time allign or not?

dallaire

Member
2007-12-01 12:51 am
I'm sure that I have asked a variation of this question before but I was wondering if there is a benefit to allign driver voice coils vertically on the baffle if one is using a 24db per octave electronic constant voltage crossover? I think if a read things correct, that things would only be in phase at the cross frequency and not at any other frequencies, if this is correct, than is there anything to be gained by alligning drivers on baffle???
Thanks,
dallaire
 

dallaire

Member
2007-12-01 12:51 am
Iain, I have not read anything about this type of phase error. When you say stay away from crossing at the frequencies near the low end roll of, how close are you talking? using a 24db per octave slope? and then would this phase problem you speak of not effect most "two way" designs out there??? I mean they are all crossing over at pretty much the limits of the two drivers right? this is where a three way setup I think is so nice, as long as "I" am not using and designing a passive crossover, I could never get it right, from what I have read, it hard enough for the pro designers!
dallaire
 

Iain McNeill

diyAudio Moderator Emeritus
A 24dB/octave slope means that both drivers are contributing to the combined acoustic output over a 2 octave band. When a driver is 25dB down, it's contribution to the total output is less than 1dB.

Over this 2 octave band, each driver has to acoustically output the correct magnitude & phase response to properly combine. The acoustic output is the sum of the magnitude & phase responses of both the electrical filter and the driver in cabinet and if the driver-in-cabinet response isn't flat magnitude and 0° phase then the electrical filter isn't defining the acoustical output.

I think this is why a lot of 2- and 3- ways with off-the-shelf crossovers sound so-so.

From my simulations, the driver-in-cabinet phase response due to the high-pass roll off causes more significant problems. One way of mitigating this is to bias the crossover frequency up a bit but the best way is to design the filter to complement the driver/cabinet.

As you say, this can be hard when all you have are R, L & C's.

What are you using for your XO?
 
Iain McNeill said:
I think it is more important to compensate the crossover for each drivers magnitude and phase response.

I'll second on that.
Blind "time-alignment" disregarding the phase relationships between drivers is useless - that's why "generic" crossovers like active with "book" slopes just won't work.

People usually think of drivers as having linear phase, but, as already pointed out, roll-offs have associated phase shifts that need to be considered.
 

forr

Member
2004-12-01 6:46 pm
Next door
Hi Dallaire,

---benefit to allign driver voice coils vertically on the baffle if one is using a 24db per octave electronic constant voltage crossover?---

The theory behind crossovers supposes all drivers to be verticaly aligned. Your question should be : "What do I lose in my drivers are not strictly aligned ?"

In his book, Dickason shows response variations for offsets of 1" and 2" at 1000 Hz for an LR 24 : about 0.25 dB and 1.25 dB.

With digital crossover, it is easy to nullify the offset.


About the resonance of the high-passed driver, a long time ago, Marshall Leach wrote an AES article about its influence on the overall response. I have been intrigued that nobody seems to have done further work on the subject so I decided to do simulations myself. I found it was quite easy. I published some results in other threads but they did not retain much attention.

If you provide
- the frequency, kind and order of crossover (here LR24),
- the current vertical offset and the resonance frequency and Q of the high-passed driver,
I can do a simulation to determine what response variations can be expected.
 

dallaire

Member
2007-12-01 12:51 am
Well guy's, I am using the "Marchand XM1 crossovers" I have twelve position rotary switches for tweet and mid level control. power supply is also Phil's design. I have chose Peerless six inch woofer HDS Nomex, four inch HDS PPB mids/woofers, and Scan Speak 9500 tweeter. I was thinking of perhaps 350hz to mid and 2500-3000 to tweeter. Nothing is built yet, still in design phase, all this input is wonderful!!!
Thanks,
vince
 
My system is somewhat similar, with the Marchand crossovers, level trimmers on each output (I'm not sure 12pos rotary gives fine enough control), Morel MDT28 tweets, Dynaudio 5" midbass units (discontinued) on semi-open baffles, and some surplus AR 8" drivers in 1.5 cubic foot sealed enclosures. Very happy with it, though I could stand a serious upgrade of the bottom end quality.
 

