DIYMobileAudio.com Car Stereo Forum banner

Thoughts on my system response - I'm pulling my hair out

3K views 49 replies 17 participants last post by  M-Dub 
#1 · (Edited)
I recently set up a PC-based RTA and had initially found it useful for better balancing my sub in my system. I have had a number of threads asking questions about things I was seeing and hearing that I think were artifacts of the calibration of my measuring system. Now I've got things straightened out, I'm looking at understanding the higher frequency response to help me get a more level sound that works for most of the different styles of music and recordings that I listen to.

What really helped me with sub integration was to measure both channels simultaneously from the driver seat, taking an average of measurements made around the head location. The following curves represent these measurements, both sub on (green) and sub off (purple):



On average, relatively flat from midrange on up, but first thing I notice on the curve are the +/- 5 dB swings on the way from about 200Hz to 10kHz. I know a little bit of peakiness is good to prevent a system from sounding sterile, but my impression is that +/- 5 dB in this type of pattern can add a lot of inaccuracy to the sound.

A couple of questions for the experts:
  • Considering I find myself re-EQ'ing the system just about every change in song or recording, is it possible that these peaks and valleys are noticeable and I'm trying to correct for them to better match the recording?
  • I've taken the time to do things right on the installation which I had always thought would minimize these types of sound quality problems, but would one expect such a peaky response from a well installed system? Should I be looking at other installation issues? As some background, the front stage is a passive Alpine SPX-17REF installed with mids in the doors (firing directly out) and tweeters on-axis in the sail panels, separated by about 24". There is no processing.

With minimial EQ capability to tweak what looks like a range of problems (3-band parametric in the Kenwood DNX7100 head unit), I set out to take some more measurements to look for a potential installation issue. Looking at left (yellow) and right (orange) channels independently:



My interpretation is that they both look pretty bad individually (although they get better when measured simultaneously above), with a significant valley on the left channel at about 120 Hz and the same on the right nearer to 220 Hz. From about 1000Hz on down to about 100 Hz, I see what looks like mirror images of the peaks and vallies in the plots.

Again for those more experienced reading these plots:
  • Does the mirror imaging from 1000 Hz on down to 100 Hz look like it represents any issue, or just a relative volume difference with the speakers? Axis positioning of the woofers in the doors potentially contributing to this?
  • Since I'm not measuring anything versus time, I would not expect that this is a phase issue, correct (although this is where my inexperienced mind jumped when I saw the mirror imaging)?
  • Since the issue looks like it spans only part of the midrange region(100 Hz on down looks consistent on both sides), thoughts on how to correct it without throwing off the midbass, or if it even needs correction (if the combined response above looks better)?

I want to thank the group in advance for any thoughts. I'm on a last ditch effort to try to get things right in this system so I can start enjoying it. With limited processing, I'm trying to ensure that all the installation issues are resolved, then I can consider whether or not I need to make an upgrade in this area. I'm just surprised that a well installed system can sound so "not good" without the processing power.

Thanks!
 
See less See more
2
#2 ·
What are you measuring your response curves with? (pink noise, etc ...)

They look really ragged. You say that you took time to do things right in the installation, but I didnt see any mention of where wyour drivers are mounted/aimed, etc.

Just by the look of those plots, you need some more EQ work for sure, and unfortunately that 3 band parametric is not going to cut it.
 
#3 ·
That's pink noise, 10Hz - 20kHz, lossless burned to CD.

I should clarify "right" with respect to the installation. I took the time to deaden, built baffles, wired clean, etc. "Right" also includes an element of driver positioning, I know, but I elected to keep the system stealth and use nearly factory mounting. Woofers are in the doors, pointing directly out, right about knee level. Tweeters are in the sails about 24" directly up from the woofers, positioned more on-axis. Maybe I was too hopeful that good sound would be achievable in the stock locations with minimal adjustment.

EQ is where I thought I was going but wanted to identify anything obvious first.

Thanks for your thoughts. Maybe I'll end up seeing you at the meet at the end of August if it ends up happening.
 
#4 ·
You could fix most of your problems with careful use of the 3 bands of eq you have. I see 4 humps that need to be cut...just 3 if your system isn't sounding too sterile with that dip in the upper midbass/lower midrange area. I have 250 and 630 cut wide at -5 with my peq to take the bonky out of the midrange.
 
