I assumed it was standard operating procedure to reverse the polarity on the tweeter in a 2 way second order crossover? But doing a google search on hookup for second orders resulted in diagrams with both drivers wired both ways. (+ to+, + to -). What is the determining factor on which is works best?
__________________
__________________
No matter, one of the mods will move it.
There isn't a short version or simple answer to this unfortunately, but FWIW:
It depends. Electrical transfer functions and the actual acoustical slopes are not always (usually are not) synonymous, since the driver's natural responses usually come into play unless you're crossing drivers with a flat response out to ~2 octaves past the nominal XO frequency. IOW, if the driver is already rolling off 2nd order by itself & you stick a 2nd order electrical on it in that general vicinity, you've got a 4th order acoustic slope, of xyz type (LR, Butterworth, Gaussian or whatever).
The assumption of flipping the polarity for 2nd order filters is based on symmetric filters (2nd order has 180 degrees phase rotation) and coincident acoustic centres. Usually not the case on a flat baffle, since the midbass units will usually have their VCs a good 20 - 25mm (or more) behind that of a normal dome tweeter. Less of an issue with cone tweeters of course, although even then there is likely to be some offset in the z (front - rear) axis.
If you've a flat baffle, you'll usually need some form of ladder delay network on the tweeter when using 2nd order acoustic slopes. You can physically move the tweeter back with a stepped baffle, although the diffraction can be a problem, and that only works optimally when directly on axis -as you move off, it can be more of an issue. Neither of these are necessarily inevitable -you can sometimes get away without them (although not often); that is circumstance dependent, e.g. the precise slope Q and driver positions etc.
Which is a bit of a long winded way of saying there isn't a single answer as there are too many variables involved.
There isn't a short version or simple answer to this unfortunately, but FWIW:
It depends. Electrical transfer functions and the actual acoustical slopes are not always (usually are not) synonymous, since the driver's natural responses usually come into play unless you're crossing drivers with a flat response out to ~2 octaves past the nominal XO frequency. IOW, if the driver is already rolling off 2nd order by itself & you stick a 2nd order electrical on it in that general vicinity, you've got a 4th order acoustic slope, of xyz type (LR, Butterworth, Gaussian or whatever).
The assumption of flipping the polarity for 2nd order filters is based on symmetric filters (2nd order has 180 degrees phase rotation) and coincident acoustic centres. Usually not the case on a flat baffle, since the midbass units will usually have their VCs a good 20 - 25mm (or more) behind that of a normal dome tweeter. Less of an issue with cone tweeters of course, although even then there is likely to be some offset in the z (front - rear) axis.
If you've a flat baffle, you'll usually need some form of ladder delay network on the tweeter when using 2nd order acoustic slopes. You can physically move the tweeter back with a stepped baffle, although the diffraction can be a problem, and that only works optimally when directly on axis -as you move off, it can be more of an issue. Neither of these are necessarily inevitable -you can sometimes get away without them (although not often); that is circumstance dependent, e.g. the precise slope Q and driver positions etc.
Which is a bit of a long winded way of saying there isn't a single answer as there are too many variables involved.
There is a right way to do it. The issue is there is more to it than the text book behaviour of filters calculated with standard values. They will interact with the speaker impedance and response to produce something other than intended unless designed with these things taken into account. Further, changing polarity is a matter of a 180 degree step, whereas phase rotates continuously around 360 and can be different at different frequencies.
a microphone and a response curve will determine easy the correct way to wire your cross over.
A second order cross over is things of the past for tweeters btw, now the standard is 3 way cross overs for tweeters.
You can get good results if your tweeter is more than 30 mm wide and can cope with some midrange to pass through it.
A second order cross over is things of the past for tweeters btw, now the standard is 3 way cross overs for tweeters.
You can get good results if your tweeter is more than 30 mm wide and can cope with some midrange to pass through it.
Take a look at this discussion and scroll down to Driver Phase Matching:
https://speakermakersjourney.blogspot.com/2016/05/lm-1-bookshelf-crossover.html
The LM-1 crossover simulation files are available over in the software section here, so you can play with alternatives.
Best,
Erik
https://speakermakersjourney.blogspot.com/2016/05/lm-1-bookshelf-crossover.html
The LM-1 crossover simulation files are available over in the software section here, so you can play with alternatives.
Best,
Erik
Any passive crossover is pretty likely to be way off if you just do the math and build it (even if you've used published impedance graphs of the drivers) . The final result should be verified with pink noise or a slow swept sinewave, at which point you can switch the phase of the tweeter and see which way gives you the less damaged frequency response.
Eh? According to whom? I assume you're talking about 3rd order electrical (T section) filters? That's news to me. The electrical filter topology on the tweeter depends (or should depend) on what you're doing, i.e. XO frequency, power-handling of the drivers involved, impedances, phase-matching and a host of other factors. 2nd order electrical is as useful now as it ever was, and still regularly employed in many, many designs, just as 3rd order is.
it is still employed but 3rd order is the future and the new standard in monitors and recording studio.
B&W might be the exception, but harbeth, rogers, usher, atc and spendor moved to 3rd order
Um, just to talk "common" vs. "necessary":
Accurate phase matching between drivers is necessary, as is a smooth frequency response. How that is accomplished given specific drivers, baffles, and other requirements may vary.
Most of the 2 way speaker designs I have seen need to have the tweeters inverted, and they do! However it's an artifact of the crossover points selected, driver's natural phase response and acoustic delay. Inverting the tweeter is not the requirement. I got lucky with the LM-1 for instance, and the drivers line up best in positive polarity.
The idea of a particular order filter (2nd, 3rd, 4th) etc. being required by certain designs is not something I have encountered, although I believe D'Appolito has stated a preference higher order designs in the MTM designs he's known for. Another point to go against the idea of a specific network order being needed is that many 2-way systems match best with asymmetrical filters, such as a 2nd order on the woofer, 3rd on tweeter. Again, while this is common, it's not the goal nor does it make it better or worse, IMHO than a 2-way with symmetrical filters.
As has been mentioned, power handling, and off-axis responses also matter. It's up to the designer to figure out what he/she cares about most.
Accurate phase matching between drivers is necessary, as is a smooth frequency response. How that is accomplished given specific drivers, baffles, and other requirements may vary.
Most of the 2 way speaker designs I have seen need to have the tweeters inverted, and they do! However it's an artifact of the crossover points selected, driver's natural phase response and acoustic delay. Inverting the tweeter is not the requirement. I got lucky with the LM-1 for instance, and the drivers line up best in positive polarity.
The idea of a particular order filter (2nd, 3rd, 4th) etc. being required by certain designs is not something I have encountered, although I believe D'Appolito has stated a preference higher order designs in the MTM designs he's known for. Another point to go against the idea of a specific network order being needed is that many 2-way systems match best with asymmetrical filters, such as a 2nd order on the woofer, 3rd on tweeter. Again, while this is common, it's not the goal nor does it make it better or worse, IMHO than a 2-way with symmetrical filters.
As has been mentioned, power handling, and off-axis responses also matter. It's up to the designer to figure out what he/she cares about most.
Last edited:
That's a misleading way of stating your "fact".
Yes, the total phase change from infinitely low frequency to infinitely high frequency is 180degrees for a 2pole filter. (it's 90 degrees for each pole of the filter).
But at the cross over pfrequency the phase has only changed by45degrees for each pole giving 90 degrees for a 2pole.
The 2pole high pass gives a change of 90degrees in one direction.
The 2pole low pass gives a change of 90degrees in the other direction.
At the crossover the two effects add up to a difference between the two signals of 90°+90° = 180°, i.e. the signals are out of phase at the crossover frequency if they are measured/listened to at the same distance.
LR2 requires a cascade of two Butterworth single pole filters. These need to be separated so that each filter sees the required source and load impedances. Or each filter needs the component values altered to correct fo non ideal impedances between the filters.
You cannot make an LR2 from a 2pole filter. i.e. an LR2 = two B1 in cascade.
If you want to build an LR4, then you need two B2 in cascade.
Last edited:
Yes Erik, it is not necessary, it is the future of good speakers which are built with unrestricted bandwidth drivers.
Epos and many others have specific drivers which will work well with 1st order or second order. However I was only making a point that modern unrestricted drivers will benefit from a steeper xo point for clarity.
Badge it is nice to see you will measure your speakers! I tried to design my xo and 4 years later I stand exactly where I started... it is very hard to do.
Epos and many others have specific drivers which will work well with 1st order or second order. However I was only making a point that modern unrestricted drivers will benefit from a steeper xo point for clarity.
Badge it is nice to see you will measure your speakers! I tried to design my xo and 4 years later I stand exactly where I started... it is very hard to do.
Last edited:
Yes, but I'm also somewhat prone to consider context i.e. whom I'm talking to. If I give the simple explanation (fudged for that purpose), it gets the point across, which was the object in this case. If I give a longer, more complex one, the only people likely to understand it are those who have a reasonable understanding of filter design in the first place, rather negating the value. In this case, I felt it to be sufficient.
Last edited:
So what? T section filters are not the 'future', nor are they the present, nor the past. They are exactly what they have always been: a T section filter, with the same advantages & disadvantages they have always had. Personally I favour using a filter that does the job required of it -other than abstract interest, I couldn't give a hoot if the electrical topology is 2nd, 3rd, 4th, or anything else so long as it achieves the design objectives set out.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.