DIYMobileAudio.com Car Stereo Forum banner

Target Curve Comparison

119K views 95 replies 35 participants last post by  Sounds_Insane 
#1 · (Edited)
This thread may spur some discussion, or it may not. I often see people ask about what final "curve" they should tune their system to. You see responses range from things like: flat, "smiley face", ELC (equal-loudness contour), gentle slope downward, the "JBL Curve" (provided by Andy W.), etc. It just so happened I decided to do a bit of research on the topic while I had some down-time recently and wanted to share my findings. Everything below was obtained from simply using Google Search.

So, let's start with the curves written out. I ended up settling on five different curves. Any explination/logic I found about why these curves exist will be shared below said curve.

1.) JBL/Andy W. Curve:

-20-60hz: +9db
-60-160hz: Transition to 0
160-3khz: Flat (0)
3khz+: Gradual Roll-off to -6db @ 20khz

* Andy has stated this target response is the "ideal response for a small listening environment" with a bump on the low-end to compensate for typical listener preference.

http://www.diymobileaudio.com/forum...main-andy%20wehmeyers%20target%20response.jpg

2.) Crutchfield Labs Curve:

-20-125hz: +6db
-125-200hz: Transition to 0
200-12.5k**: Flat (0)
12.5-16khz: Roll-off to -3db
16-20khz: Roll-off -6db (-9db total)

**This range includes a +1db bump at 400hz, a -1db dib at 10khz, and a +1db bump at 12.5khz.

* This target curve was in a recent edition of Crutchfield Labs. Here is the only comment I found to the logic behind it, "This represent's Jeff's ideal sound curve, one that helps overcome some of the limitations that appear when you listen to music in a moving vehicle."

Crutchfield Car Stereo Proving Ground

3.) