stevie

Todd, I admire your enthusiasm, but I think you will run into problems trying to build a passive version of an active setup. There's no doubt it can be done, but you would really need a good knowledge of passive crossover design and some good measuring equipment to be successful. Working from crossover value tables never works. Believe me, I've been designing crossovers for over 10 years - for some very well-known companies in fact.

The problem with the general Lexus setup is that the bass/mid is positioned 15" from the HF unit. That will screw up any manufacturer's passive crossover, no matter how good it is, because it will have been designed for a setup where both units are in close proximity on the same plane.

I have never seen this mentioned in manufacturers' literature or in the various forums on the Internet. However, if you increase the distance between the mid and HF or the distance of one driver from the baffle, you must redesign the crossover to compensate.

The problem you describe of lower mid frequencies coming from the bass/mid driver is a sign that you have a phase problem with the crossover. If a crossover is reasonably phase correct, you will hear everything from the tweeter, including the bass (listening to one channel, of course).

You will be able to sort this problem out using an active crossover as long as you have the option of phase adjustment (0 - 180 degrees) or delay, which is basically the same thing.

If I had to choose between a bit of hiss and a phase inacurrate crossover, I'd choose hiss every time. After all, you only hear the hiss on quiet passages (over to you, Percy).

By the way, one of the signs that your crossover has phase problems is easily heard in the sound of a modern snare drum. Unless the crossover design is reasonably accurate, the sound of the snare will be thin, emphasizing the higher frequencies. Get the crossover right, and the snare will sound fat and deep (with plenty of HF content, of course). This is one of the ways I tune a speaker crossover. I have a particularly well balanced pair of headphones that I use for comparison. If you get the snare right, most other things fall into place.

Another sign of a phase inaccurate crossover is that the lower treble (upper range of female vocals) sounds harsh. Many people blame this on the tweeter when in fact the crossover is to blame.

Enough rambling- from me, too.-

Stevie

Rico's Revenge

We have checked the phase of all the speakers with a phase detector and everything comes out right.

As far as the passive crossovers go, I've always been a fan of doing things simple. It seems that everytime you add a 12 volt piece, your potential for disaster increases. However, it all comes down to the sound. If I can't get it right with passives (and adding a couple of design points as a bonus) then I'll be forced to go with the active. Are there any others besides the ones Percy suggested that you would consider?

Other than that, if Percy and yourself can keep the dialog going, I would appreciate the opportunity to learn.

Thanks

Todd

BTW...here's a site with some pics of the car

http://www.streetsourcemag.com/ViewO...ProfileID=6898

The pics aren't updated with the trunk vinyled out but you'll get the idea.

Percy

Stevie,

Correct me if I'm wrong, but the only two areas that you'll get a PERFECT phase response are the following...

Headphones. No crossover necessary.
Live performance or actually in the performance. (no xover...grin!)

If memory serves me correctly (hey...getting old ) doesn't a 12db/oct xover delay the signal by 90 degrees? (or was it 45?) In any case, there is a delay with the best of passives unless there's some way to correct for this.

Active crossovers. Here you have two types. One is digitally controlled, analog filters. That is, the signal still goes through passive components. Examples of this are the McIntosh MEN456/McIntosh MEN455. Mc uses the analog filtering for the "ultimate in headroom" (their advertising...not mine!) versus the 16 bit (then) digital limitation of the digital filter networks. This may explain the noise (slight hiss) and their selection of components used (ne5532 op amps/tl074 op amps/caps, etc). The hiss (it's noticable...you may get points docked from it if the judge is picky) can be remedied by swapping out the analog output stage with Black Gate capacitors or Sanyo OsCons. Just by doing this the hiss is reduced to 5 to 10 percent of what it was. VERY significant and I would now rate this as an "acceptable" piece. But, I'm very picky about this type of stuff!

Since they're still using analog filtering networks, the phase of each xover network will, more than likely, not be perfectly in line with each other. There is a certain amount of delay with a 12db/oct versus a 24db/oct xover. Can't remember the exact phase numbers off the top of my head but they do exist since they're still pumping through passive components. Yes, it's "active" in that there are op amps bumping up the signal (level adjustments/frequency adjustments) but it's "passive" because of the components.

