Hi all,
I have learned to do speaker measurements using Speaker workshop, a mic and an impedance jig. I have modelled crossovers and built a few two-way passive-xo speakers. They sound good. Descriptions are posted here somewhere. A lot of you have helped and guided me.
After all this, I now come to a point where I am simply unable to answer a very basic question about xo design. Why is a deep reverse null important? Am embarassed to ask this, because it seems to be something Which Everyone Knows About. And I too like to see a deep reverse null when I flip the polarity of one of the drivers while modelling the crossover.
But if I get a crossover which is giving me a sufficiently flat modelled in-phase combined SPL, but is giving me a notch of just 3-5dB when I flip polarity, does it matter, in audible terms for the guy listening to music with my speakers?
I can understand that the reverse null is a good engineering test, because it allows me to verify many things. For instance, it allows me to verify that the woofer and tweeter are in phase. When I build the crossover and flip the polarity and measure the SPL, I should be able to see the reverse null -- this tells me that the crossover has been built correctly, and the measured reality closely matches the modelled behaviour. I understand all this, but my question is: does it matter to the guy listening to the music if I get a shallow reverse null, provided that the SPL curve is flat and tonal balance is "good" (for appropriate values of "good")? Does a messed-up phase relationship between the woofer and tweeter, with a flat enough combined SPL, have an audible impact on the sound?
Please help?
I have learned to do speaker measurements using Speaker workshop, a mic and an impedance jig. I have modelled crossovers and built a few two-way passive-xo speakers. They sound good. Descriptions are posted here somewhere. A lot of you have helped and guided me.
After all this, I now come to a point where I am simply unable to answer a very basic question about xo design. Why is a deep reverse null important? Am embarassed to ask this, because it seems to be something Which Everyone Knows About. And I too like to see a deep reverse null when I flip the polarity of one of the drivers while modelling the crossover.
But if I get a crossover which is giving me a sufficiently flat modelled in-phase combined SPL, but is giving me a notch of just 3-5dB when I flip polarity, does it matter, in audible terms for the guy listening to music with my speakers?
I can understand that the reverse null is a good engineering test, because it allows me to verify many things. For instance, it allows me to verify that the woofer and tweeter are in phase. When I build the crossover and flip the polarity and measure the SPL, I should be able to see the reverse null -- this tells me that the crossover has been built correctly, and the measured reality closely matches the modelled behaviour. I understand all this, but my question is: does it matter to the guy listening to the music if I get a shallow reverse null, provided that the SPL curve is flat and tonal balance is "good" (for appropriate values of "good")? Does a messed-up phase relationship between the woofer and tweeter, with a flat enough combined SPL, have an audible impact on the sound?
Please help?
Not my area but...
The notch depth may depend significantly upon the exact placement of the monitoring microphone to get exact phase cancellation at the exact frequency that is is supposed to cross over at.
Yes? No?
The notch depth may depend significantly upon the exact placement of the monitoring microphone to get exact phase cancellation at the exact frequency that is is supposed to cross over at.
Yes? No?
Absolutely true. In fact, that's one of the reasons why some designers don't take a very deep notch very seriously -- they know that this is a function of mic positioning, and the mic position may not much match with (i) the listener's ears, and (ii) in-room phase response after taking all the reflections into account. The real-world listener in a real room can't hear the deep notch of the flipped-polarity configuration.
tcpip,
not sure I understood you correctly. Are you saying a listener
can't tell the difference between a normal polarity situation
and an inverted one ?
not sure I understood you correctly. Are you saying a listener
can't tell the difference between a normal polarity situation
and an inverted one ?
In my experience it is very audible and a good phase match (deep null) gives a much more solid image and presentation.
The null in and of itself is also important, not just with regards to the depth of the notch but also the shape of the notch too. A good notch should be deep, but also symmetrical whilst following a typical null shape.
Yes the mic position is important with respect to the notch depth, but this is also a very good indicator of how wide your primary listening lobe is. If the drivers are integrated properly at the correct frequency and have a nice deep symmetric and text-book shaped null then you can pretty much say that it will have a reasonably wide primary lobe. This means you can move your head up and down quite a bit without the drivers suddenly going out of phase and creating a large shift in tonal balance. Or looked at another way, if the drivers are properly integrated at the listening position, then looking for the reverse null will show excellent cancellation even if the mic is too high or too low. Because the drivers are integrated correctly the design therefore shows good tolerance to the mic position.
The wider the primary lobe is, the more even the power response of the loudspeaker tends to be too.
There are a few reasons why your notch might deviate from the ideal result, all of which should really be corrected if possible. The first is that simply the phase tracking between the drivers is poor. This means that while the high pass and low pass acoustic functions are very good, the phase response at the intended listening position is isn't. This tilts the primary listening lobe either up or down and makes the loudspeaker far more susceptible to the 'head in a vice' situation whilst also making the loudspeaker sound less coherent when listened to at your normal listening height.
The second reason for the null not being quite right is if the acoustic transfer functions are not anywhere near being where they should be. Sometimes this is necessary. You will see designers using asymmetric slopes as a way of bringing the drivers into phase with one another, whilst sacrificing slope integrity. If you've got a slightly shallow/steep function on one of the drivers, or have pushed the xover frequency slightly up or down, this will create a less than ideal reverse null shape, even if it is deep. What happens here is that you start altering the shape and symmetry around the primary listening lobe, often narrowing the lobe and creating a loudspeaker that is fine when listened to above the listening axis, but really shouldn't be listened to below, or visa-versa. Using a slightly asymmetric crossover is absolutely fine as it only affects things to a small degree and is necessary, but if you've really screwed up then you've got a problem. Not only does this create issues with the primary lobe and hint at the phase alignment being off, but it also shows that the drivers are producing sound where you weren't expecting them too. In other words this means your tweeter could be attenuated far less than it should be, or the region of break-up on the mid bass might not be attenuated enough.
