This is a theoretical question about designing a minimum phase passive crossover. If you could put together a simple “recipe” for such a design, would it go something like this:
1.pick two drivers with wide overlap and good frequency response
2.Use first order network to connect the drivers to achieve 6db slope through the critical region (inductor on woofer, capacitor on tweeter)
3. Flatten impedance curves of both drivers with zobels. (Do the drivers need to have the same impedance once flattened?)
4. physically align driver acoustic centres relative to listening position
Is this the right direction to achieve minimal phase shift over a wide range? How closely does the driver rolloff have to follow a 6db slope to achieve this?
1.pick two drivers with wide overlap and good frequency response
2.Use first order network to connect the drivers to achieve 6db slope through the critical region (inductor on woofer, capacitor on tweeter)
3. Flatten impedance curves of both drivers with zobels. (Do the drivers need to have the same impedance once flattened?)
4. physically align driver acoustic centres relative to listening position
Is this the right direction to achieve minimal phase shift over a wide range? How closely does the driver rolloff have to follow a 6db slope to achieve this?
Hi,
Its a nebulous task with real drivers.
If you want something simple :
Use a bass driver with a full range mid/treble unit e.g. FR88EX.
( http://zaphaudio.com/smalltest/ )
Arrange the bass driver sensitivity to be higher than the FR by the
amount of BSC (baffle step compensation) you think you will need.
Design the box for the bass driver, ideally sealed with a Q of around 0.6.
Cross them over with a 1st order series network at the baffle step frequency.
(This should be quite a bit higher than the FR Fs ....)
Google and investigate all terms above not understood.
Sod Zobels and such guff, the above does not need them.
The above will get you as near to minimum phase issues as is sensibly possible.
rgds, sreten.
Its a nebulous task with real drivers.
If you want something simple :
Use a bass driver with a full range mid/treble unit e.g. FR88EX.
( http://zaphaudio.com/smalltest/ )
Arrange the bass driver sensitivity to be higher than the FR by the
amount of BSC (baffle step compensation) you think you will need.
Design the box for the bass driver, ideally sealed with a Q of around 0.6.
Cross them over with a 1st order series network at the baffle step frequency.
(This should be quite a bit higher than the FR Fs ....)
Google and investigate all terms above not understood.
Sod Zobels and such guff, the above does not need them.
The above will get you as near to minimum phase issues as is sensibly possible.
rgds, sreten.
Last edited:
Except for the possibility of different bass alignments i agree totally with Sreten.
Sreten's bass alignment but active instead of passive series (and a bit of fiddling to deal with baffle step), Tysen fits the recipe.
Except for the use of ML-TL loading, this EL166 MTM fits the receipe.
One of the reasons these will work, but a mid-woof/dome is VERY difficult, is because the acoustic centre of a driver moves with frequency & with dynamics. By XOing quite low you can get the drivers essentially co-incident.
dave
Lets not forget, that if you want a minimum phase system (not "minimum phase crossover") you will have to allign the acoustic centers of the drivers! This means recessing the tweeter voice coil to approx. the depth of the woofer voice coil.
David S.
David S.
Hi,
Zobels are "needed" for 1st order parallel crossovers, but for a
first order series c/o they are more of a hindrance than a help.
Simply don't do any useful for series so simply leave them out.
rgds, sreten.
Linear phase filters will not equate to a linear phase system. Driver phase will still be there. I'm also not sure if a linear phase highpass and linear phase lowpass would have the same group delay (and linear phase networks will have group delay). A first order total response (driver and network) can add to flat response and flat phase.
Still, we are fooling ourselves if we ignore the different group delays due to the relative driver depths. What was the objective here? Linear phase system or linear phase crossover?
David S.
Hi,
When c/o at the baffle step frequency voice coil position error is minimal.
The whole thing is an approximation, and near as you can sensibly get.
Whether the whole is laudible or not is a different matter, but its an approach.
Any frequency response errors will introduce phase errors by definition.
Getting BSC right also fixes its phase error as well as the response error.
rgds, sreten.
When c/o at the baffle step frequency voice coil position error is minimal.