#5 ·
Thanks - this is the other part of the problem: a weak PEQ. Unfortunately, the Kenwood PEQ is rather limited to the following:
  • BASS: 40, 50, 60, 70, 80, 100, 120, 150; Q = 1.00, 1.25, 1.50, 2.00
  • MIDRANGE: 500, 1000, 1500, 2000; Q = 1.00, 2.00
  • TREBLE: 10k, 12.5k, 15k, 17.5k; no Q
And it's only one in each of the ranges.

Other than maybe the 500 peak and some of the 100 peak (which is already cut by about 3 dB or so), I really don't think I've got much that matches up.

I have brought up 1500 wide at one point, but then I end up with a hole around 200-300 that I can't fix.
 
#7 ·
Welcome to the wacky world of car audio! You have found out the dirty secret, that the installation techniques don't really have alot to do with what you end up with for sound alot of times.
The speaker locations have alot to do with the ragged response you see on the plots. Having the mids in the doors firing directly out puts the main focus of the speakers more at your knees. There are a multitude of things the sound probably refects off of before reaching your ears or the mic. That would explain the peaks and dips your getting in the midrange. Having the tweeter farther away from the mid can cause some comb filtering around the x-over frequencies where you would experience a peak or dip depending on the location in the car the mic is positioned.
If your getting reflections off the edge of the speaker opening in the door panel there is not much you can do about that without relocating the speaker or doing some modification to the panel somehow.
I would try changing the phase on each mid one at a time first and see if this helps. Listen to the sound after each change to see if it helps before committing to the change. Next try doing the same with the tweeters. This will be a much less noticeable change as our hearing is less sensitive to phase in the higher frequencies. After you find the settings that sound the best try changing the phase on both mids vs. both tweeters at the same time. Choose the setting that sounds the best to you.
Now after you go through all this the sub may not blend very well so you need to try changing the phase on it and use the better sounding of the two.
After all this if the sound and response curves are still not that great then try some eq. Try lowering the peaks instead of raising the dips. Our ears are less sensitive to dips in response than peaks. Since you have limited eq choose the worst offenders and tame them down. Keep in mind that after you work on the phase of the speakers that the response will be different than what you posted here.
This is just a brief touch on tuning and there are many more things you can get into to work on the sound with such as x-over points if you were running an active system, time alignment, etc. Just keep in mind that the usual method is to do everything you can to help the sound before using eq. Eq can't fix location, sound arrival times and reflections. One last note, the response curve you get with rta will not always be the best sound of the system. Your ears should always be the final judge. If the rta says its better but your ears say its worse then trust your ears over the rta. You listen to a system with music not with pink noise. An rta can never be as sophisticated as your ears.
I hope this helps a little.

Tim
 
#8 · (Edited)
Considering I find myself re-EQ'ing the system just about every change in song or recording, is it possible that these peaks and valleys are noticeable and I'm trying to correct for them to better match the recording?
This is not abnormal. Music is not mastered with any set standard or consistent methodology. Different tracks can have wildly different bass, highs and midrange. There is no single correct way to set a system to produce all music perfectly. I find myself always adjusting my EQ to balance out the sound to improve my enjoyment.

You're not alone in finding yourself adjusting EQ for fine tuning based on how each track sounds. I do this myself. I'm going to be buying the JBL MS-8 processor which has the ability to store 'presets' of custom user EQ curves. This should help reduce the time spent re-EQ'ing tracks.
 
#9 ·
If you have to adjust the eq for every song or album you're doing something wrong. I level everything out by ear with frequency centered pinknoise and this creates a nice universal curve that makes just about everything sound as good as it's gonna sound.
 
#10 ·
If there is a "universal curve" that makes everything sound good, I've never heard it. There are some songs that have wildly different bass or highs and even mids from other songs. There is no "universal" microphone, mixing board, mastering technique or other production rules that are followed by every artist. You aren't likely to find a "universal curve" that makes everything sound "correct" or even "good".

