I have been studying furiously here and with books and tutorials and am now working on my first crossover design using various online calculators here and elsewhere for the component starting point for my 2nd order 2 way crossover elements. I am creating it in XSIM to watch the SPL and impedance plot effects of my layout and various additional circuit blocks.
I need some help to understand what I am seeing and to understand the best sequence to design my circuit. I started with picking a crossover point based on the driver characteristics and computing the 4 cap/ind values for my 2nd order 2 way circuit. In XSIM I massaged the crossover point to get a smooth transition and the applied an L pad resistor network on the tweeter to match the sensitivities and get and pretty good overall flat system response.
My questions:
1) The apparent graphical crossover point in XSIM using the FRD and ZMA files for my two drivers is different by around 2K hz from that generated by the standalone DIY Audio calculator. Why is this? Is it because of the non ideal impedance curves for the actual drivers where the impedance is not say 8R at the crossover point?
2) I see that a properly designed tweeter L pad network reduces the high frequency level as expected and does not affect the crossover behavior as I expected. Do I then assume that the add of a proper L pad should not disturb the selection of the crossover components?
3) As I try some of the other refinements to the circuit such as flattening the impedance curve of the woofer with a Zobel network does seem to affect the crossover component selection. Does that mean that all of the other steps such as Zobel flattening, notch filters, baffle step correction, etc (except the tweeter L Pad) cause an iteration in the starting selection for the basic crossover component selection?
Thanks,
George
I need some help to understand what I am seeing and to understand the best sequence to design my circuit. I started with picking a crossover point based on the driver characteristics and computing the 4 cap/ind values for my 2nd order 2 way circuit. In XSIM I massaged the crossover point to get a smooth transition and the applied an L pad resistor network on the tweeter to match the sensitivities and get and pretty good overall flat system response.
My questions:
1) The apparent graphical crossover point in XSIM using the FRD and ZMA files for my two drivers is different by around 2K hz from that generated by the standalone DIY Audio calculator. Why is this? Is it because of the non ideal impedance curves for the actual drivers where the impedance is not say 8R at the crossover point?
2) I see that a properly designed tweeter L pad network reduces the high frequency level as expected and does not affect the crossover behavior as I expected. Do I then assume that the add of a proper L pad should not disturb the selection of the crossover components?
3) As I try some of the other refinements to the circuit such as flattening the impedance curve of the woofer with a Zobel network does seem to affect the crossover component selection. Does that mean that all of the other steps such as Zobel flattening, notch filters, baffle step correction, etc (except the tweeter L Pad) cause an iteration in the starting selection for the basic crossover component selection?
Thanks,
George
Hey George,
First, where did you get the FRD and ZMA?
1 - They should be measured, mounted in the speaker.
2 - Kind of sort of. Sometimes using a raw resistor IS The right answer. 🙂 So in general, you want to shoot for a final driver impedance of around 8 Ohms. If your tweeter is 4, for instance, and you need to pad it down, then you might want to design an L Pad to give you a final of 6-8 Ohms, to make it easier to drive.
3 - Zobels in the woofer are incredibly helpful in the crossover design, so they probably should come first, same if you have a very high inductance tweeter like you might with a ribbon.
However, if you are designing for tubes and you want to flatten the crossover point, yes, you can do that after everything else is done.
First, where did you get the FRD and ZMA?
1 - They should be measured, mounted in the speaker.
2 - Kind of sort of. Sometimes using a raw resistor IS The right answer. 🙂 So in general, you want to shoot for a final driver impedance of around 8 Ohms. If your tweeter is 4, for instance, and you need to pad it down, then you might want to design an L Pad to give you a final of 6-8 Ohms, to make it easier to drive.
3 - Zobels in the woofer are incredibly helpful in the crossover design, so they probably should come first, same if you have a very high inductance tweeter like you might with a ribbon.
However, if you are designing for tubes and you want to flatten the crossover point, yes, you can do that after everything else is done.
Forgot,
The crossover point for two drivers is more than where they overlap. For a 2-way speaker, the recommended lower bound of the tweeter and the dispersion of the mid-woofer are important considerations. For this reason, picking two drivers with significant overlap is helpful. If you pick two drivers that force you to cross at a certain point you can end up with unwanted compromises.
The higher you crossover, the less power the tweeter will be subject to, and higher the dynamic range and power handling. However, too high a point and the mid-woofer's dispersion will be too narrow, and lead to poor off axis response.
Last, you'll need to measure the acoustic offset of your drivers before designing your crossover.
