Strange problem - Phase Cancellation?

[stevem]

I have been switching back and forth between a 2-way and a 3-way speaker system configuration. All the drivers each have their own power amp channel, and are actively crossed over.

I noticed that when using the 3-way configuration, my 1/3 octave spectrum analyzer indicates a drop in response of about 9 db right at the upper crossover frequency (1.23khz). If I invert the phase of the mid driver, the notch disappears, but another notch appears around the lower crossover frequency (297Hz). I do not see this phenomena when using the system as a 2-way with the crossover at 1khz.

The tweeter and the woofers are positive to the black terminals, while the midrange is positive to the red terminals. I have compensated for this in the wiring to the divers from the power amps.

Does anyone have an idea what may be causing this and how to correct it?

[Zilch]

It's your tweeter that's out of phase, not the mid.

When you invert the mid, you put it in phase with the tweeter, but out of phase with the woofer.

Invert the tweeter, instead.

Does that work?

The phase cancellations are not only electrical, but also acoustical, from the physical alignment of the drivers....

[stevem]

I thought of that, but if the tweeter is out of phase, then wouldn't it also be out of phase with the woofer when used as a 2-way? I don't notice any cancellations that way at all.

I use a digital crossover that has delay functions, so the drivers should be acoustically aligned. That shouldn't be the problem either.

[scott fitlin]

What slopes are you using?

[stevem]

What slopes are you using?

48 db/octave Linkwitz-Reilly.

[boputnam]

Get or borrow a Galaxy "Cricket" and individually test each transducer for what it's "seeing", phase wise.

http://www.galaxyaudio.com/galaxy/Products4.html

Sweetwater's got 'em - http://www.fullcompass.com/Products/pages/SKU--11726/index.html.
Call 800 356-5844 x1114 "George". He is my rep, and does me right! :thmbsup:

I think I like Zilch's guesstimate, but personally I stopped such "guesstimates" a long time ago, and now take measurements, instead. Measurements are unambiguous and focus your trouble-shooting.

The Cricket takes no prisoners...

[boputnam]

I thought of that, but if the tweeter is out of phase (with the LF), then wouldn't it also be out of phase with the woofer when used as a 2-way?Not necessarily. But, we need more info on how you run this as a "two-way" (only LF and HF, only LF and MF - what!!??). Until you provide that, I'll move out on this limb, here...

Just "guessing" :o: , but if the MF is out-of-phase with the HF (thus presenting the cancellation), maybe the LF is in-phase with the HF, and therefore when in two-way mode (LF and HF only) creates a summing in the area of frequency overlap between the LF and HF.

Your observation:
I don't notice any cancellations that way at all.... suggests the LF and HF are in-phase.

It is common in a 3-way (BUT NOT THE RULE) that the MF is alone out-of-phase.

Do you know?

[stevem]

Not necessarily. But, we need more info on how you run this as a "two-way" (only LF and HF, only LF and MF - what!!??). Until you provide that, I'll move out on this limb, here...

The 2-way is the LF and HF. The 3-way adds the mid-range. Thanks for the heads-up on the Cricket. It's inexpensive enough, and at least I'll know for sure how everything is phased. Something is not quite right with my system, and it does sound like a phase anomaly. When I invert the phase on the mids, it sounds a little different but not much. Can driver proximity effect phase? My drivers are almost touching each other (from the bottom, WWMT).

[baldrick]

Am I way of when I say that the phase changes 180' every 6db on a filter? It means, i.e. if you have a 2 way 12 db/ocate filter the phase is correct, if 18 db/ocate than you should change the phase on either tweeter or woofer, if 24 db/ocate than OK and so on... could this be the reason why you have to revert the phase on your 3 way system?

I've dealt with very much caraudio earlier (mainly JBL of course) and thats when I found out that when i changed the slope from i.e. 18 to 24db/ocate I also had to change the phase. The same experinse I've had with surround amplifiers with sub-preout and the small/large speakers button. When the amp have a 6/18 db filter you need to change the phase on either the sub or your woofer when running speakers in "large", otherwise the sub and fronts will phase out :)

[stevem]

I thought that the 48 db/octave Linkwitz-Reilly slopes were supposed to sum flat and in-phase. Is this correct?

[boputnam]

:dont-know

You've wired the MF differently than the rest, so I'm getting confused on what the -48dB slopes are doing across the spectrum.

