PDA

View Full Version : Asking for help/explanation on x-over problem



pangea
09-18-2005, 10:05 AM
Could someone please help me explain what might have happened here and why?

That magical 3-D holographic soundstage had somehow disappeared from my MTM hybrids, (as shown in my avatar) and instead the music has been coming directly from the speakers and for a while I didn't have a clue what was wrong.
Then I ran the test tones and discovered that there was almost nothing coming from the 2235's which puzzled me at first, then I checked both my Behringer Ultradrive's and found that one of the 2235's had become inverted somehow, although I know everything was fine before. OK, I have been experimenting a bit with different characteristics and slopes for a while and somehow I always seem to end up coming back to the "Bessel" and at the moment I'm using 12dB/oct through out.

So naturally I fixed that problem and I also let the Ultradrive align it self once more, since I still wasn't happy with the soundstage, but after I let the Ultradrive do it's thing it still wasn't OK, the soundstage was still fuzzy, even though the bottom end was much more firm again and with all the power I was used to from before.

Then I noticed that the Ultracurve tried to compensate a dip at about 90Hz all the way upp to 1000Hz, which was later explained by the fact that the Ultradrive had inverted both 2235's with 0.09mS delay. The 2123's were normal with a 0.43mS delay, as were the 2441's normal at 0mS and the 075 were inverted with a delay of 0.50mS.
Both 2235 and 2123 sit in vented boxes.

Then it hit me the 2235 and the 2125 should both be wired with the same polarity, should they not, since they blend from 90Hz to where they meet the 2441 at 1000Hz, as was kindly suggested by Mr. Widget, to work as a kind of sonic "glue" and which it does well BTW.
The distance between the 2441 and to the 2235 and the 2123 is equally 34cm from center to center, (therefore the chosen x-over frequency 1000Hz).

So now I wonder why the Ultradrive wants to give the 2235 and the 2123 opposite polarities? Has it something to do with the second order slope perhaps?

However, when I run both 2235 and 2123 with the same polarity, the soundstage becomes absolutely rock steady and all of a sudden every instrument can be located down to an inch.:D

I would also appreciate if you can see anything wrong with my thinking.

BR
Roland

scott fitlin
09-18-2005, 10:23 AM
In my manual for the BSS 366T, they say that if you use the auto alignment, it will compensate for known phase differences, except when using 3rd order filters, so my guess is your unit is trying to do the same.

Either there is a way to tell your Behringer not to do this, or you can set up your parameters manually.

pangea
09-19-2005, 01:33 AM
In my manual for the BSS 366T, they say that if you use the auto alignment, it will compensate for known phase differences, except when using 3rd order filters, so my guess is your unit is trying to do the same.

Either there is a way to tell your Behringer not to do this, or you can set up your parameters manually.

OK, thanks! That's what I thought. So, it seems the 12dB slope, is what's causing that polarity shift.

But how might that affect the overall performance and in particular, the polarity of the other drivers, when I'm forcing the 2235 and the 2123 to play with the same polarity, due to the fact that they blend from 90 to 1000Hz. There was no problem giving them the same polarity manually and store the settings.

But I'm thinking now, if I Should perhaps go for the 3rd order slopes instead, or maybe the 1st order and let the Ultradrive decide the polarities, without me having to interfere in that process?
There is no such thing as a 6dB "Bessel", am I right!?

Any suggestions?

BR
Roland

scott fitlin
09-19-2005, 09:18 AM
I have to study this for a bit! I know 3rd order filters are inherently out of phase, especially in the midband using two crossover points, which is why when using 3rd order you invert the polarity of your mids to bring them back into phase acoustically. With the BSS, it assumes you have done this, and therefore doesnt perform phase compensation when using 3rd order filters so you maintain the correct acoustic response.