If you find that all your tracks have very similar sound, then something in your system isn't producing the sound correctly. Possible causes are crappy speakers, clipping/limiting pre-amp stage, clipping amp stage etc. When you don't have matching volumes, bass, mids or treble, many times you can force a similar sound by compressing/limiting/clipping the hell out of the sound. Another possibility is your speakers have a very static character to the sound where everything sounds the same no matter what is played.
 
#18 ·
I some questions forTspence.

When do you find time to EQ every track AND pay attention to your driving?

How long does it take to EQ each individual track?

What about songs that have 1 minute or longer super quiet intros? Do you EQ for the intro, then re-EQ for the rest of the song?

I am just curious as to how one goes about changing the equalization for each and every track they listen to.
 
#21 ·
Using an eq is a band aid approach to your sound quality problems. Location of speakers, the type of speakers and your source unit will have the biggest impact on SQ. I've tried using eq's with limited success. I could never get all the different genre of music I listen to to sound acceptable; consequently, I only use bass and treble to change the tone slightly and I've been a lot happier with my system.
 
#30 ·
Pulling your hair out unfortunately won't have much of an acoustic effect at those frequencies. :)

Seriously though, for trustworthy measurements you really need a spatial average of some sort... try moving the mic around an inch or two and correlate the measurements you get at each location. You'll notice many of the peaks/dips "adjust themselves" when you do this. It's amazing what a difference an inch or two can make...

It's also a good idea to listen to your problem frequencies (1/3 octave test tones are great for this) to hear them first hand and when you apply eq. Try moving your head around a little and see what happens while listening, too. You might be surprised to find some of your biggest RTA problems aren't actually audible, while others will truly have you pulling your hair out when you hear how bad they sound and how much the sound changes when you move your head just a fraction of an inch...
 
#47 ·
Thanks, I already shave my head anyway, so it would be interesting to pull my hair out. :)

The measurements were done with spatial averaging. I used to take six independent measurements and let TrueRTA average; now I just sit in the back seat with earplugs in and rotate the mic around the driver's head location about 3-4 inches in each direction.

I'll give a shot to listening to some tones around the problem frequencies. I seem to have trouble listening to relative differences using tones, but I can give it a shot--can't hurt.

Thanks again.
 
#33 ·
ZOMG Spence! I wish you could understand this.

There is a whole other world of SQ out there waiting for you if you would just let us help you.

I know exactly what your going through because I used to be the same way. I evolved from where you are now.

Soooo... what if the track has a high hat that you feel should be louder?
You boost those frequency up right? Well now the higher octaves of the female vocals may sound unbalanced. This is just an example.

You don't have the masters. You cant boost or cut instruments. An EQ is to get you speakers to sound good in your car. You shouldn't have to play with is much unless you change speakers or cars.
 
#35 ·
Yeah, some tracks are a trade-off where you won't get everything to sound correct. I'm not so much trying to increase individual instruments as much as I'm trying to get everything to sound as it should sound. Sometimes that's impossible due to the quality or acoustic variance between sounds in the mix.

Some tracks are supposed to be jacked around because it's intended. Sometimes I get lucky on shuffle mode and get two very similar quality tracks back to back and don't want to touch the EQ. :)
 
#40 ·
Yeah, some tracks are a trade-off where you won't get everything to sound correct. I'm not so much trying to increase individual instruments as much as I'm trying to get everything to sound as it should sound. Sometimes that's impossible due to the quality or acoustic variance between sounds in the mix.

Some tracks are supposed to be jacked around because it's intended. Sometimes I get lucky on shuffle mode and get two very similar quality tracks back to back and don't want to touch the EQ. :)
I'm curious as to how you know what it's supposed to sound like?


I use an RTA and it gives me a very good indication as to what's really going on.


I believe you're just chasing the potential of your setup going about things the way you are.
 
#39 ·
Yes. But only high quality ones. My computer speakers pretty much make everything sound close. My car system has so much clean power and dynamics that it's really easy for me to hear differences and want to fix things. I sometimes wish I didn't think I need to fix the sound. I admit I may have a bit of a compulsive problem there.
 
#44 ·
Dang, I let a day go by and I almost forget what I asked. Let me sort through the responses and see if I can pull some info and other questions out. Be right back...
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top