If it were me - and while awaiting the Cricket - I'd experiment by cross-wiring the HF and measure the response.
Maybe yes, maybe no...

[stevem]

:dont-know

You've wired the MF differently than the rest, so I'm getting confused on what the -48dB slopes are doing across the spectrum.

If it were me - and while awaiting the Cricket - I'd experiment by cross-wiring the HF and measure the response.
Maybe yes, maybe no...

I'll try reversing the phase on the tweeters. Should I check the diaphragms to see if they're wired correctly? I ordered the Cricket this afternoon. Full Compass had the best price too.

[boputnam]

1. I'll try reversing the phase on the tweeters.
2. Should I check the diaphragms to see if they're wired correctly?
3. I ordered the Cricket this afternoon. Full Compass had the best price too.

1. Let us know... :hmm:
2. Do you think they have been refurbished? Or are you just reasonably thorough to want to anyway (me too... :yes:). It couldn't hurt, so long as you are comfortable tearing into them. The markings on the diaphragms are not always intuitive...

btw - what are all the transducers in this system? You've not posted that, far as I can see...

3. Good to know, and thanks for the feedback. I really like those guys, and George in particular has provided me, for one, with outsanding support. :applaud:

[stevem]

I checked the system again. When I reverse the polarity on the midrange, the dip at the tweeter/midrange crossover frequency disappears, but a dip at the woofer/midrange crossover frequency appears. When I just reverse the polarity on the tweeter, the dip at the tweeter/mid crossover disappears. It's looking like you an Zilch are correct.

I bought the compression drivers used, so I guess anything is possible. I did peek inside once just to confirm that the diaphragms were the correct ones. They appeared so.

The drivers in the system are as follows:

Woofers - 1401Nd (one pair per side)
Midrange - 2012H
Tweeter - 2450SL with PT-F1010HF waveguide

[boputnam]

Often we find here that ears are the most important guide. If the reversed polarity on the HF is more pleasing - and removes the cut (cancellation...?) in the frequency response - go with it. Double-check when the Cricket arrives and let us know. Hell - we should form a rental pool for the Cricket! :rotfl:

When you get done with the triage, post some pics of these - they "sound" really intriguing...

[Don C]

I can't see why this is taking so much discussion. The solution seems obvious to me. Maybe I just don't get it though. Seems to me that the notch is because the tweeter is out of phase with the midrange in the range where they both have output. When you swap the polarity if the midrange you fix the problem at the mid-high transition, but create it again at the woofer-mid transition. So put the midrange polarity back where it was and swap leads on the tweeter instead.


I have been switching back and forth between a 2-way and a 3-way speaker system configuration. All the drivers each have their own power amp channel, and are actively crossed over.

I noticed that when using the 3-way configuration, my 1/3 octave spectrum analyzer indicates a drop in response of about 9 db right at the upper crossover frequency (1.23khz). If I invert the phase of the mid driver, the notch disappears, but another notch appears around the lower crossover frequency (297Hz). I do not see this phenomena when using the system as a 2-way with the crossover at 1khz.

The tweeter and the woofers are positive to the black terminals, while the midrange is positive to the red terminals. I have compensated for this in the wiring to the divers from the power amps.

Does anyone have an idea what may be causing this and how to correct it?

[stevem]

I can't see why this is taking so much discussion. The solution seems obvious to me. Maybe I just don't get it though. Seems to me that the notch is because the tweeter is out of phase with the midrange in the range where they both have output. When you swap the polarity if the midrange you fix the problem at the mid-high transition, but create it again at the woofer-mid transition. So put the midrange polarity back where it was and swap leads on the tweeter instead.

The question then becomes, why is the tweeter's output out of phase when it is wired in-phase? I'd like to know why this is. Besides, maybe we all have nothing better to do! :D

[Zilch]

The question is: Why does adding the midrange throw the tweeter out-of-phase?

Now you know why I don't mess with three-ways.

Well, uhmmm, generally speaking, that is.

[Never mind. Rest of crossover parts just came in.... ;) ]

[boputnam]

The question is: Why does adding the midrange throw the tweeter out-of-phase? You think so, Zilchster?

It may be that the HF is out-of-phase with the MF. In the area of overlap between the two, there is cancellation (as you suggested above). This cancellation may also be happening with the two-way runs (LF and HF only), but is less of an issue probably due to the roll-off of the LF response.

Don't for sure know until the Cricket arrives...

