Hi, I’ve heard that if the crossover network is designed properly, reversing polarity of one driver give a deep null at the crossover region on the frequency response. However, what if one does it without measurements, just voicing by ears? Can we detect the deep null by listening? Also, does deep null have any advantages? Are there any designs that are benefit by deep null?
A deep null does not indicate a good crossover, it just shows that they are in phase at the crossover frequency.
Not everyone uses this techique. Some typical reasons it is done are... When you are sweeping a microphone by hand trying to work out what's going on, and when using Xsim and the user hasn't turned on individual phase responses.
Not everyone uses this techique. Some typical reasons it is done are... When you are sweeping a microphone by hand trying to work out what's going on, and when using Xsim and the user hasn't turned on individual phase responses.
From what I've noticed, for broadband content (music), a HF that is not time-aligned sounds 'closer' than (precedes) the woofer, and the effect goes away when it is appropriately delayed. Please note that I'm talking about an active crossover here.
Yes, using test tones / sweeps and it would not allow you to compensate it no matter how much EQ you throw at it. It is just never enough.
It maybe difficult to detect this null using regular music because:
- It may or may not contain the frequency of the null
- And even if it does, the effect could be masked by other frequencies.
Unlikely.
Last edited:
By my personal experience, a deep null in the midrange region will simply give you the listening impression of a scooped-mid tone, for example.
T
T
This is interesting. Because I have a Fostex driver FE108 in a horn housing. The speaker as a dome attached with three small pins. The frequency sweep shows a total change of phase around 4 KHz. Other modern designs have a similar phase change.
The mid high tone has a very curious sound (to my dislike of course). The speaker sweep shows a 3-5 dB rise there, but the sound is different, some voices just seem to lack body!
Maybe you re marks describes it,
Hi, yeah, spatial effect and adjusting delay in DSP real time allows to notice it.
I noticed listening distance needs to be sensible to hear it. There doesn't seem to be any difference if listening other side of the room, at least with my system. I do not know if this is common for every situation or some specialty with my system.
If tweeter polarity is flipped to make great cancellation on-axis the sound becomes spacious with my system, again listened close enough, another spatial effect. Usually, when crossover works as it should, there is constructive interference on-axis and destructive interference off-axis (vertical) when transducers are not coincident. The situation flips with polarity, destructive interference on-axis and constructive interference off-axis. This changes ratio of direct / reflected sound around crossover frequency which makes the impression of spaciousness, more room sound than direct sound.
Which brings a thought that this setup could be used to listen early reflections, and / or wonder about frequency bandwidth audibility of the crossover.
Last edited:
I have to add, that with 3-way speaker I have there seems to be effect in height localization with the low crossover and it's timing, "bass" height seems to adjust with delay. I think it's very likely that first vertical reflections have effect on this; as delay is adjusted the vertical off-axis nulls turn to another angle and if there is significant reflection from ceiling/floor, or null toward ceiling/floor, it can affect the height perception. Null toward first specular reflection on ceiling would make the sound appear lower in height than higher frequency content. When timing is right, or balanced, how ever one wants to look at it, right for the situation, then the height of "bass" seems to be same as with all the other content. Similar effect is with the high crossover I think. Anyway, perhaps there is multiple things at play but in general errors in timing I hear as spatial effect with relatively close listening distance. Listening too far and there is about no difference, reflections of room dominate impression, a hazy blob of sound regardless of delay settings.
Last edited:
Hi, 3-5db difference breaks system balance on itself already. If there is resonance with the horn it could further contribute to the sound and be responsible of weird phase stuff. Nevertheless, I believe you it probably won't sound very good with this kind of details in measurements.
I've observed that when a system is out of whack its not possible to get good balance, it always seems to be wrong tune it one way or another due to some resonance or frequency response peak or something taking the attention. When the system is reasonably problem free, tuning the balance is now easy and it's possible to make very small adjustments to further refine the balance. I'm using DSP, which makes this kind of observations possible, because the balance can be adjusted at will without extra cost.
I'm not sure how easy it is to hear any of this with passive crossovers, if changing parts/perspective takes time or is impossible altogether. One would never hear alternative and thus never learn there is something not quite right. With DSP instant comparison is possible, which makes the stuff audible, two audible perspectives on some aspect, switchable at will, and one can reason with the perceived difference. With only one perspective it's not possible to gather information like this, other than by gut feel if it feels right or wrong, but it's hard to say what is it that you hear unless there is great amount of experience on listening such things. Changing perspective, like adjusting delay parameter in realtime, allows to hear such things.
Last edited:
You'll likely hear it clearly using a relatively slowly upswept sine wave signal on-axis in near field, which in my experience is the single most revealing subjective test signal that can be used. An upswept sine signal can easily help to detect multiway loudspeaker polarity issues and other driver-induced issues (like with complex diaphragm designs incorporating wizzer cones flapping out of phase with the larger main cone on full-range drivers).
One can hear a lot while listening intently on-axis during an upsweep. Depending on the steepness of the electrical or digital crossover filters used, you can hear different disturbances during the upsweeps.
If you like to listen to badly done loudspeakers, then I suppose, yes. Otherwise, it's always something to be avoided--deep nulls in the resulting loudspeaker output on-axis.
There are many loudspeakers that use apparently out-of-phase (i.e., polarity) midrange and/or bass drivers, but by using appropriate digital or all-pass delay in conjunction with that polarity reversal, will yield a total in-phase summed response through the crossover interference band, and even perhaps a half-cycle faster damping step response. But it takes acoustic measurements to do this effectively as in all other cases.