Now for the other type. Digitally controlled, digital filters. Examples of this are the Sony xdp4000x, Sony XESP1, Sony XESZ50, Alpine F#1 Status PXAH900. If you've ever taken a look inside one of these units, they're all populated by IC's, or integrated circuits. Yes, you'll see a few capacitors and resistors, but they're mostly for the power supply or some sort of decoupling/grounding. The signal itself goes through an op amp (buffer), then goes to some sort of a/d converter (analog to digital), then to the actual processing and then back out to a digital to analog converter, then finally the op amp buffer (different one). At no time should the signal actually pass through a passive component that's used as a filtering network. It's all done by integrated circuits and digitally.

Here's a pic of the xdp4000x opened up.

http://www.matronics.com/xdp4000x/xdp-4000x-nocasea.jpg

In theory, this type of processor should have a much lower noise floor (the lower it is, the more details you can pick out in a recording) and shouldn't have any phase response problems. Once again in theory, "perfect" phase response.

Hissing versus accuracy in phase response....

Let's say that you're using a home speaker (2 way) with home electronics. The tweeter is crossed over via a 6db/oct (shallow) slope while the midwoofer has no crossover. The tweeter is so many degrees behind the midwoofer due to a delay with the crossover. My guess is that this wouldn't be objectionable since we're pretty much used to listening to this, that is, a certain amount of phase deviation.

Now lets take another speaker, also 2 way. Tweeter is crossed over at 24db/oct and the midwoofer is crossed over at 24db/oct. This is fairly typical of the car audio enviroment. If both drivers are in phase with each other, my guess is that there wouldn't be anything objectionable with this either.

If the drivers are out of phase with each other (180 degrees apart) then you would hear some nasties. Vocals will start losing coherence, staging will be pretty much all messed up.

As long as the drivers are within a certain amount of degrees within each other then the listening will be acceptable. Once when they're really far apart (180 from each other) then that's where things get nasty.

Hissing noise. As long as it doesn't sound like it's from an AM radio then it should be good for most people. Depends on your tolerance to the low level grunge and dirt that's floating around the system acoustically. If you're ear is sensitive to lots of minor defects (mine is!) then it'll probably drive you nuts after a while. Also, if your car is being judged, it depends on his/her tolerance (the judge) to the "hiss" noise.

Component selection. I'd agree with Stevie on the xover, but the component itself will make a difference. Soft dome versus metal domes.

http://www.dynaudiona.com/mainmenu/b...f/Audience.pdf

Take a look at the graph on page 7. Very interesting! Notice the ringing effects of the "typical" metal dome. Page 6 is also good on the rear damping cross section of the tweeter. That explains why there's so much back space in some of the Morel and Dynaudio designs. Mainly for damping and getting rid of any nasties/reflections.

This was a long ramble!

Percy

stevie

Hi Todd,

>We have checked the phase of all the speakers with a phase detector and everything comes out right.

When I was talking about phase I didn't mean in phase or out of phase. That represents a phase difference between the drive units of 180 degrees and is easily corrected by reversing the connections on the tweeter. I meant that the phase difference between the drive units at the crossover point is, say 72 degrees, when is should be zero.

You can adjust for this by redesigning the crossover. But you cannot really get it right unless you have some kind of measuring equipment to help.

Take the crossover designed for your MB Quart system. It has been designed on the assumption that the bass/mid and HF are mounted on the same panel close together. The phase difference between the drive units at the crossover point will be zero (keeping things simple) and both drivers will sum over the crossover area to give you a nice flat frequency response over the half octave to octave where they are operating together.

Now, if you move the HF unit back or forward by as little as half an inch you will get a phase shift at the crossover point that is quite considerable. Just guessing, but that would cause a phase shift of around 110 degrees at 3 kHz. Now move the tweeter away from the bass/mid another 12 inches and you will get even more phase shift.

Your nice crossover is now no longer phase coherent and you have a phase shift of, say 145 degrees at the crossover point that you cannot compensate for simply by reversing the polarity of one drive unit. The frequency response will be very ragged throughout the crossover region and the sound will be 'edgy'. Most people would blame this edginess on the tweeter, because the effect is similar to using a harsh sounding HF unit. You will also get the effect that you have described, of sounds moving from one drive unit to another.

If you are very, very lucky, the total phase shift caused by moving the HF unit will be close to 180 degrees at your crossover point - in which case reversing the polarily of the HF unit will do the trick.