[DavidF]

All drivers should be in the same polarity for the 24/48dB (and therefore in phase) on the theoretical side of things. In actual use you have drivers with different horizontal acoustic centers. Also, the acoustical response may be combining with the electrical response and providing something other than the predicted 48dB roll off/on. How these variables effect the total is the essence of problems in DYI.

Hopefully the combined response characteristics (acoustic/electrical) are still near the desired 48dB. The very narrow bandwidth overlap should be minimal in terms of effecting phase issues, ie only in a narrow segment of the overlap. I suggest wiring all drivers with the same polarity (with the expected polarity confirmed as suggested above). Start at -0- delay in the crossover. Then tweak the delay or phase on the high xover in the attempt to minimize the cancellation effect. Also, be sure you are measuring on the listening axis. There has to be some nulling effect in the vertical direction, up or down, due to the vertical driver offset on the baffle but the lobe should be close to -0- degrees if the xover overlaps are conforming to the predicted response. Go way up or down and you should expect some cancellation effects.

Finally, don’t be too keen on getting the best measured result. Rely on your ears as well.

DavidF

[stevem]

You think so, Zilchster?

It may be that the HF is out-of-phase with the MF. In the area of overlap between the two, there is cancellation (as you suggested above). This cancellation may also be happening with the two-way runs (LF and HF only), but is less of an issue probably due to the roll-off of the LF response.

Don't for sure know until the Cricket arrives...

I just double checked the system as a 2-way. There is no notch at the crossover frequency with the same polarity on woofer and tweeter. When I invert the tweeter, the notch appears. I'm hoping the Cricket arrives Monday.

[Mike Caldwell]

Hello
I have the cricket set of polarity testers. It works great when testing directly to a driver. When testing a passive system through the crossover network sometimes you will get an inconsistent reading primarily on a high end driver or tweeter. I guess some tricky passive eq and impedance compensation networks can fool the pulse that the unit puts out. When directly connected to a tweeter for testing I would recommend a cap in line between the sending unit and the tweeter it's not a bad idea to use a cap with a compression driver test either. It puts out a '"cricket" like sounding pulse every couple of seconds that does contain a far amount of low frequency.
As for whats right or wrong of having all or some of the drivers in phase or inverted from each other it's just a matter of how it sounds. Many designs do it either way.

Mike Caldwell

[stevem]

Thanks for the tips Mike, I appreciate it. My system has no passive components, so that shouldn't be a problem. I'll use a capacitor on the compression driver for safety. I'll let you know how I make out.

[Zilch]

Am I way of when I say that the phase changes 180' every 6db on a filter?It's 90° per pole (6 dB), actually, and the series protection cap alone, itself a single-pole filter, will shift it 90°....

[Earl K]

Am I way off when I say that the phase changes 180' every 6db on a filter?
Yes Baldrick, you have it wrong. But not to worry, we all have a lot to learn .


It's 90° per pole (6 dB), actually, and the series protection cap alone, itself a single-pole filter, will shift it 90°....

(A) For the sake of clarity ;

(1) Phase changes induced by filters, typically occur over in the "Stop-Band" not the "Pass-Band" . ( The transitional area between these bands is very loosely called "the crossover point" by some of us. )
- So , a protection capacitor of sufficient value , placed in series with a driver , will have a negligable phase shift up in the drivers pass-band ( the area we want to hear ). Sufficient value equates to; some size of capacitance that places F3 at about one & a half to 2 octaves below the frequency area that one doesn't want to be phase shifted .

(2) The 90° shift ( rotation ) occurs when the effect of the filter has been fully realized .

(3) At F3 , the phase shift is 45° ( for an N =1 pole filter ) .

(B) A Formula ;
- To find a frequency dependant phase angle in a simple series circuit ( when either a capacitor or a coil is placed inline ) use the following ;

Ø ( Phase Angle ) = arc tan ( E1 / E2 ) ( I've bench tested this formula and it does work very well )

E1 = the measured voltage across the load.
( lets say the measured voltage across a load resistor [ instead of the driver ] in a single pole crossover )

E2 = the measured voltage across the inline capacitor or the inductor .

( A person needs a scientific calculator that will give cosines, tangents, arc-tans, etc. , a DMM [ set to read AC voltage ] and a sine wave generator capable of delivering output of a couple of volts ) .



BTW ; measuring changes of phase angle within parallel circuits just isn't this easy . That formula involves measuring the changes in current .