The crossover point for two drivers is more than where they overlap. For a 2-way speaker, the recommended lower bound of the tweeter and the dispersion of the mid-woofer are important considerations. For this reason, picking two drivers with significant overlap is helpful. If you pick two drivers that force you to cross at a certain point you can end up with unwanted compromises.
The higher you crossover, the less power the tweeter will be subject to, and higher the dynamic range and power handling. However, too high a point and the mid-woofer's dispersion will be too narrow, and lead to poor off axis response.
Last, you'll need to measure the acoustic offset of your drivers before designing your crossover.
What is the correct sequence to apply the various circuit blocks in your crossover (basic network, Zobel, baffle step compensation, notch filters, L pad, etc)?
Thanks,
George
Thanks,
George
I use the mnfg curves if available, Spl trace if not - for speaker selection and purchase. Then I use DATS, EMM6,sound easy for TS parameters and for FRD and ZMA curves in the boxes.
George
This is a rough guide, as the truth is, we often iterate, going back and forth. For instance, Zobels are first, but often tweaking them can help improve driver phase matching later.
1 - Driver Zobels if and only if needed.
2 - Level setting devices
3 - Crossover filters
4 - Notch filters
5 - Impedance flatterners (which are not universally regarded as required)
1 - Driver Zobels if and only if needed.
2 - Level setting devices
3 - Crossover filters
4 - Notch filters
5 - Impedance flatterners (which are not universally regarded as required)
All your questions are best understood while working on a specific set of units within a simulation program. Follow the rules of preparing frd's and zma's so they reflect the driver properties and the enclosure they are meant to be in.
Clearly measuring FRD and ZMA in the treated boxes with the actual drivers to use for crossover design will get you a much better finished result that will behave more like your modeled system curves but what are your thoughts on measuring the speaker in quasi anechoic conditions (outdoors say) vs the actual listening position? A commercial speaker builder has to generalize but if I know where they are going I do not I suppose.
George
George
The first step is using FRD/ZMA data that reliably simulates the drivers you will be using in your intended baffle dimensions and driver placement on that baffle.
Baffle width/ height affects the frequency that baffle step applies (change from low frequencies wider than the baffle radiating 360 degrees (wrapping around the baffle = 4pi space) to projecting forward off the baffle 180 degrees (2pi space radiation).
The placement / position of the driver on the baffle affects "diffraction" which is a ripple overlay on the "infinite baffle" native response of the driver. The driver's infinite baffle response has peaks and dips and these combine with baffle step PLUS diffraction peaks and dips.
You must also have phase data at least in the FRD response so that your crossover will "sum" correctly.
Lastly - you need to know your Z offset on the baffle if you are not time aligning the voicecoils of your drivers
Otherwise - (frd/zma) garbage in / out applies and your crossover design will be flawed.
The good thing about the XO tools is they let you see very quickly your XO design.
The downside - people "rush to XO" and don't understand the significance of the FRD/ZMA data they are loading that reality will be quite different from simulation if they don't prepare their FRD/ZMA as above.
There are probably step by step tutorials to do the above. I was going to write one if not one day.
Baffle width/ height affects the frequency that baffle step applies (change from low frequencies wider than the baffle radiating 360 degrees (wrapping around the baffle = 4pi space) to projecting forward off the baffle 180 degrees (2pi space radiation).
The placement / position of the driver on the baffle affects "diffraction" which is a ripple overlay on the "infinite baffle" native response of the driver. The driver's infinite baffle response has peaks and dips and these combine with baffle step PLUS diffraction peaks and dips.
You must also have phase data at least in the FRD response so that your crossover will "sum" correctly.
Lastly - you need to know your Z offset on the baffle if you are not time aligning the voicecoils of your drivers
Otherwise - (frd/zma) garbage in / out applies and your crossover design will be flawed.
The good thing about the XO tools is they let you see very quickly your XO design.
The downside - people "rush to XO" and don't understand the significance of the FRD/ZMA data they are loading that reality will be quite different from simulation if they don't prepare their FRD/ZMA as above.
There are probably step by step tutorials to do the above. I was going to write one if not one day.
It is a little complicated. If you are doing a 2-way, then you can 100% do this in place, with gated measurements, and in fact, you probably should!
However, for 3-ways, you need the quasi-anechoic measurements to get accurate phase and amplitude through the crossover point.
The problem with quasi anechoic however, is you still have to add the baffle step.
So, my advice, if you are making a 2-way, is measure in place, and watch for baffle step issues.
Best,
E
- Status
- Not open for further replies.