way off frequency response...

neomagus00

i just went and took off all the equalization on my HU, then took these measurements - is this normal, for the response to be a 30 decibel range??? i mean, yeah, there's a couple gaps in the grid (to be filled in soon), but damn...

now, granted, i did experience the whole PLD phenomenon as i did this - i'd have, say, a 7.5 kHz tone on and lean forward, and it'd get way louder, so i know that where i held the spl meter matters - i'll set it on a stand next time.

bknauss

What are you using to measure? Doesn't look like an automated measurement device since the points look like they were chosen out of thin air.

Also, where are you measuring? ... Where are the speakers that you are measuring?

There is a lot of room for error.

neomagus00

Originally posted by bknauss
There is a lot of room for error.

i agree, which is why i did it again with more and better-chosen sample points. i placed the meter where my head goes and pointed the mic towards the front of the car, and still got one focked-up graph, included below. i'm gonna do it a third time, with the meter in the center of the car, to eliminate PLD problems. i should have that tomorrow.

these are all measurements of the mm6 install i have, minus the equalization i usually use. i just played a tone and read the meter, then graphed the 40 odd points i got.

exalted512

the extreme lows and highs could be due to the amp not being able to produce the Hz
the really high could be b/c your tweeters are not directly on axis
-Cody

neomagus00

true. this still leaves 2 primary problem areas - right around 1000 Hz, and down low. they're xoverd at like 70 Hz, so what's up with the frequency response being that high in the 50s??

exalted512

dunno, polk's dynamic balance is supposed to prevent that
maybe some type of interference?
is the car running when you do this?
i hope youre smart enough to know the car should be off...lol
-Cody

neomagus00

yeah, the car's off, i just turn it on for a while every 15 or so measurements - my battery doesn't last so long when the system's turned up.

what is polk's dynamic balance??

exalted512

should be in your mm6 booklet
if i remember correctly its supposed to even out the frequency curve and eliminate distortion...it might just try to eliminate distortion....i dont remember...lol
-Cody

bknauss

What meter are you using? If its the Radioshack one, there's a good amount of documentation on the web for what frequencies cause problems... as in there are a lot of frequencies that are off by +/-5dB.

I would measure the frequency response at 1 foot on axis from one of the speakers. If its still COMPLETELY off like it is now, then there's something definitely wrong with the mic/meter.

Thom

One big contribution to the in- car response is reflections. Pretty much all of the surfaces in a car are hard, reflective materials (plastic or vinyl wrapped panels) which aren't at all good for sound. The many reflections are causing many (if not all) of the comb filters (narrow valleys) on that graph. If you've got the time go out and re- measure and place some thick foam here and there and look at the difference it can make.

Where are the speakers mounted? What are you using to make the measurements? Pink noise?

geolemon

To be honest, that really isn't that bad.
I had a worse measurement in my Civic, when first tweaking in there.

Much of this could be due to phasing anomolies, as described here (and some info on how to contend with them):
http://www.betteraudio.com/geolemon/phasing/phasing.htm

While the problem is exacerbated by having multiple speakers, you can see that even with a simple 2-speaker setup, and theoretically perfect crossover/phase response, you still have phasing anomolies, due to your offset seating position in the car.

Some of these phasing anomolies are due to reflections, as Thom mentioned... the secondary reflected sound having an inherently different pathlength than the first.

There are two easy ways to clean this up:
1) go the "all the way" route, and turn your car into a single-seater, with the seat right in the center.
2) apply delay to the near side speaker to virtually equalize the arrival times at your ears.
Alpine head units mostly have this feature built in now, but Pioneer and others are also starting to include it. Bear in mind, this will mess things up for the passenger side of the car even more - but probably you don't have an audiophile sitting there anyway.

As far as the reflections go - there's not a lot you can do about it.
You can roll the windows down, enjoy the fresh air... and of course get enough wind noise to kill your SQ. But hey, great on a nice day.

But seriously, one somewhat effective thing you can do - tint your windows.
And you might consider doing multiple layers of light tint, over a single layer of dark, because it will absorb higher frequency sound better.

Just some thoughts.

Interesting, I thought, that your measured plot resembled my simulated plots that I illustrated that article with quite a bit. .
Might be something to look into...

EQ won't help phasing issues, actually, by nature...
... there's an equal and opposite cancellation force as you increase the amplitude of any frequency being cancelled.

neomagus00

meter - radioshack

speakers - mm6, mids in lower door panels, tweets in dash

sound - i'm using specific test tones to measure, so for the 2 kHz data point i used a 2 kHz sine wave.

interior - most of the car is either leather or padded in some way, including the doors and the roof; the only hard stuff is the dash and the windows.

correction factors - i could find about a dozen different correction factors (like -2 at 750, etc.), and instead of simply applying them, i extracted an equation for the corrections and applied that. for reference, the correction equation that i got, to be added to whatever has been measured, is (4 X 10^-8)(Hz^2) - (8 X 10^-4)(Hz) + (0.0469). this fits the measured data to within 1.5 dB, close enough for my work.

phasing - for the most recent graph, i held the meter along the centerline of the car, approximately equidistant from each speaker. the mic was equidistant from each mid to within an inch, same with tweeters; mid-tweet difference may have been up to 6 inches. so, phasing shouldn't have been a problem until the highest frequencies.

i'm gonna add a bunch more frequencies (like 150 more) and see what i get. if i get a saw-tooth graph, i'll know its phasing and i may have to get a different HU .

this one's much better - the 100-1000 range has evened out a lot, probably because of where i held the meter. 1k-4k is still low, and beyond there i don't really care right now. below like 75 i didn't care about this time either, cause it's been xovered there and i'm not worrying about it right now.

bknauss

neo, you might want to think about building a Wallin Jig/Preamp/Mic and using Speaker Workshop. It should cost under $40 for everything, and assuming you've got a laptop or a computer you could place near the car, it would help out IMMENSELY with the measurements.
http://www.gti.net/wallin/audio/audio.html
http://www.speakerworkshop.com/

Looks like the response is starting to figure itself out

Thom

Originally posted by neomagus00
meter - radioshack

speakers - mm6, mids in lower door panels, tweets in dash

sound - i'm using specific test tones to measure, so for the 2 kHz data point i used a 2 kHz sine wave.

interior - most of the car is either leather or padded in some way, including the doors and the roof; the only hard stuff is the dash and the windows.

Leather is about as reflective as anything in the car- plastic or vinyl wrapped panels/ dash, console, etc. A slight bit of padding (a layer of carpet) will only help very high frequencies, the lower the frequency you want to affect the thicker the foam/ padding/ carpet has to be.

Thom

Are the mids behind stock door panels?