:)

[boputnam]

(A) For the sake of clarity ;

(1) Phase changes induced by filters, typically occur over in the "Stop-Band" not the "Pass-Band" . ( The transitional area between these bands is very loosely called "the crossover point" by some of us. )
- So , a protection capacitor of sufficient value , placed in series with a driver , will have a negligable phase shift up in the drivers pass-band ( the area we want to hear ). Sufficient value equates to; some size of capacitance that places F3 at about one & a half to 2 octaves below the frequency area that one doesn't want to be phase shifted .

(2) The 90° shift ( rotation ) occurs when the effect of the filter has been fully realized .

(3) At F3 , the phase shift is 45° ( for an N =1 pole filter ) .

(B) A Formula ;
- To find a frequency dependant phase angle in a simple series circuit ( when either a capacitor or a coil is placed inline ) use the following ;

Ø ( Phase Angle ) = arc tan ( E1 / E2 ) ( I've bench tested this formula and it does work very well )

E1 = the measured voltage across the load.
( lets say the measured voltage across a load resistor [ instead of the driver ] in a single pole crossover )

E2 = the measured voltage across the inline capacitor or the inductor .

( A person needs a scientific calculator that will give cosines, tangents, arc-tans, etc. , a DMM [ set to read AC voltage ] and a sine wave generator capable of delivering output of a couple of volts ) .

BTW ; measuring changes of phase angle within parallel circuits just isn't this easy . That formula involves measuring the changes in current .

:)I was gonna post all that, but "get a Cricket" is easier to type... :p

[Zilch]

I was gonna post all that, but "get a Cricket" is easier to type... :p

:rotfl:

[stevem]

The Cricket has arrived. What a neat little testing device. It works perfectly. I have a fairly complex (for a home) system, and it really came in handy to see if everything was hooked up correctly. Since I make all my own interconnects, it was great for testing them too.

Anyway, I discovered that my system is wired correctly, but the whole thing was in reversed polarity. I was able to trace the inversion to the balanced XLR outputs of my Bryston preamp. It's an older model, and was built when Bryston was using pin 3 hot. They later switched to the pin 2 hot convention.

The weird anomaly I described in my first post is still present however. I would expect this since all I did as flip the polarity of all the drivers. It is also present with 24 db/octave L-R slopes as well. I have to do a little more checking to see if I can figure this out.

[Earl K]

Steve,

- It does apear to be a cancellation caused by path length differentials .

- The wavelength of your problem frequency equates to about 11" .

- 1/2 of that ( a 180° difference ) is 5.5" .

- Are the voice coils 5.5" apart ( horn to woofer ) / or something close to this value ?

- I know you've said your crossover has delay / but / you haven't told us how much delay you've entered into the woofers parameters window . What's the value ?

- The other possibility is one of those Oopps !! situations . :p . I assume you're using some sort of dsp-based crossover to achieve such steep slopes. Look through your crossovers internal settings ( go deep into the menu layers ) . If you're using the units presets / compare point by point the different polarity assignments that occur between the different programs. I've seen some units that automatically swap polarity on the midrange output when put into the 3 way mode ( versus the dual, 2 ways ) . I may be thinkiing of the original DBX Drive Rack. When they first came out , I test drove one for a few months. I didn't buy it because it had implemeted too many ( non-industry conventions, which I felt was dangerous ). This is the basis for my suspicion of perhaps some Oopps factor . FWIW, after this demo unit had sat on my desk for a couple of months , it went intermittent. :o:



:)

[stevem]

Tweeter (horn) to midrange is 8" center to center.

Woofer delay = 0, midrange delay = .6", tweeter delay = 1.5".

The DSP I'm using is a BSS Omnidrive Compact Plus FS 366. I have all the polarity settings the same, and all the individual driver phase controls set to zero. They do have a "phase compensation" feature for adjusting phase response at the crossover point (especially for 3-way systems), but I have it defeated. I don't know if there are any other adjustable phase or polarity features on this unit.

I suspect you're correct about the wavelength problem. I have the mid up against the horn until it's almost touching. This seems intuitively like a good idea (point source, and all), but I never really see it done on commercial systems. Is there a min/max distance drivers should be spaced from each other on the baffle?

[Woofer]

I checked the system again. When I reverse the polarity on the midrange, the dip at the tweeter/midrange crossover frequency disappears, but a dip at the woofer/midrange crossover frequency appears. When I just reverse the polarity on the tweeter, the dip at the tweeter/mid crossover disappears. It's looking like you an Zilch are correct.

