1 - 20 of 70 Posts

Patrick Bateman

·
Joined
·
6,013 Posts
Discussion Starter

Facebook isn't a great forum for technical discussion, there's not enough space to post a proper reply. So here goes:

When two sources of sound are one quarter wavelength apart, they'll reinforce each other, and when they're half a wavelength apart, they'll nullify each other. This is how transmission lines work; basically we separate the front and the back wave by one quarter wavelength, to make it louder at one narrow band of frequencies.

Here's an example of this in action.

A typical sedan is about six feet wide, and the cabin is about 54" wide, or 1.37 meters wide.

The width of the cabin leads to two critical frequencies:

At 63hz and below, the two midbasses will reinforce each other.

At 126Hz, the two midbasses will be out-of-phase.

IE, the sound from your LEFT midbass radiates to the right side of the car, and when it arrives at the right side, the RIGHT midbass is 100% out of phase with the wavefront radiated from the LEFT, and vice versa.

A typical midbass is high passed at 80Hz, so this means that your midbass has a big ol' null right where you'd want it the least, at the very bottom of it's bandwidth.

Note that this isn't an "all or nothing" thing. It's not like we have a suckout that's super narrow at 126hz. It basically means that our left and our right midbass will augment each other as the two approach 63hz, but as you travel an octave higher, they'll interfere with each other.

One thing that's a bummer about waves, is that when two waves augment each other you get 6dB more output, but when they interfere with each other you get a null. IE, destructive interference is a way WAY bigger problem than constructive interference. With destructive interference, you can potentially have no output whatsoever at a given frequency. Of course, in the real world, that rarely happens. But you WILL see dips that can be as big as 10-15dB.

If you've ever wondered why those speakers you have on your desk have nice midbass and they're tiny, while the speakers in your car are twice as big and aren't as punchy, the answer is geometric.

On top of all that, we also have cabin gain kicking in at around the same frequency. (I'll get into cabin gain in the next post.)

Patrick Bateman

·
Joined
·
6,013 Posts
Discussion Starter
Cabin gain is (fairly) simple:

The way that horns work, for the most part, is that they take sound and radiate into into a narrower beam. For instance, 500Hz is 24" long (0.6 meters) so when a horn has a mouth that's bigger than that, all the frequencies above 500Hz are radiated into a narrow beam. This raises the output on axis. Basically you have the same amount of energy radiated into a smaller space. You can get crazy high efficiencies because the space is so much smaller. For instance, if you reduce the angle of radiation from 180 degrees to 90 degrees, your efficiency goes up 6dB, which is like quadrupling your amplifier power. If you reduce it from 360 degrees (radiating in all directions) to 90 degrees, you get a 'bump' of 12dB.

In a car, it's like you're sitting inside of a bass horn.

For instance, 60Hz is 18.75 feet long (5.7 meters.) Due to that size, the wave is constrained by the size of the cabin.

Patrick Bateman

·
Joined
·
6,013 Posts
Discussion Starter
If your subs are in the trunk, and your midbass is in the doors, there's going to be a significant delay between the subs and the door speakers.

For instance, my Accord is over SIXTEEN feet long. Subs in the trunk are about eight feet away from the door speakers.

So, let's do the math:

140Hz is eight feet long. But the critical frequencies are at one-half-wavelength and at one-quarter-wavelength.

1) at one-half-wavelength, the sub will be out of phase with the midbasses. This frequency occurs at 70Hz in my Accord, but your frequency will depend on the distance from your subwoofer to your midbass.

2) at one-quarter-wavelength and below, the output of the sub will augment the output of the midbass, because the two speakers are nearly in phase with each other. This means that the subwoofer will augment the midbass at 35Hz and below. Of course, this is fairly meaningless, because your door midbasses aren't generating much output at 35Hz.

Patrick Bateman

·
Joined
·
6,013 Posts
Discussion Starter
If all of this is making your head hurt, it should, it's hideously complex.

In a nutshell, the octave from about 80hz to 160hz is a freaken' mess. We have one midbass cancelling out the other midbass, we have the sub cancelling out the midbass, and then we have a tremendous amount of constructive interference that starts to happen below 80hz.