You can compensate for phase shift by introducing a delay in the HF unit to bring the phase at the crossover point back to zero. That is why a good electronic crossover has either a delay facility or a 0 - 180 degree phase adjustment. A theoretically perfect crossover (as provided by an electronic crossover) assumes that your drive units have a perfectly flat frequency response and that they are time aligned, which is rarely the case.

There is a trick you can use to get the phase response of your crossover right without using expensive measuring equipment. Just use a test disk and a sound level meter. Reverse the polarity of the HF unit and adjust the phase control on your crossover so that you get a very deep null at the crossover point. You are looking for a dip of at least 24 dB, although 30 dB is better. Then change back the connections on your HF unit. You now have a phase coherent crossover.

Stevie

stevie

Hi Percy,

>Correct me if I'm wrong, but the only two areas that you'll get a PERFECT phase response are the following...

You'll see from my response to Todd that I wasn't talking about perfect phase throughout the frequency range, but a phase coherent crossover - one that meets the targets for the crossover type.

>If memory serves me correctly (hey...getting old ) doesn't a 12db/oct xover delay the signal by 90 degrees?

You're not getting old, Percy. Second order filters have a 90 degree phase angle at the crossover frequency.

>Let's say that you're using a home speaker (2 way) with home electronics. The tweeter is crossed over via a 6db/oct (shallow) slope while the midwoofer has no crossover. The tweeter is so many degrees behind the midwoofer due to a delay with the crossover. My guess is that this wouldn't be objectionable since we're pretty much used to listening to this, that is, a certain amount of phase deviation.

It's a bit more complicated than that. You have phase shift introduced by the crossover, you have phase shift introduced by the drivers where they roll off from flat, you have phase shift due to the driver offset (the different acoustic centers of each driver), and you have phase shift due to the distance between the drivers. The trick in designing a good crossover is to get all these various phase shifts to compensate for each other.

There is no audible problem with an HF unit being delayed with respect to the midrange driver. What is objectionable is when the target phase response at the crossover point is way out - when, for example, you are targeting 180 degrees and you actually get, say 45 degrees because the drive units are not time aligned at the crossover point or because the crossover slopes are not accurate.

A first order crossover is critical because the drive units are working together over several octaves. I'm sure you're aware that Dynaudio are particularly keen on first order. If you look at their crossover circuits you will see lots of components whose sole purpose is to maintain the phase response of the crossover.

As a side note, the smaller Dynaudio drivers are closely time aligned with the tweeters when mounted on a baffle, thanks to the large dome in their center. Normal speakers have their acoustic center delayed quite a bit because their cones are deeper.

I wouldn't say categorically that you always have to meet the ideal. A certain amount of excess phase shift is sometimes acceptable, particularly at low frequencies. Whether it is or not depends on a lot of things, including the steepness of the slopes and the type of filter. Some filters are more forgiving than others.


>Now lets take another speaker, also 2 way. Tweeter is crossed over at 24db/oct and the midwoofer is crossed over at 24db/oct. This is fairly typical of the car audio enviroment. If both drivers are in phase with each other, my guess is that there wouldn't be anything objectionable with this either.

The fourth-order Linkwitz-Reilly is relatively forgiving of phase errors but incorrect phase can still produce large dips in the frequency response. However, the degradation in the sound due to phase errors is much greater that the frequency response variations would lead you to expect. I think that perhaps the brain is sensitive to phase distortion because it is something that doesn't occur in nature (as you remarked at the beginning of your last post).

In Todd's case I'm pretty sure his sound problem is caused by incorrect phase, because what he described is typical of a crossover with phase errors.

>Soft dome versus metal domes.
Take a look at the graph on page 7. Very interesting!

For all the wrong reasons. That Dynaudio tweeter has a sensitivity of 76 dB at 2 kHz - 10 dB down on 10 KHz! It's basically unusable. The metal dome is around 20 dB louder. Also, the frequency response of the metal dome tweeter isn't typical of a good, modern metal dome tweeter. They're comparing apples and pears. It's misleading advertising as well as being technically inaccurate. You should tell them about it.

Stevie

Rico's Revenge

Good morning gentlemen!

Thanks for the 1st cup of coffee morning reading material! LOL!!