I bought the compression drivers used, so I guess anything is possible. I did peek inside once just to confirm that the diaphragms were the correct ones. They appeared so.

The drivers in the system are as follows:

Woofers - 1401Nd (one pair per side)
Midrange - 2012H
Tweeter - 2450SL with PT-F1010HF waveguide


.... and with that last statement, lies the problem.

Generally the diaphragms in Compression Drivers work backwards as you would normally expect them to, so yes, do wire your Compression Drivers opposite polarity to what you would normally.
So you were almost on the right track. Wire red to positive for bass and mids and black to positive for your tops.
Well at least that's how it works in the rock'n'roll world.
Good luck.

[Earl K]

Horn to midrange is 8" center to center.

Woofer delay = 0, midrange delay = .6", Horn = 1.5".

- I'm having a hard time visualizing this being workable . Your delay numbers suggest the 1401nd is deeper than either the 2012 or the 2332/2450 .

- Which of these 3 components has the deepest dimension ?
- That component becomes your Zero Reference point.
- Typically ( in these configurations ) the horndriver combo is the deepest component to which all the others are synched/aligned (delayed) .


:)

[Tom Loizeaux]

Remember that all crossovers, passive or active, produce a phase shift of 90 degrees per 6dB of slope. When you switch to a 3-way you need to figure the amount of shift at your slope, and how it affects the other phase relationships.
It may require that you invert one of your drivers each time you switch back and forth.

Tom

[Earl K]

Tom, Steve is dealing with individual slopes which are all multiples of 24 db/octave.

- That means that polarity is his primary concern , along with getting the individual delays entered correctly .

- He doesn't have to bother about counting 90° phase shifts per crossover pole. ( FWIW, the shifts are in fact 45° per driver / per pole / measured at F3. They are only 90° when referenced between 2 drivers at the F3 point. The 2 drivers' "phase" shifts rotate in opposite directions away from the 0° point reference. This is why 2, 45° shifts appear as a single 90° offset/shift .)


:)

[stevem]

- I'm having a hard time visualizing this being workable . Your delay numbers suggest the 1401nd is deeper than either the 2012 or the 2332/2450 .

- Which of these 3 components has the deepest dimension ?
- That component becomes your Zero Reference point.
- Typically ( in these configurations ) the horndriver combo is the deepest component to which all the others are synched/aligned (delayed) .


:)

Actually the 1401Nd is the deepest driver. Yes, usually most compression drivers and horns are the deepest, but the PT-F1010HF is extremely shallow. The 2012H mid driver is just a little shorter than the woofer. I might be a little off in my delays, but they're probably pretty close.

[Earl K]

Actually the 1401Nd is the deepest driver. Yes, usually most compression drivers and horns are the deepest, but the PT-F1010HF is extremely shallow. The 2012H mid driver is just a little shorter than the woofer. I might be a little off in my delays, but they're probably pretty close.

- Oh, OK ( so much for that thought ).

- Steve for the moment I'm really stymied .

- Does your BSS model have an auto circuit for setting delays ? . ( I know some of the higher-end models do .)

- You now need an additional 5.5" offset ( to account for the 1/2 wave cancellation, when measuring at 1230 hz ) . Add this additional amount to the horn delay and remeasure using some sinewaves ( at your crossover point as well as 1/3 octave below & 1/3 octave above your Fc ).

- I'm now thinking the problem may be linked to acoustic centers that are existing in locations that are significantly different that the voice coils of the drivers. Given your components, this doesn't make a ton of sense to me / but oh well.

- An error ( when compared to your mechanical alignment ) of almost 1/2 foot seems extreme, but I'll keep an open mind.

- BTW ; I have a bunch of literature that places the acoustic center for lower frequencies in CD type horns out towards the bell lip. How far out into the bell - depends on the LF loading point of the horn. Maybe this is where this 5" plus has disappeared to .

[stevem]

- Oh, OK ( so much for that thought ).

- Steve for the moment I'm really stymied .

- Does your BSS model have an auto circuit for setting delays ? . ( I know some of the higher-end models do .)

- You now need an additional 5.5" offset ( to account for the 1/2 wave cancellation, when measuring at 1230 hz ) . Add this additional amount to the horn delay and remeasure using some sinewaves ( at your crossover point as well as 1/3 octave below & 1/3 octave above your Fc ).