It's this weird car audio paradox, where it's ridiculously easy to produce gobs and gobs of output in the two octaves from 20 to 80Hz, but producing sound between 80 and 160Hz is a bit of a nightmare.

There are about a hundred different ways to fix this:

1) The simplest solution by far is to simply use REALLY BIG midbasses and lots of EQ and just power through it.

2) We know that the left and the right midbass will be out-of-phase at a specific frequency based on the width of the car. Due to this, we can use asymmetrical high pass filters. Basically the high pass filter introduces a delay. By making them asymmetrical, we can improve the midbass on one side of the car at the expense of the other side. The asymmetrical slopes move the null from one side of the car to the other. You will never get this right without a microphone, so go buy one and tinker with this. Thanks to Erin H on the FB group for mentioning this solution.

3) We know that the left and the right midbass will be out-of-phase at a specific frequency based on the width of the car. So you can use an all-pass filter to introduce a delay to one speaker. I believe Audiofrog may have a patent on this, so do your research before you go and use this solution.

4) You could use DSP delay to address it. Note that DSP delay works across all frequencies, so if you improve things at one frequency, it will likely degrade things at another.

5) You could use multiple midbasses.

dgage

·
Registered
Joined
·
2,008 Posts
Good stuff Patrick. Thanks for the nice summarion and thanks to you and ErinH for the potential solutions.

I hadn’t thought of an all-pass filter but that is interesting and one I might have to play with.

I’m assuming multiple midbasses would bring about their own issues even if they’re right next to each other.

I’m thinking of rebuilding the bottom of the door panel in my Acura TL to be an enclosure for an 8”.

ckirocz28

·
Registered
Joined
·
1,463 Posts
3) We know that the left and the right midbass will be out-of-phase at a specific frequency based on the width of the car. So you can use an all-pass filter to introduce a delay to one speaker. I believe Audiofrog may have a patent on this, so do your research before you go and use this solution.
I'm pretty sure they can't patent all-pass filters or their usage for this or any other purpose, they could patent a one button midbass "fix" in their dsp software that uses all-pass filters for the fix. But this midbass problem is a perfect fit for an all-pass filter.
Multi Sub Optimizer kind of does this sort of fix of midbass and bass using any eq or crossover type you tell it to use, although I haven't tried it for anything, yet.

dgage

·
Registered
Joined
·
2,008 Posts
I haven’t tried an all pass on the midbass but would that really fix both sides or just one seat?

ckirocz28

·
Registered
Joined
·
1,463 Posts
I haven’t tried an all pass on the midbass but would that really fix both sides or just one seat?
Most likely, just one seat, but that's why dsp's have different selectable presets.

tonny

·
Registered
Joined
·
282 Posts
OR just fix it with one mono midbass in the centre off the car

Playing those frequenties at least up to 150 mono is no problem in most cases even for SQ competition....

bnae38

·
Registered
Joined
·
1,834 Posts
I have a fairly narrow apf on my right midbass to address this at about 170hz. It works to fix the null but I often wonder if it's worth it since it seems to affect staging a bit.

Patrick Bateman

·
Joined
·
6,013 Posts
Discussion Starter
I haven’t tried an all pass on the midbass but would that really fix both sides or just one seat?
It depends

The "conventional" solution to fixing the center image in a car is to use delay.

Delay affects all of the frequencies.

But, realistically, it might be better to only delay the frequencies that are really out of phase.

This is one of those subjective things where you really need to try it and see which solution you prefer.

Here's an example:

Let's say you have a sedan, and in your sedan the left speaker is 51cm closer than the right speaker. The "traditional" solution would be to delay the left speaker by 1.5 milliseconds. The reason that this works is that sound travels 34cm in one millisecond, so delaying the left speaker puts the soundstage in focus.

But here's another solution, pioneered by Audiofrog. (Andy, chime in if I got this wrong.) We know that the left speaker is 51cm closer. 333hz is 102cm long. (speed of sound is 34,000cm per second, which means that 333Hz is 102cm long.) Because of the path length difference, we know that the biggest issue with the midbass in our car is going to occur at 333Hz. Basically at that frequency, the left and the right midbass are out-of-phase. This is because the pathlength is 51cm, 333hz is 102cm long, therefore the left and right are out-of-phase at 333Hz.

