Happy New-Year,
In all the on line calculators a passive24 dB high pass filter can be seen as two 12 dB serie filter and is always made the same way: first serie cap is always the littliest capacitance value then the second "hub" (cellule) has a bigger capacitance value.
Is there a reason why a designer may want to invert the two 12 dB "hubs" of a 24 dB filter by putting first the bigger capacitance value with its dedicated shunt voice-coil then the second "hub" that is usualy the first ?
A sim on a crossover filter soft is saying that is the same but is it ? Or there is reason to invert sometimes ? Parts precision after sorting out vis a vis of the low pass filter for instance, anything else ?
In all the on line calculators a passive24 dB high pass filter can be seen as two 12 dB serie filter and is always made the same way: first serie cap is always the littliest capacitance value then the second "hub" (cellule) has a bigger capacitance value.
Is there a reason why a designer may want to invert the two 12 dB "hubs" of a 24 dB filter by putting first the bigger capacitance value with its dedicated shunt voice-coil then the second "hub" that is usualy the first ?
A sim on a crossover filter soft is saying that is the same but is it ? Or there is reason to invert sometimes ? Parts precision after sorting out vis a vis of the low pass filter for instance, anything else ?
If the Ls are different and the Cs are different, then there will be a significant difference in the result
from reversing the order of the filters. This is due to the loading effect of the second filter on the first,
and also of the driver's loading on the second filter. If you work out the equations, they will be different
in the two cases.
from reversing the order of the filters. This is due to the loading effect of the second filter on the first,
and also of the driver's loading on the second filter. If you work out the equations, they will be different
in the two cases.
Thanks Rayma,
Are these differences comming from the caps datas like diferent voltages and capacitances and the esr amount due to different number of // caps in each of the two hubs ?
Is the first shunt coil resistance is for instance chosen for the Q damping of the high pass bump and the serie resistance of one hub in spite of the other in relation of the better spl atenuation vis a vis of the high pass of the driver below (are the ears can hear few m ohms? ) ?
Or do you talk of something else?
Is it often seen in loudspeaker brands?
Are these differences comming from the caps datas like diferent voltages and capacitances and the esr amount due to different number of // caps in each of the two hubs ?
Is the first shunt coil resistance is for instance chosen for the Q damping of the high pass bump and the serie resistance of one hub in spite of the other in relation of the better spl atenuation vis a vis of the high pass of the driver below (are the ears can hear few m ohms? ) ?
Or do you talk of something else?
Is it often seen in loudspeaker brands?
Even for perfect L and C there are math differences in the filters from the different sequences.
A cascaded passive filter like this is actually rather poor and cannot create a proper filter shape.
But it is cheap and simple. Active line level filters are much more accurate. Of course the speaker
designer may not want a perfect second order filter shape either. I think these days third order
HP filters are more often used than second.
A cascaded passive filter like this is actually rather poor and cannot create a proper filter shape.
But it is cheap and simple. Active line level filters are much more accurate. Of course the speaker
designer may not want a perfect second order filter shape either. I think these days third order
HP filters are more often used than second.
The key thing here is that passive filters are based on interactions of reactance with the impedance of the rest of the components in the circuit.
Each component in a passive filter affects not only the frequency response, but also the impedance that the preceding components must work with.
If you put the smaller cap nearer the driver, there will not be any value of larger cap that you can put before it and still realise your target curve, because the impedance it has to work with will be too far out of spec.
As an aside, I'd encourage anyone to avoid using the simple online crossover calculators and get into proper modelling software like VituixCAD. This allows you to see for yourself what changes to the value of each component do to both the impedance and frequency response of a circuit.
It also allows you to take account of the many things that the simple calculators completely ignore. For example, the impedance of a driver is often not a constant 8 (or 4, etc) Ω at all frequencies. This means you probably don't want a textbook alignment anyway, as that won't end up flat when that non constant impedance is accounted for. You can also account for things like bafflestep, and look at off axis responses as well as on-axis too.
Each component in a passive filter affects not only the frequency response, but also the impedance that the preceding components must work with.
If you put the smaller cap nearer the driver, there will not be any value of larger cap that you can put before it and still realise your target curve, because the impedance it has to work with will be too far out of spec.
As an aside, I'd encourage anyone to avoid using the simple online crossover calculators and get into proper modelling software like VituixCAD. This allows you to see for yourself what changes to the value of each component do to both the impedance and frequency response of a circuit.
It also allows you to take account of the many things that the simple calculators completely ignore. For example, the impedance of a driver is often not a constant 8 (or 4, etc) Ω at all frequencies. This means you probably don't want a textbook alignment anyway, as that won't end up flat when that non constant impedance is accounted for. You can also account for things like bafflestep, and look at off axis responses as well as on-axis too.
... worked a little on that since, having several loudspeakers on the bench doesn't help to go fast but listenning music instead !
I certainly must have to measure the coils with Arta Limp around the cut off of circa 2500 hz which is a 7.4 ohms there, but anyway it is not usual to see a 24 dB high pass with at beginning 6 ohms tamming resistors // to a 1.5 uF lift, then a 14 uf first cell before the first shunt coil then a much lower second cell at circa 7 uF near the tweeter...
Dunno if someone ever seen such odd topology. I must measure this tweeter on axis then put it in Xsim with the rigth coils values. Or at least Vituix without the +/- 180° 10% implement as I can not do that in my appartment.
I certainly must have to measure the coils with Arta Limp around the cut off of circa 2500 hz which is a 7.4 ohms there, but anyway it is not usual to see a 24 dB high pass with at beginning 6 ohms tamming resistors // to a 1.5 uF lift, then a 14 uf first cell before the first shunt coil then a much lower second cell at circa 7 uF near the tweeter...
Dunno if someone ever seen such odd topology. I must measure this tweeter on axis then put it in Xsim with the rigth coils values. Or at least Vituix without the +/- 180° 10% implement as I can not do that in my appartment.