dallaire

Member
2007-12-01 12:51 am
I don't know how the marchand x-overs compare to what else is out there, but I can say they are a lot better than the passives that came in my Definitive tech speakers. I'm waiting as I'm posting this for the UPS guy to deliver my new book "Joe D'appolito, Testing Loudspeakers" after reading it, I hope to have more knowledge on the subject, I'm told it is a good read!
dallaire
 

Iain McNeill

diyAudio Moderator Emeritus
The Marchand XO's use a state variable filter much like the Rane AC23 I've used for years (which sounds great!). The advantage here is that the high-pass and low-pass functions are locked in unison. You cannot make the filter pair that is not constant voltage. However, to successfully deploy this XO, the drivers must be flat magnitude and 0° phase for an octave above and below crossover. A tough challenge.

You're going to want to bias the crossover frequency higher than what seems appropriate. Simulation will help to evaluate this but phase deviations in the drivers are going to produce audible ripple in the combined response. With Conrads suggestion of connecting pots to adjust the XO frequency maybe you could tune this over time with listening although matching left & right will be a pain. Maybe you could get a used AC23 on EBAY for $100 like I did.

I made two identical crossovers with a relay switching system to allow me to A-B audition changes in real time. I can't tell you how educational this has been.

Don't get discouraged. you're approaching the fun part! You have excellent drivers here and there's no reason this speaker can't be world class.
 

dallaire

Member
2007-12-01 12:51 am
Thanks again Iain, I need to learn more about what you are talking about when you say the drivers must be flat magnitude, do you mean have flat "freq. response"? and zero degress phase for one octave above and below the cross point? I understand "at" the cross point, phase relationship, but need to read the book I just recieved yesterday "Testing Loudspeakers" Joe D'Appollito's book, perhaps it will give some insight to what you are speaking about in regards to the crossover/phase/octave thing? I am very patient and will educate myself before building amything, as I want it to be as good as I can make it. Is there any testing I can do with mic and software to check the correctness of what your refering to? or is this something that would require a scope? I have R plus D software from Acoustisoft and there rec. mic and pre anp/soundcard setup.
dallaire
 

Iain McNeill

diyAudio Moderator Emeritus
that's right, magnitude & phase are the two components of the frequency response.

Joe's book, as the title implies, focuses on speaker driver testing and doesn't really discuss crossovers in depth. But there's a lot of good stuff here and in the many DIY websites.

Not familiar with your RpluS software. Is it the same as the ETF system on their website? ETF has everything you need for characterizing your speakers & cabinets incl. mag + phase measurements of the speakers.

Many of the cabinet design programs will accept data from measurement systems like ETF so you can accurately predict what the cab will do before you cut wood.
 
This thread looks like a perfect place for me to expose my ignorance.

I am currently doing a 3-way OB active setup (24dB/oct.). I need a little help and it might pertain to this subject.

I'm having trouble with the phase(?) between the mid and woofer. When moving forward/backward from the baffle there seems to be phase lobes(?) from the woofer; these are present even when the mid is turned off. The problem is when I adjust the phase of the mid/woofer for one spot, when I move forward a bit it is out of phase. I can fix this by changing the phase again, but then the original listening spot is out of phase.

My question is, since there are these peaks/nulls with just the woofer playing, is this a room interaction issue or is it something that can be compensated for with time alignment?

Also, say for example the woofer voice coil is 2" further back then the mids voice coil, what would be the approximate delay the mid would need?

:confused:
 

Iain McNeill

diyAudio Moderator Emeritus
Sounds like room modes to me - the only way you can get interference from a single driver is from a reflection off a wall or large object. If the phase nulls are on axis then it's definitely a room mode because the relative distance from mid to woofer isn't changing.

speed of sound is 340m/s. it travels 1 ft in 1.1ms or so. 1" is about 100us delay. No much huh?

I think you got to fix a lot of other stuff before you'll hear that!

Look at the phase response of your 24dB/octave filters & consider the phase response of the drivers in the cabinet and make sure that the total phase response throughout the crossover region matches your design target.