As discussed here I have implemented a good approximation to a 4th order LR to a Seas tweeter by estimating the existing Q and frequency, and using an LT to raise the cut frequency and make the response align to the 2nd order Butterworth - and then overlaying a further second order.
This gives a pleasing result with my miniDSP crossover.
I'd like to use this technique again - but with a conventional (if external) crossover. Is there any software that would do this, or should I just try to replicate the response by simply fiddling in XSim with trial and error? I don't need to introduce any gain, its all attenuation and raising the cut frequency.
(And yes that leaves me with the Z offset to mess up phase, which I deal with in pure delay in DSP - but it will be a starting point, and for some tweeters with very low Fs, I might still look for around 1.2kHz 4th order - like my Wavecors for example - which should reduce the phase error somewhat).
This gives a pleasing result with my miniDSP crossover.
I'd like to use this technique again - but with a conventional (if external) crossover. Is there any software that would do this, or should I just try to replicate the response by simply fiddling in XSim with trial and error? I don't need to introduce any gain, its all attenuation and raising the cut frequency.
(And yes that leaves me with the Z offset to mess up phase, which I deal with in pure delay in DSP - but it will be a starting point, and for some tweeters with very low Fs, I might still look for around 1.2kHz 4th order - like my Wavecors for example - which should reduce the phase error somewhat).
Passive or active analogue crossover?
In the active case, you can do the same as you did digitally, you just don't get the inaccuracies due to the finite sample rate anymore because there is no sampling. Linkwitz's website shows how to calculate Linkwitz transform circuit values and you can find information on second-order Butterworth high-pass filter sections all over the place.
What you have now is a digital approximation to a Linkwitz transform circuit, when you build the circuit on the late Siegfried Linkwitz's site, you have the real thing.
In the active case, you can do the same as you did digitally, you just don't get the inaccuracies due to the finite sample rate anymore because there is no sampling. Linkwitz's website shows how to calculate Linkwitz transform circuit values and you can find information on second-order Butterworth high-pass filter sections all over the place.
What you have now is a digital approximation to a Linkwitz transform circuit, when you build the circuit on the late Siegfried Linkwitz's site, you have the real thing.
Last edited:
It's the same result in the end. Your precision is going to come from your generation of the target curve.
If you simply use the rolloff you want the woofer to be producing, your response measured for use in Xsim should be clean in the low end.. otherwise you could generate the difference curve and apply only that.
This may not be intuitive but as much as the changed phase due to delaying a tweeter appears on the tweeter, it only affects the other driver.
Analogue. I wish to replace the active system with an (outboard) passive crossover.
Can you present the transfer function (TF) of your MiniDSP settings? In case we know that TF, it could be replicated -to some extent- passively. The impedance can be taken from the Seas datasheet I presume. Otherwise, without any data and measurements, we are basically shooting in the dark.
VituixCAD
Has filter blocks that will calculate proper inductance/capacitance
According to impedance and desired cutoff.
Cut off and impedance values can be scrolled in real time.
no endless fiddling or trial and error.
After calculation, values and DCR can be set to real world coil values/tolerance
Since real driver impedance is used or ZMA file
the impedance can be fine tuned, no roller coaster 25 ohm + nonsense peaks
Xsim is useful but extremely limited and time consuming
Is it not possible to low the spl end of the woofer instead by choosing a coil for the passive low pass with more Z, like an air coil, then avoiding the LT on the tweeter ?
You do know that analogue crossovers can be active, don't you 😉? In fact, in the good old days when DSPs were still expensive, most active crossovers were analogue.
I never heard of a passive Linkwitz transform circuit, but I'll see if I can come up with an LRC circuit that has the required behaviour. I don't see any reason why it couldn't be done, especially not when bass has to be reduced.
Something like this could maybe work as a passive bass-reducing Linkwitz transform circuit, at least it has a pair of poles and a pair of zeros. I have no idea how to calculate the component values and combining it with a passive LC high-pass will complicate it further.
Thank you all. Yes I know that active can be analogue - I want to get the speakers into a state where they need a conventional 2 channel amplification and I can give them to one of my kids.
I currently don't have a target function I can show for the high pass on the tweeter (which has the LT to raise its 2nd order and then a further 2nd order) because the minidsp console only shows the crossover and peq parts separately. It would be nice if there was a utility to take the settings from the export XML and graph them. This is the transform on the tweeter. Fs is 1200 and the impedance peaks across that area from 6Ohms to 7.5Ohms according to the H519 datasheet, though that's less relevant if there are other circuits between the tweeter and this part.
I guess what I need is something that will be a good starting point for a low shelf, and I can play with it in xsim or vituix until its reasonable?
The low shelf is an attenuation at low frequency, passthrough at high, the 2nd order high pass is standard, there will be an lpad for level matching and I guess an impedance compensation. Could the low shelf be as simple as an LPAD with a capacitor shunt - surely that is first order at best?
I currently don't have a target function I can show for the high pass on the tweeter (which has the LT to raise its 2nd order and then a further 2nd order) because the minidsp console only shows the crossover and peq parts separately. It would be nice if there was a utility to take the settings from the export XML and graph them. This is the transform on the tweeter. Fs is 1200 and the impedance peaks across that area from 6Ohms to 7.5Ohms according to the H519 datasheet, though that's less relevant if there are other circuits between the tweeter and this part.
The low shelf is an attenuation at low frequency, passthrough at high, the 2nd order high pass is standard, there will be an lpad for level matching and I guess an impedance compensation. Could the low shelf be as simple as an LPAD with a capacitor shunt - surely that is first order at best?
A few years ago I wrote a MATLAB script to compute component values in a passive line-level Linkwitz Transform. I just looked at that script, and found the following in the comments:
% Since inductors are the most difficult components to tune, we will select the value
% for one of the inductors and then derive the other component values from it.
% (Could choose the value of R2 instead, to suit minimum input impedance requirements,
% but then the inductances get HUGE.)
Just a heads-up that a passive LT is possible, but might not be realizable.
% Since inductors are the most difficult components to tune, we will select the value
% for one of the inductors and then derive the other component values from it.
% (Could choose the value of R2 instead, to suit minimum input impedance requirements,
% but then the inductances get HUGE.)
Just a heads-up that a passive LT is possible, but might not be realizable.