When using other filter types, you have to disengage the phase compensation, it is in the menu, if you dont want it to do this. Its on/off!

pangea
09-19-2005, 09:40 AM
I have to study this for a bit! I know 3rd order filters are inherently out of phase, especially in the midband using two crossover points, which is why when using 3rd order you invert the polarity of your mids to bring them back into phase acoustically. With the BSS, it assumes you have done this, and therefore doesnt perform phase compensation when using 3rd order filters so you maintain the correct acoustic response.

When using other filter types, you have to disengage the phase compensation, it is in the menu, if you dont want it to do this. Its on/off!

OK, could that explain, that the soundstage was still a bit fuzzy, when I tried the 3rd order, but when I tried the 1st order slopes, it all came together.
But I did also take away the HP filtering from the 2123, so that it can go down to its natural roll off and I also let the 2441 go all the way up and at the same time I let the 075 meet from above with a 1st order slope and the x-over point set at 20.000Hz.

It works great and for the moment I'm very happy.

BR
Roland

frank23
09-19-2005, 12:39 PM
There is no such thing as a 6dB "Bessel", am I right!?

No, 6dB is 6dB because it is only one component and you cannot tune two or more components in relation to each other like linkwitz, bessel, butterworth etc.

frank

Earl K
09-19-2005, 01:35 PM
Then it hit me the 2235 and the 2125 should both be wired with the same polarity, should they not, since they blend from 90Hz to where they meet the 2441 at 1000Hz,,,

- Yes,,, typically any MTM uses dulpicate woofers. With identical woofers one would never think of running them with different polarities .


So now I wonder why the Ultradrive wants to give the 2235 and the 2123 opposite polarities? Has it something to do with the second order slope perhaps?

- As Scott has suggested ; just disarm the auto phase-flip feature / it's useless for MTM work ( do keep the auto time-align engaged, assuming they are separate ) .


<> :cheers:

EDIT:

- The fact that your UltraDrive machine choose different polarities for the two woofer types just means it detected something it didn't like and attempted to fix it. It likely detected some "lobing" of some sort. Microphone position can also greatly effect this. Lobing is a fact of life with MTM systems. There is a stable listening window in which the imaging is unbeatable / outside that window , it starts to break down . The machines' solution to lobing was worse than the "problem" it detected. Remember "automatic whatevers" are just some electronic manisfestation of the programmers own crossover philosophy ( with his "logic" burned into the chip ) . You can bet ( in this case ) the programmer wasn't thinking about MTM crossover design.


- Back in the second post of your MTM thread , I stated you were going to have to try out different slopes to find the best fit between the "M(s)" & the "T" section . There is no road-map for this stuff / just instincts and intuition.
- What slope are you now using on the 2441 ?

- And yes, I would choose a filter type that gives the best transient or step response / since these systems actually let one hear these choices ( when properly implemented ) .

pangea
09-20-2005, 10:07 AM
No, 6dB is 6dB because it is only one component and you cannot tune two or more components in relation to each other like linkwitz, bessel, butterworth etc.

frank


Ahhh, yes, I knew there was something and now I know why. Thanks for the info.

While we are on the subject, if you don't mind explaining it in simple terms. Why is there no 3rd order Bessel?
I kinda like the Bessel and use it where ever I can. :D

BR
Roland

pangea
09-20-2005, 11:11 AM
- Yes,,, typically any MTM uses dulpicate woofers. So, one would never, ever think of allowing duplicate woofers to run with different polarities .



- The slope is really a secondary issue / though one type may produce more lobing than others. Experimentation , is still the key here .

- The fact that your UltraDrive machine choose different polarities for the two woofer types just means it detected something it didn't like and attempted to fix it. It likely detected some "lobing" of some sort. Microphone position can also greatly effect this. Lobing is a fact of life with MTM systems. There is a stable listening window in which the imaging is unbeatable / outside that window , it starts to break down . The machines' solution to lobing was worse than the "problem" it detected. Remember "automatic whatevers" are just some electronic manisfestation of the programmers own crossover philosophy ( with his "logic" burned into the chip ) . You can bet ( in this case ) the programmer wasn't thinking about MTM crossover design.