- I'm now thinking the problem may be linked to acoustic centers that are existing in locations that are significantly different that the voice coils of the drivers. Given your components, this doesn't make a ton of sense to me / but oh well.

- An error ( when compared to your mechanical alignment ) of almost 1/2 foot seems extreme, but I'll keep an open mind.

- BTW ; I have a bunch of literature that places the acoustic center for lower frequencies in CD type horns out towards the bell lip. How far out into the bell - depends on the LF loading point of the horn. Maybe this is where this 5" plus has disappeared to .

I haven't had a chance to do this yet, but since each driver is in it's own enclosure, I can move them relative to each other. This is going to be my next step to see if changing distances changes the cancellations.

Your idea about varying the delay is a good one. I'm going to experiment with the delay also to see what, if any, effect that has.

By the way, the BSS I have does have an auto delay function, but when I upgraded the software to the FS 366t version, a software bug is keeping it from working properly. BSS is supposedly fixing the problem, but I don't see anything happening yet. My only other alternative is to revert back to the older software, but up until now I haven't felt that it was worth the effort.

[boputnam]

The Cricket has arrived. What a neat little testing device. It works perfectly. I have a fairly complex (for a home) system, and it really came in handy to see if everything was hooked up correctly. Since I make all my own interconnects, it was great for testing them too.

Anyway, I discovered that my system is wired correctly, but the whole thing was in reversed polarity. I was able to trace the inversion to the balanced XLR outputs of my Bryston preamp. It's an older model, and was built when Bryston was using pin 3 hot. They later switched to the pin 2 hot convention.Sorry to break in here, doods, but this is a great post. It merits a re-read for everyone of us that thinks we know what our systems are doing! :nutz:

Now, back to the mystery...

[boputnam]

...when I upgraded the software to the FS 366t version, a software bug is keeping it from working properly. BSS is supposedly fixing the problem, but I don't see anything happening yet. My only other alternative is to revert back to the older software, but up until now I haven't felt that it was worth the effort.Now I'm intrigued.

Did the system do this before the firmware "upgrade"...? :hmm:

[Mike Caldwell]

The Cricket has arrived. What a neat little testing device. It works perfectly. I have a fairly complex (for a home) system, and it really came in handy to see if everything was hooked up correctly. Since I make all my own interconnects, it was great for testing them too.

Anyway, I discovered that my system is wired correctly, but the whole thing was in reversed polarity. I was able to trace the inversion to the balanced XLR outputs of my Bryston preamp. It's an older model, and was built when Bryston was using pin 3 hot. They later switched to the pin 2 hot convention.

The weird anomaly I described in my first post is still present however. I would expect this since all I did as flip the polarity of all the drivers. It is also present with 24 db/octave L-R slopes as well. I have to do a little more checking to see if I can figure this out.

Ahh.... the old pin 2 or pin 3 hot dilemma, for a few years it was about a 50 50 guess at best. Even more so if mixing US and European built equipment.
Now most of the world ahs settled on the pin 2 + standard. The older MX series and maybe the EX series of power amps from QSC had standard pin 2 + xlr's but the 1/4 inch tip ring sleeve inputs were wired with the tip - and the ring + . Something to check if your using any of them with the 1/4 inch inputs. Something tells me the last of the MX series was tip + ring -.
On piece of processing equipment like an EQ for instance that was built with pin 3 + inputs and outputs and it was connected to a system that was all pin 2 + and if using standard wired balanced xlr cables to and from the unit you would still be in the correct polarity.

Mike Caldwell

[stevem]

Now I'm intrigued.

Did the system do this before the firmware "upgrade"...? :hmm:

It supposedly did, but I never tried it before I upgraded the software. :banghead: I had an e-mail conversation with the techs at BSS, and they told me about the problem. They are supposed to have a fix in the works, but I got the impression from the user groups on BSS' website that it's not a priority with them.

[stevem]

Ok, I had a chance to play around with this a little more. I discovered that separating the tweeter from the midrange had no effect on the cancellation. Next I tried changing the delays on the midrange and tweeter, and this did have an effect.

I went back and recalculated my delays, and I found that I had made a mistake. The shallowest driver is the midrange, not the tweeter, so it needed more delay than I had given it, by about 2". After increasing the delay on the mid to 2.5", most of the depression at the crossover was gone. There is still a couple of db, but this now seems unaffected by polarity reversals, so it's probably do to other factors. Anyway, it seems that the delay was the biggest contributor to the effect, although I would have thought the cancellations frequency would have been higher given that a 2" wavelength is about 6.5khz.