I am truly a small fish swimming in an ocean of knowledge when it comes to cross-overs. I'm just hoping I can learn from you guys. 98% of all the competitors out there don't have a clue about this stuff, so any edge I can get is worth it.

I talked to my friends at the shop last night and I will print off this whole discussion to take to them later. When we were talking about active crossovers, the Sony piece that Percy brought up was mentioned as well as the old XEC-1000. But they thought that if I really wanted to do it, that the old Harmon Kardon piece was one of the best ever built and suggested that I try to find one of those (good luck right?).

Have either of you had experiene with the HK unit? I use HK exclusively for my home electronics (with Definitive Technologies speakers) and I love it.

Thanks again!

Todd

Percy

Stevie,

Is there a program out there that will take all the mounting locations, drivers, and measured (note - actually measuring the drivers!) and then plugging them in to get a crossover design? There's just too many variables in car audio, especially the mounting locations. The stock xover included with manufacturers just seems like a "band aid" fix, especially if you're mounting the drivers on a slightly different axis or distance from one another.

Todd,

When they mentioned the HK piece, did they give a model number? Would like to see what the HK can really do.

Percy

Rico's Revenge

Percy, it was an older unit...I think the CXO-1?

stevie

>Is there a program out there that will take all the mounting locations, drivers, and measured (note - actually measuring the drivers!) and then plugging them in to get a crossover design?

Yes, there are quite a few. I have used a DOS program called Netcalc for quite a few years, which is very good, but I don't think it's available any more. I don't bother with it much now, as I usually have a good idea what I want to do straight off the bat. I use it occasionally for calculating notch filters or sometimes to give me a handle on a particularly difficult drive unit.

You have to feed in the frequency response, phase response and impedance of all the drive units. Then you just work on the crossover as you would if you had it on a breadboard. It won't produce a crossover for you; it will just model the system according to the circuits and values you choose. Some programs have a 'seek' function, where you enter your circuit and get the computer to adjust the values to meet your target slope and crossover point.

Unfortunately, you really need an FFT measuring system to get the driver measurements it needs to work.

I'd send you a copy of Netcalc (it fits on a floppy) but it has a protection system that will only allow you to install it twice to hard disk. And I've already lost one life.

Probably the best program at a reasonable price is LSPcad, which does crossovers and enclosures. The manual itself, which is downloadable, is very informative and is a recommended read for anyone who wants to know more about speaker system design, even if they don't intend to buy the program.

You can download a demo version and the manual at: http://www.libinst.com/LspCAD.htm.

>There's just too many variables in car audio, especially the mounting locations. The stock xover included with manufacturers just seems like a "band aid" fix, especially if you're mounting the drivers on a slightly different axis or distance from one another.

That's the conclusion I've arrived at, too. I think a lot of it is pot luck. Of course, a kick system should work well with the manufacturer's crossover because the drivers are mounted on the same baffle, but the way the drivers are installed in a Lexus is likely to be problematic.

So, although your points on passive/ active are quite valid, I think that active might be the best way for Todd to go. Assuming he can find a device that doesn't hiss, of course- With a 24 dB crossover he will be able to dial out phase errors using the phase control. He will also be able to swap out components easily, which is not straightforward when the crossover is passive.

Stevie

engin_ear

iTrader: (1)

"That's the conclusion I've arrived at, too. I think a lot of it is pot luck. Of course, a kick system should work well with the manufacturer's crossover because the drivers are mounted on the same baffle, but the way the drivers are installed in a Lexus is likely to be problematic."

Another two cents, for whatever it's worth:
First of all, great discussion.
Yeah. Pot luck is made even worse due to a bunch of other things that you have to fight in the car. Unfortunately, correcting the relative phase of drivers in a single home enclosure by compensating with the crossover phase only tends to compensate at sweet spots - yeah, you can compensate for your listening position, but as the drivers are separated more and more, it makes more of a mess. (That is not to say the phase should be ignored - it needs to be set up correctly.) Serendipity is often what makes one speaker system sound better than another. This gets WAY worse in the car domain. Separate the drivers (by necessity!) and now it's an acoustical nightmare, too complex to analyze the system as a whole, freq/phase, with all of the reflections added in, given all of the different physical materials used in a car. My point here is that, until someone comes up with a simulation program to take all of this into account and analyze the system as a whole (I don't think Netcalc can take the car shape and materials into account?), I agree with you Stevie, absolutely it's pot luck. In the car you can set the phase of any of the drivers to some nominal reference point and it could still sound lousy at your listening position. Almost same problem when setting up an EQ...first use the science, then tweak anyway, heh, heh.