- Back in the second post of your MTM thread , I stated you were going to have to try out different slopes to find the best fit between the "M(s)" & the "T" section . There is no road-map for this stuff / just instincts and intuition.
- What slope are you now using on the 2441 ?

- And yes, I would choose a filter type that gives the best transient or step response / since these systems actually let one hear these choices ( when properly implemented ) .

- As Scott has suggested ; just disarm that silly auto phase flip feature / it's useless for MTM work ( keep the auto time-align engaged, assuming they are separate ) .



<> :cheers:

I was afraid of that and the 2235 and the 2123 would "kill" each other off with opposite polarities, right?

I have positioned the microphone at exactly the double distance than the distance between the woofers. (cc 34cm - 0 - 34cm, which makes 68cm and I have chosen the distance from baffle to mic 136cm) I cannot explain my reasoning for this, other than that it felt right and that I figured the two lobes would become uniform at that point. As I said don't know if this makes any sense. :o:

After having experimented with various slopes, I found that the imaging, definitely is best when I'm using the 1st order slopes throughout and let the Ultradrive align the drivers, without me having to interfere with the process too much.

I'm not sure if that is a good thing to run the woofers at 6dB and at the same time, having them meet the 2441 at 1000Hz x-over, at a different slope, say 3rd order, or is that OK?

I am somewhat concerned that the Selenium horn, which I'm using, might run out of steam at 500Hz, with the 6dB x-over at 1000Hz (optimal with the chosen MTM distances). The Selenium horn is said to work down to 500Hz, but I would prefer, not to be forced to use it to its limits.

I have to agree totally, when the imaging is good, it is EXTREMELY good!:applaud:
I've listened to one or two speakers over the past 30+ years and I have NEVER heard speakers with better imaging, or detail for that matter, regardless the price tag and here I'm using a bunch of old surplus JBL stuff, ... go figure! :D
I love this stuff!!! I also love when people brag about their heavy duty mega buck speakers, what ever the brand and I invite them to come listen to my old DIY stuff...:D
Need I say more? :blink:

BR
Roland

Earl K
09-20-2005, 12:12 PM
I'm not sure if that is a good thing to run the woofers at 6dB and at the same time, having them meet the 2441 at 1000Hz x-over, at a different slope, say 3rd order, or is that OK?

I am somewhat concerned that the Selenium horn, which I'm using, might run out of steam at 500Hz, with the 6dB x-over at 1000Hz (optimal with the chosen MTM distances). The Selenium horn is said to work down to 500Hz, but I would prefer, not to be forced to use it to its limits.

- If you're concerned about power-handling issues then try using a 2nd order Bessel on the 2441/Selenium horn. This filter still maintains good transient qualities , as well as having virtually nil group delay qualities . Another idea is to put a DC blocking cap in the circuit 1.5 octaves down from an 800 Fc point . This provides some protection , as well as offers a bit of phase rotation around the crossover point and below. A 34uf cap is about what you need for this. It's something to try.

- I wouldn't manually ( or allow the machine to automagically ) flip the polarity of the horn circuit. There's a difference to the blending which I don't like / though, maybe your setup will be different .

frank23
09-21-2005, 12:28 PM
What one should read and what I have been treasuring for many years, is the article by mr linkwitz of which I have a copy at home, but I haven't been able to find a digital version. It's not even on his own very very informative website:

17 - Siegfried Linkwitz, Active Crossover Networks for Noncoincident Drivers, JAES, Vol. 24, No. 1, January/February 1976. Reprinted in Loudspeaker Anthology, Vol.1, AES 1978, Abstract

what also might be interesting is this, but I have never seen it:

16 - Siegfried Linkwitz, Passive Crossover Networks for Noncoincident Drivers, JAES, Vol. 26, No. 3, March 1978, Abstract

The first article is about "The spatial seperation between drivers in a loudspeaker system that affects the radiation pattern over the frequency range where more than one driver contributes to the total acoustic output." The article is 7 pages.

His site is www.linkwitzlab.com

Frank