[boputnam]

Well, this is turning out OK.

You got a brand new Cricket, and we got to hear about it's use.

We got to learn a whole lot more about phasing and interference than we usually do. Many are clueless of the impacts of this when they are struggling to improve their sound.

Good stuff.

[GordonW]

What we're overlooking here, is the mass-rolloff of the woofer itself.

All drivers, at their upper limit, have a second order (12 dB/oct.) low-pass rolloff...

At the 2-way crossover point, the woofer ITSELF is probably rolling off, adding phase shift. That, combined with the time delay difference between the woofer and tweeter, is causing enough delay to "bring the phase circle around" to the point where the polarity comes out correct. However, the midrange you're using DOES NOT have a rolloff at the same point... it's still "flat" out to well beyond the mid-tweeter crossover point. No phase shift... therefore, you're on a completely different place on the "phase circle"... and hence, the need to reverse tweeter polarity.

Don't sweat it... it's just simple physics. Nothing wrong. Just wire it to where the response is closest to flat in either case, and let 'er rip...

Regards,
Gordon.

[Earl K]

After increasing the delay on the mid to 2.5", most of the depression at the crossover was gone. There is still a couple of db, but this now seems unaffected by polarity reversals, so it's probably do to other factors. Anyway, it seems that the delay was the biggest contributor to the effect, although I would have thought the cancellations frequency would have been higher given that a 2" wavelength is about 6.5khz.

Okay Steve,

- What the above tells me is that your delay corrections are only partially accomplished. You either need to add or subtract some delay from the horn circuit. It's impossible to know because I don't know which relative side of 0° the differential presently resides ( comparing the 2012 to the horn circuit ).

- Your words indicate that you've brought the phase differential between the horn & 2012 , smack-dab into a phasing "grey zone" that is somewhere between full addition of signals & the full subtraction of acoustic signals.
- Because your polarity reversal ( it's the same as adding 180° to the differential ) had minimal to no effect , this means you should be sitting at an approx. phase differential of either 120°( or 300° ). Alternatively, the differential could be 90° ( or 270° ). Any 180° polarity flip would still land any "new" differential into a neutral buffer-zone where "very minimal acoustic addition" occurs .

- The 1230 hz crossover wavelength is approx. 11" in length at 25° centigrade ( I believe that's the temp of reference ).
- Significant increments of this wavelength to remember are :

- Each 30° increment will equate to around .9187 " in distance ( Off-hand, I don't know the closest / most relevant equivalent fraction / & it may be a moot point if your dsp only accepts decimal input )
- 6 of these 30° increments ( 180° ) equates to 5.5".

- If you're happy with the acoustic transition between the 1401nd & the 2012 then leave the delay for the 2012 as it is and prepare to add delay to ( or subtract delay from ) the horn signal ( adding more delay might be your only option / unless the existing horn offset equates to around 2 " ) .

- You won't know if the present phase differential is closer to the 0° point or 360° point ( on the phase-wheel ) / until you add some more delay ( to the horn circuit ) / and track whether you get acoustic addition or subtraction.
- If you get significant acoustic addition on top of what you already have, it means the differential most likely presently sits around 300°. ( If it were at 240° , a movement to a new offset of 300° would be just barely measurable / what with our typical resolutions & of "ad-hoc" methodologies ) .
So;
(1) Add a single 30° increment more delay ( .9187" ) to your horn circuit and make note of whether this adds acoustic signal, or subtracts signal .
- Add another 30° increment and make note if you get even more acoustic summation .
- If you add 2 delay increments and you start to lose signal ( via acoustic subtraction ) it means you're moving your phase differential into the "subtractive phase zone" which exists between 120° to 240° ( see the underlined notes on page 2 )
- But, if the differential is sitting at around 300° , then adding 2 increments of delay ( 1.8374" ) will bring you very close to achieving full signal alignment.
- Alternative method; if your horn delay is still at 1.5" / just defeat the existing horn delay . If you're sitting at 300° , removing that 1.5" will move the signal back towards the 180° point on the phase-wheel / which you'll be easily able to measure as significant acoustic subtraction. If this happens re-engage the delay and get ready to add more on top of the existing (1.5" ??) .
- You may get lucky and find that you're sitting at 120° of differential and by simply removing the existing horn delay, the differential moves back into an area of acoustic summation that exists 45° either side of 0°.