Kudos to all of you guys - it's amazing that some of these systems sound as good as they do!

stevie

>Another two cents, for whatever it's worth:

The more the merrier. I think we are all learning a few things from the different perspectives represented here.

>Unfortunately, correcting the relative phase of drivers in a single home enclosure by compensating with the crossover phase only tends to compensate at sweet spots - yeah, you can compensate for your listening position, but as the drivers are separated more and more, it makes more of a mess.

The advantage you have with home speakers (or sound reinforcement speakers) is that you are not sitting on top of the speaker system. The angle between the woofer and tweeter is relatively narrow - only a few degrees. This means that a speaker with a well designed crossover will be smooth both on and off axis, which reduces the sweet spot effect you mention.

I can tell you from experience that, even with a typical domestic speaker with a 6.5 inch driver and 1 inch dome, you will not get an accurate frequency or phase response at 1 metre. You need to get back to around 2 meters before you get an accurate picture of what you hear at the listening position. But it is perfectly possible to design a textbook crossover for a domestic speaker, as many speaker manufacturers have been doing for years.

>This gets WAY worse in the car domain. Separate the drivers (by necessity!) and now it's an acoustical nightmare, too complex to analyze the system as a whole, freq/phase, with all of the reflections added in, given all of the different physical materials used in a car.

Tell me about it!

>My point here is that, until someone comes up with a simulation program to take all of this into account and analyze the system as a whole (I don't think Netcalc can take the car shape and materials into account?)

Crossover design programs don't take room reflections into account. You have to cut all of these out before you feed the information to the program. In a living room it doesn't matter too much. The brain is normally pretty good at ignoring room reflections and listening through to the original sound, although it can be fooled by reflections very close to the sound source. I think this is the only reason we can actually listen to music in the car (you remember those frequency curves Percy published recently?)

I have to admit, I wouldn't have a clue how to input the information from a car speaker system into Netcalc. Where would you measure from? The listening position is above the axis of the tweeter. Do you design for this position, or do you design for a mid point between the drivers like in a domestic two-way system? What do you do about the other channel? The freqency response of the bass/mid for each channel is different because you are listening to the left and right speakers at completely different angles.

When I designed the crossover for my GS, I placed the pod on the floor (in the house), positioned the tweeter 15 inches away and designed for the driver's listening position, i.e. above the tweeter. A fourth order Linkwitz-Reilly worked quite well. Because I did it outside the car, what I've got is only an approximation, but it sounds reasonable. I have no idea if this was the right way to do it. Unfortunately, I haven't been able to take any measurements from inside the car.

>Almost same problem when setting up an EQ...first use the science, then tweak anyway.

That's where active crossovers can be a great help. You can do a lot of things with a passive crossover that you can't do with a standard active crossover, but tweaking a passive x-over means soldering lots of different components on a board and a bucketload of measurements. Not to mention listening tests.

So how do you guys with active crossovers, EQ and digital delay set your systems up?

Stevie

Percy

Stevie,

For tuning I try to keep it as simple as possible. I usually go in first with a test disc and the RTA. Pink noise and A weighted settings, which is pretty much quick and dirty. Level it off as much as possible at the level you're used to listening at. This is where all the little tricks of inverting the phase (via controller), sliding some frequency settings over, the levels and slopes. If I don't have an RTA, (which I don't now) then I go in with test tones and the B&K meter. Takes longer since I have to graph out everything, but it works.

Mic position is where the head would be and the mic is pointed at the windshield.

The fine tuning is performed by ear, after all the nasty initial tuning. Only tune for about 15 minutes at a time, otherwise the ears tire out and after that you'll start doing nutty things to compensate.

I'd be willing to bet that a MLSSA measurement would look like a MESS when it came to car audio. Delays all over the place. Absolute nightmare. Wonder if there's any affordable MLSSA gear out there.