Therefore, one solution, instead of using DSP delay, is to use an all pass filter on ONE side that's centered at 333Hz. Basically delay it about one quarter wavelength (0.75 milliseconds) so that the left and the right side are no longer out-of-phase at 333Hz.

This solution probably won't sound as awesome as traditional DSP delay, but it won't wreck the soundstage as badly for the passenger. It's basically applying delay at the frequency where the pathlength difference is most offensive.

And if all of that makes sense, you can see why there's a lot of good reasons to try and equalize your pathlengths as much as possible. Basically the closer you get to equal pathlengths, the lower in frequency you can apply the delay, which will make the left AND the right seat sound more consistent.

dgage

·
Registered
Joined
·
2,008 Posts
Good stuff guys.

Can you guys speak to multiple midbasses? I know in home audio, if you keep the drivers within half the cone diameter, they act as one. But with the short path lengths in the car, I don’t think that is necessarily the case. Has anyone played with multiple midbasses in the car? Maybe 3 5.5” instead of an 8”.

ckirocz28

·
Registered
Joined
·
1,463 Posts
Good stuff guys.

Can you guys speak to multiple midbasses? I know in home audio, if you keep the drivers within half the cone diameter, they act as one. But with the short path lengths in the car, I don’t think that is necessarily the case. Has anyone played with multiple midbasses in the car? Maybe 3 5.5” instead of an 8”.
It's actually based on quarter wavelength distance, not cone diameter. Coupling is the term for that effect. And, for the same reason it could help with low frequency nulls (pathlength differences), it will also hurt imaging/staging. But I haven't tried it, and T/A may correct the imaging/staging while still helping with nulls.
I just think the all-pass approach is better all around.

209555

·
Joined
·
327 Posts
Stop making my brain hurt.

Thanks for the write-up. I'm brand new to active with DSP and am putting together a system as we speak. I need to learn everything I can between now and sometime next month when I turn the power on for the first time.

ckirocz28

·
Registered
Joined
·
1,463 Posts
Stop making my brain hurt.

Thanks for the write-up. I'm brand new to active with DSP and am putting together a system as we speak. I need to learn everything I can between now and sometime next month when I turn the power on for the first time.
The brain pain is far from over my friend, it's just getting started, welcome to DSP land! Just go ahead and plan to spend several hundred hours of quality time with your dsp, and buy a battery charger.

ckirocz28

·
Registered
Joined
·
1,463 Posts
And an inverter to keep your laptop plugged in in the car.

209555

·
Joined
·
327 Posts
And an inverter to keep your laptop plugged in in the car.
Damn, I didn't even think about a battery charger! Makes sense. May as well test and tune without burning tank after tank of fuel. I had no idea it would take so long to dial it in. Do you have any suggestions? I am converting to LTO. I have 18 Yinlong cells for a total of 120ah of Lithium. This chemistry likes to rest at a little above 14V. What would you recommend? I'm buying a DC power supply as well, I suppose I could just build a lead to the battery with a plug and use that?

209555

·
Joined
·
327 Posts
I'll delete my posts so as not to clutter the thread after ckirocz28 responds.

ckirocz28

·
Registered
Joined
·
1,463 Posts
Damn, I didn't even think about a battery charger! Makes sense. May as well test and tune without burning tank after tank of fuel. I had no idea it would take so long to dial it in. Do you have any suggestions? I am converting to LTO. I have 18 Yinlong cells for a total of 120ah of Lithium. This chemistry likes to rest at a little above 14V. What would you recommend? I'm buying a DC power supply as well, I suppose I could just build a lead to the battery with a plug and use that?
I don't know about LTO, but check NOCO chargers, they're pretty damn good and may have a setting that works for those.

ckirocz28

·
Registered
Joined
·
1,463 Posts
Damn, I didn't even think about a battery charger! Makes sense. May as well test and tune without burning tank after tank of fuel. I had no idea it would take so long to dial it in. Do you have any suggestions? I am converting to LTO. I have 18 Yinlong cells for a total of 120ah of Lithium. This chemistry likes to rest at a little above 14V. What would you recommend? I'm buying a DC power supply as well, I suppose I could just build a lead to the battery with a plug and use that?
It doesn't have to take that long to tune, I was kind of joking, kind of.

1 - 20 of 70 Posts