(2) If you are sitting at 120° of differential, then 2 (30°) increments of added delay (1.8374" ) will put the mid/horn signal into a zone of almost full cancellation ( since you've moved smack-dabb into a 180° phase offset ) You won't even begin to leave this subtractive zone until you've added at least an additional 2 more increments of delay / though you may need 240° worth of 30° increments . ( That's 8 times 30° / or 8 *.9187 / = 7.3496" / which does seem extreme )


- I've included John Eargles words & pictures on phase addtions . They somewhat illustrate my message , though I feel I should draw some that clearly demarcate these areas that I call the neutral buffer-zones, as well as their relative locations to 0° & 180° .
- I would have liked to see the "phase-wheel" divided into 4 clearly defined zones that use 3 colors to illustrate what I am on about ( there are 2 neutral buffer-zones that would need the same color and would "buffer" the additive & subtractive zones ) .
- Take a look at the included .gifs & see if you get the bigger picture from these pages .

- This is an exercise in trial & error / and of course, always involves some mental gymnastics .


<> :)

[stevem]

Thanks for all the responses everyone, I appreciate the help.

Earl, you've given me quite a lot to think about. I see what you're saying about adding delay to change the phase relationships between the drivers. It makes sense. You put a lot of effort in explaining it so I'll know how to apply the delay, and how to make some sense of the result. Thank you. Now my problem is the time to experiment with it, because I'm about to leave for a week's vacation. It might have to wait until I get back.

I also should have mentioned another feature of the BSS, but until now I didn't know what, if any, the relevance was. In addition to the ability to select "normal' and "invert" for each driver's polarity, you can also adjust the phase of drivers from 0 degrees to 180 degrees in 5 degree increments. With the unit configured as a 3-way (LMH), you can adjust the mid and the woofer's phase, but not the tweeter. In the 2-way configuration (LH) you can adjust the phase of the woofer. I assume that the tweeter is not adjustable, because the tweeter becomes your reference point, and you are always adjusting the midrange and/or woofer to it. Would you suggest using these adjustments instead of varying the delays? I haven't even tried them to see what effect they have.

I also wonder about trying to adjust phase so you don't have cancellation at the crossover. How do the different phase relationships between drivers effect the overall sound of the system? Is it better to have all the divers in phase with each other, and not worry so much about the crossover points? Is absolute phase reversal audible? I sure think I hear a difference. Am I just being a "tweak" to worry about it?:blink:

[Earl K]

I also wonder about trying to adjust phase so you don't have cancellation at the crossover.
- Then, when you're back from your holidays try my suggestions. ( Or try using the smaller increments that you mention are built into the unit )

- How do the different phase relationships between drivers effect the overall sound of the system?
- Remember that all this talk / is only about phase relationships that exist within the crossover zones. When two acoustic sections are slightly out of phase / you'll get less summation .
- Most people will say that the typical listener is unaware of phase relationships per se / they do point out that what people can hear are the overall level changes caused by phase relationships. .



Is it better to have all the divers in phase with each other, and not worry so much about the crossover points?
- Please, remember you're talking about phase relationships that occur only in the crossover zones / nowhere else ! All three of your drivers have the same polarities. Your new Cricket demonstrated that.
- Out in the pass-bands of the 3 drivers ( which is the greatest percentage of what you listen to ) there is minimal to no phase change to worry about. Your drivers are all moving in tandem / which happens to be the beauty of using crossovers that utilyze slopes that are multiples of 24 db / octave .


Is absolute phase reversal audible? I sure think I hear a difference. Am I just being a "tweak" to worry about it?

- The term is absolute polarity !
- I wish you guys would stop using these two terms as if they were interchangeable .
- Some people claim to hear reversed polarities making a difference on percussive sounds .
- I do find the entire concept of listening to a system that flips polarities in one area anathema to my concept of listening. I understand the necessity to achieve decent summations in the crossover area / but I still don't like the concept. Of course this is the "Normal" situation with just about anything you might buy & listen to, in Home HiFi. ( The whole idea, still drives me nuts which is why, I prefer to make my own DIY systems )


:) Anyways, Bon Voyage !

[Tom Loizeaux]

Guys, this is a great thread! It's a little hard to follow...but that's what makes it so interesting. A special thanks to Earl for sharing his amazing insight and knowledge.
I suspect we would all benifit from figuring out how to tweak the electrical alignments of our home systems.
Thanks again.

Tom