHz dip.
Emailing me the tweeter measurement would do too: wesayso [at] hotmail [dot] com.
I'll show what I would do to find out. The up/down peak would remain. I'm talking about the smaller peaks after the main signal. It probably is diffraction related from not having the tweeter flush. Do we have other sample measurements using this tweeter somewhere? Anyone using the same tweeter?
The picture I remember seeing seemed to have a sharp cut-off of the waveguide. Not smoothed round curves.
Emailing me the tweeter measurement would do too: wesayso [at] hotmail [dot] com.
I'll show what I would do to find out. The up/down peak would remain. I'm talking about the smaller peaks after the main signal. It probably is diffraction related from not having the tweeter flush. Do we have other sample measurements using this tweeter somewhere? Anyone using the same tweeter?
The picture I remember seeing seemed to have a sharp cut-off of the waveguide. Not smoothed round curves.
Last edited:
Turn off the "Normalise to peak at each frequency" for another big difference and look at smoothing at 1/24. Check one at a time to see differences.
I'd be more interested in turning normalisation off.
Last edited:
I don't know if this helps. There's a burst of excess energy at around 10kHz. I think that's diffraction caused by not flush mounting my tweeter.
I will email you my REW file. Have to leave soon. It's cold (-8F) outside.
No Smoothing applied in the FR measurements.
Wavelets at 1/12 resolution.
Wavelets at 1/24 resolution.
I assume these are from the original setup? The one without delay?
Maybe we need to look at 3 ms span after peak. It was clear from other plots the first reflection effects were visible after ~3 ms.
No, it is not.
The information is relevant, but not factual.
Actually, it is possible to hear 0.25 micro seconds of time delay.
This was shown by the late Richard Heyser.
he started this whole time alignment thing.
Can you back up your denunciation of the widely cited work of Blauert and Laws by describing how Heyser measured that .25 us perception threshold with a group of people in double-blind testing? Were his tests published in a peer-reviewed journal as prestigious as the Acoustical Society of America like B&L?
While Heyser's "day job" was with NASA, perhaps you could identify any qualification he might possibly have had as a human-factors researcher we can trust.
B.
Last edited:
Thanks. I'm just exploring deeper the concept of time-alignment. I haven't read the JBL article yet but I do agree with you that all speaker designs are tradeoffs in one way or another. There's no such thing as a perfect speaker.
I don't quite understand what you mean by I'm assuming your original traces are as per factory?
All my measurements are made with drivers mounted onto the box. I do not use factory plots because they are all mounted on IEC panel.
Thanks for the tip. I simply don't have the space for Ground Plane measurements. My lab is very cluttered. It's not only speakers but many power amplifiers too.
There use to be a magazine called "Audio"
He was the tester of loud speakers for them.
There, you can read about him.
you can also find his work in the JAES.
(I made an error. I meant to say 0.25 inches).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.