Now while you might assume that these things would show up in the on axis 'in phase' wired drivers, the truth is it probably wont. Yes, increasing the output of a driver, by reducing the order of the xover etc, will give you a bit of tilt in the output where there's additional energy, but if the drivers aren't actually properly in phase with one another then it can quite easily be hidden. Remember, if the reverse null is pretty poor, then the summation with the polarity correct is also going to be pretty poor too. This means you will be seeing some cancellation even with the drivers wired properly in phase. You may lower the order of one of the drivers, increasing output around the xover frequency, but in doing so push the drivers further out of phase with one another such that the summed response hardly changes because more cancellation is occurring.
Seeing the 'reverse null' of a loudspeaker tells you far more about it than you'd first think and it is an excellent way of assessing whether or not the crossover has been optimally designed. Some may say it's not worth fixating on, but this is like saying it isn't worth having your car tuned up correctly such that is miss fires 25% of the time. The reason why most designers post the reverse null is because they know it shows whether or not they've done their job correctly.
Note that odd order xovers don't show a reverse null when properly designed, but just like the standard 4th order LW does, if you're getting cancellation occurring with an odd order design, it too isn't quite where it should be.
The null in and of itself is also important, not just with regards to the depth of the notch but also the shape of the notch too. A good notch should be deep, but also symmetrical whilst following a typical null shape.
Yes the mic position is important with respect to the notch depth, but this is also a very good indicator of how wide your primary listening lobe is. If the drivers are integrated properly at the correct frequency and have a nice deep symmetric and text-book shaped null then you can pretty much say that it will have a reasonably wide primary lobe. This means you can move your head up and down quite a bit without the drivers suddenly going out of phase and creating a large shift in tonal balance. Or looked at another way, if the drivers are properly integrated at the listening position, then looking for the reverse null will show excellent cancellation even if the mic is too high or too low. Because the drivers are integrated correctly the design therefore shows good tolerance to the mic position.
The wider the primary lobe is, the more even the power response of the loudspeaker tends to be too.
There are a few reasons why your notch might deviate from the ideal result, all of which should really be corrected if possible. The first is that simply the phase tracking between the drivers is poor. This means that while the high pass and low pass acoustic functions are very good, the phase response at the intended listening position is isn't. This tilts the primary listening lobe either up or down and makes the loudspeaker far more susceptible to the 'head in a vice' situation whilst also making the loudspeaker sound less coherent when listened to at your normal listening height.
The second reason for the null not being quite right is if the acoustic transfer functions are not anywhere near being where they should be. Sometimes this is necessary. You will see designers using asymmetric slopes as a way of bringing the drivers into phase with one another, whilst sacrificing slope integrity. If you've got a slightly shallow/steep function on one of the drivers, or have pushed the xover frequency slightly up or down, this will create a less than ideal reverse null shape, even if it is deep. What happens here is that you start altering the shape and symmetry around the primary listening lobe, often narrowing the lobe and creating a loudspeaker that is fine when listened to above the listening axis, but really shouldn't be listened to below, or visa-versa. Using a slightly asymmetric crossover is absolutely fine as it only affects things to a small degree and is necessary, but if you've really screwed up then you've got a problem. Not only does this create issues with the primary lobe and hint at the phase alignment being off, but it also shows that the drivers are producing sound where you weren't expecting them too. In other words this means your tweeter could be attenuated far less than it should be, or the region of break-up on the mid bass might not be attenuated enough.
Now while you might assume that these things would show up in the on axis 'in phase' wired drivers, the truth is it probably wont. Yes, increasing the output of a driver, by reducing the order of the xover etc, will give you a bit of tilt in the output where there's additional energy, but if the drivers aren't actually properly in phase with one another then it can quite easily be hidden. Remember, if the reverse null is pretty poor, then the summation with the polarity correct is also going to be pretty poor too. This means you will be seeing some cancellation even with the drivers wired properly in phase. You may lower the order of one of the drivers, increasing output around the xover frequency, but in doing so push the drivers further out of phase with one another such that the summed response hardly changes because more cancellation is occurring.
Seeing the 'reverse null' of a loudspeaker tells you far more about it than you'd first think and it is an excellent way of assessing whether or not the crossover has been optimally designed. Some may say it's not worth fixating on, but this is like saying it isn't worth having your car tuned up correctly such that is miss fires 25% of the time. The reason why most designers post the reverse null is because they know it shows whether or not they've done their job correctly.
Note that odd order xovers don't show a reverse null when properly designed, but just like the standard 4th order LW does, if you're getting cancellation occurring with an odd order design, it too isn't quite where it should be.
Am still studying the details in your reply, but thanks! 🙂
Does an asymmetric-slope xo show an asymmetric reverse null?
Does an asymmetric-slope xo show an asymmetric reverse null?
Am I right in understanding that an asymmetric notch in the case of a symmetric xo is indicative of a primary lobe going off up or down?
What does a shallow symmetric notch tell us?
What does a shallow symmetric notch tell us?
You don't tend to end up with an asymmetric notch with perfectly symmetric slopes. What happens is that the notch depth, on axis, reduces as one goes further 'out of phase' and this tilts the primary lobe up or down.
Some bandpass enclosure alignments exhibit from 6 to 24 dB acoustical slopes with no series crossover components used at all.
An electrical crossover can be asymmetric, but result in a symmetrical acoustical crossover, depending on the driver and enclosure parameters. They would exhibit a deep symmetrical null when polarity is reversed on one driver.
Last edited:
Yes, when I was talking about slopes, i was referring to acoustic slopes, not electrical. I agree.
- Status
- Not open for further replies.