Chris
Last edited:
I hope I have understood the question correctly:
I use the polarity reversal as a test whether the drivers of a multi-way loudspeaker are phase coherent at the crossover frequency with correct polarity. If the polarity is reversed, there should be a maximum drop and the individual frequency responses of the drivers should not show more level than the sum frequency response at any point.
Of course the correct polarity is restored afterwards. 😎
However, I don't do this by ear, but with measurement equipment.
Many greetings,
Michael
Translated with DeepL
I didn't try far away or the other side of the room, but got the feeling of the treble sounds hitting first and registering first in the brain when the bass sound is the one that is originally meant to arrive first. In other words, simply changing the delay of the HF changes the tonality of the overall speaker.
Though I was listening to more of the direct sound, what you say probably applies even in my case, as the treble reverberations are that primarily contribute to the big sound (spaciousness) of a room, as perceived by our brains.
Hi,
Prescott try replay a pink noise on headphone and then on a loudspeaker with destructive interference at play ( null happening from phase mismatch between drivers for any possible reason) and move around your loudspeakers it'll soon appear if you can hear it or not and how it sound.
Best bone outside ( far from any boundary) and with loudspeakers pointing to the sky if you can. Very informative test about DUT's directivity behavior if you can perform it.
As pointed previously it can help identify issue in electronic filter implemented too.
Tmuikku, Newvirus2008,
It's a 'trick' used by some mastering engineers to 'widen' sound. I find it a bit boring as idiosyncratic but it's based on pssychoacoustic research.
Behringer had a unit dedicated to it in the 90's ( can't remember it's name atm) and things like BBE sonic 'maximisers' and variations 'embelishing' box use this tricks on saturated content ( harmonic content generator as politely called by brands 🙂 ) mixed back to original signal: it 'widenth' content ( even for mono source).
Prescott try replay a pink noise on headphone and then on a loudspeaker with destructive interference at play ( null happening from phase mismatch between drivers for any possible reason) and move around your loudspeakers it'll soon appear if you can hear it or not and how it sound.
Best bone outside ( far from any boundary) and with loudspeakers pointing to the sky if you can. Very informative test about DUT's directivity behavior if you can perform it.
As pointed previously it can help identify issue in electronic filter implemented too.
Tmuikku, Newvirus2008,
It's a 'trick' used by some mastering engineers to 'widen' sound. I find it a bit boring as idiosyncratic but it's based on pssychoacoustic research.
Behringer had a unit dedicated to it in the 90's ( can't remember it's name atm) and things like BBE sonic 'maximisers' and variations 'embelishing' box use this tricks on saturated content ( harmonic content generator as politely called by brands 🙂 ) mixed back to original signal: it 'widenth' content ( even for mono source).
IF, the xover is fully in phase when polarities are correctly matched, it should be very easy to hear a difference with pink noise when you swap one driver's polarity. Just use one speaker...should be able to hear it anywhere, even the next room.
If you can't hear a difference, it's an indication the xover is not very well phase aligned.
OK, so say a difference can be heard with pink.....but without measurement, which one is correct?
When there is a null from a swapped polarity, I find after some period of listening, sometimes immediately, sometimes after a day or two,,,....,
well, some songs I routinely use for test songs, just sound somehow off....
This in on comparing different 5-ways, and while trying different xover points, ...........I sometimes screw up processing settings and reverse a polarity.
Xover points are typically in the zipcodes of 100, 300, 900, and 4000Hz. A null in any of them makes its annoying presence known after a while.
Have to admit, I find it kinda retroactively pleasing to know something is off, and then to troubleshoot and measure a section's polarity is wrong.
Makes me think my ears actually work Lol
If you can't hear a difference, it's an indication the xover is not very well phase aligned.
OK, so say a difference can be heard with pink.....but without measurement, which one is correct?
When there is a null from a swapped polarity, I find after some period of listening, sometimes immediately, sometimes after a day or two,,,....,
well, some songs I routinely use for test songs, just sound somehow off....
This in on comparing different 5-ways, and while trying different xover points, ...........I sometimes screw up processing settings and reverse a polarity.
Xover points are typically in the zipcodes of 100, 300, 900, and 4000Hz. A null in any of them makes its annoying presence known after a while.
Have to admit, I find it kinda retroactively pleasing to know something is off, and then to troubleshoot and measure a section's polarity is wrong.
Makes me think my ears actually work Lol
Last edited:
Agreed, As of course, frequency magnitude response must be correct through xover as well.
What a deep null does indicate, is that drivers are likely both in phase and in time alignment, through xover....
which are definitely part of the requirements for a xover to be considered good, imso.
It would take a keen eye. If I wanted to see time I'd rather use another method 😉
Last edited:
Agree Allen but if there is a time alignement issue it will be revealed this way ( of course the comment 'eq should be effective though' is not true as it is z offset and phase related) :
https://www.audiosciencereview.com/forum/index.php?threads/tannoy-system-600-speaker-review.11919/
https://www.audiosciencereview.com/forum/index.php?threads/tannoy-system-600-speaker-review.11919/
I guess you are kinda cryptically referring to the fact that visual measurement nulls can occur at every full period around the correct one, and can be hard to eyeball which is deeper. If so, fully agree.
And fwiw, I'd never advocate using measured nulls as a way of developing xovers.
I'll only say, if you can't hear a null with pink, when swapping one section's polarity, xover isn't very good.
Best method I know for assuring both phase and time alignment, continues to be the time-honored 'phase trace overlay method'.
Simply get the phase traces of each section measured separately, to lay on top of each other throughout the xover summation range as best as possible, and bingo.