The whole thing is an approximation, and near as you can sensibly get.
Whether the whole is laudible or not is a different matter, but its an approach.
Any frequency response errors will introduce phase errors by definition.
Getting BSC right also fixes its phase error as well as the response error.
rgds, sreten.
With active solutions though we can plug in the small delays after we do our best to get the smallest CTC possible.
Thanks, I have put that under "why didnt I think of that before". Pretty simple but very effective solution.
Here's a point.
How exactly does the phase of a driver vary with frequency?
Now, I don't claim to be an expert, but here's what I can figure...
At low frequencies, the driver's cone is moving as a piston. The acoustic centre will be somewhere within the cone.
Move higher up the frequency range, to where the cone is no longer acting as a piston, and what's the phase done?
If the driver is now radiating only from its dustcap (unlikely, I know, but it works for the point I'm making), the acoustic centre has moved with frequency.
As it has moved away from you, it's acted as a delay, altering the phase, according to the distance moved and the wavelength.
Kinda makes worrying about driver depths seem a bit simple, doesn't it? 😉
Chris
Within the drivers operating range should we even care what the phase is doing?
I see the phase change in measurements all the time, its just something I accept and move on. I only care about the phase at the XO point.
I see the phase change in measurements all the time, its just something I accept and move on. I only care about the phase at the XO point.
Not exactly.
Each driver can be thought of as a minimum phase bandpass. If the response is fairly smooth then the phase will retard (fall) on the top end and advance (rise) on the low end. If the bandpasses are narrow or the orders are high, then achieving linear phase will be next to impossible, the network phase shift will be swamped by driver bandpasses. You can try and equalize response upward to get to a first order total but it is hard work (Thiele does this though).
When I say minimum phase bandpass, that is relative to its center frequency delay, and that is dominated by its effective point of origin or acoustic center. The voice coil is approx. the acoustic center. If you want a linear phase system you must take care of the driver depth delay, it is significant. Don't worry about whether the sound is radiating from the dustcap or whatever, the unit will appear as a minimum phase bandpass of some type (this statement based on 30 years of measurements, I do claim to be an expert).
When discussing "active" networks for linear phase, I assume others mean DSP networks. Conventional active networks will not easily give linear phase response or adjustable delay. Linear phase usually means FIR type filters.
What was the original goal?
David
Hi, a simple recipe, which I think he has got, rgds, sreten.
Not simple at all :
http://www.meta-gizmo.net/tri/speak/STEEN.html
http://www.linkwitzlab.com/crossovers.htm
http://www.musicanddesign.com/Duelund_and_Beyond.html
Not simple at all :
http://www.meta-gizmo.net/tri/speak/STEEN.html
http://www.linkwitzlab.com/crossovers.htm
http://www.musicanddesign.com/Duelund_and_Beyond.html
Last edited:
Do it digitally with the Bodzio Ultimate Equalizer.
Duelund crossover are not minimum phase or transient accurate.
Duelund crossover are not minimum phase or transient accurate.
Last edited:
Amount of BSC demand is not fixed.
Probably an opamp style active BSC circuit could be more usefull i guess.
Might relay on the natural rolloff of a fullrange, and complement it with a bass (actualy if its a good fullrange than subbas would be a better name) unit.
The suggestions could be wrong naturaly.
Lot depends on personal preference.
Probably an opamp style active BSC circuit could be more usefull i guess.
Might relay on the natural rolloff of a fullrange, and complement it with a bass (actualy if its a good fullrange than subbas would be a better name) unit.
The suggestions could be wrong naturaly.
Lot depends on personal preference.
With the scheme that sreten suggests, the XO is low enuff that the drivers can be placed such that they are all within a 1/4 wavelength making them essentially co-incident.
A wide choice of drivers exist that allow a broad overlap making 1st order XO feasible. Add in the subtractive nature of a series XO and one would have to work hard against the goal to not have it all just fall into place.
dave
A wide choice of drivers exist that allow a broad overlap making 1st order XO feasible. Add in the subtractive nature of a series XO and one would have to work hard against the goal to not have it all just fall into place.
dave
- Status
- Not open for further replies.