Percy

Percy

Stevie,

I just took a look at the Esotec D260 Dynaudio tweeter, and from the graphs they have published (in paper form only) from the Madisound catalog, the D260 is pretty much flat from 2khz to 20khz. +/- 2dB, which is very good. The slightly lower end D28 will also do this. Makes me wonder why Dyn published the other graph in their Audience brochure - the graph should be much better than the one they published!

The esotar T330d will, at a slightly looser tolerance of +/- 3dB, will go from about 1.4khz to a bit past 30khz, though the official rating is supposed to be only 28khz.


Todd,

Morel Supremos. Finally got a chance to read through the review. According to the author, it's the best they've tested so far. They haven't compared it to the MD330D yet, otherwise the scale would be tilted! They did have a MLSSA graph published and from the looks of it, there is still a bit of resonance/undamped energy going on. If you have a copy of the article (I didn't buy it yet), you'll notice the graph has all of the energy still present up until 0.5 milliseconds. After 0.5ms, there is still a bit of acoustical energy being produced, way up until 2.5ms. With the Dynaudio Esotar T330D, acoustical energy is completely absorbed by 0.5ms. There's nothing after 0.5ms, which pretty much tells me the rear chamber of the tweeter has been properly designed to absorb the backwave of the tweeter. The esotar will respond better to transients, or at least is "prepped and primed" for more material. I would think of this as a "settling time".

Adam,

I know you were looking at the Esotar's for use in your system. In what Ron430 affectionately terms as "THE BOOK" (GS Audio Primer), I basically took apart the tweeter to fit in the A Pillar. (Don't worry Stevie...I'm still thinking of how to fit in the Morel!) I experimented a bit with the T330D that I have at home and removed the face/mounting plate. DO NOT DO THIS if you plan on using these in the car. Reason...as soon as I removed the front plate, at least 40 to 50 percent of the energy was lost. The reason for this is the bevel surrounding the tweeter - it acts as a bit of a reflector. Think of a flashlight. Lets take the Streamlight Stinger (military) or the classic MagLite. Removing the front plate of the tweeter is like removing the reflector. Yes, the light energy is still there, and it's running at the same intensity (at source) but the brightness and focus of the light is greatly dimished. Same thing with the Esotar T330D/MD330D tweeter. This also applies to the other tweeters out there that have a bevel surrounding the dome area. So, leave the plate intact, otherwise you'll lose ALOT!

Which makes me wonder, what would happen if there was a custom bevel surrounding the MD100 in the A Pillars?

Percy

stevie

Hi Percy,

> Makes me wonder why Dyn published the other graph in their Audience brochure - the graph should be much better than the one they published!

I'd blame it on the marketing bozos.

> If you have a copy of the article (I didn't buy it yet), you'll notice the graph has all of the energy still present up until 0.5 milliseconds. After 0.5ms, there is still a bit of acoustical energy being produced, way up until 2.5ms.

That is pretty poor. I've seen much better from very cheap tweeters. Depends how it was measured, though.

> With the Dynaudio Esotar T330D, acoustical energy is completely absorbed by 0.5ms. There's nothing after 0.5ms, which pretty much tells me the rear chamber of the tweeter has been properly designed to absorb the backwave of the tweeter.

At the price I'd expect nothing less. I'd just observe that it is risky to compare waterfall curves from different sources. With slightly different settings you can get quite different results.

>(Don't worry Stevie...I'm still thinking of how to fit in the Morel!)

It was a Scan-Speak- Anyway, I have now decided to replace my Audaxes, which I'm not really happy with, with some Morel MDT 41s. It means taking a saw to them to cut the back chamber down but, as I have them lying around in the workshop, I might as well use them. It is also quite a good tweeter and will hopefully still be so after I've butchered them.-

Stevie

Adam_Peng

Percy,

Thanks for the info.

If I get the MD330D in the near funture, I will put them in the kickpanel area (I am a kickpanel guy...). It is a very tough task to make a kickpanel to accommodate both MD330D and MW150 in such a restricted area. I will just leave it up to my installer to worry about that.

Intalling MD330D in the kickpanel has one potential threat...My wife's high-hill. The protection grille of my current MD-100 is dented because my wife once accidentally step on it. With the loose protection like the MD330D, how can MD330D survive in the same situation....which makes me pretty nurvous. Maybe a high-voltage electric-fense might help...^_^

Adam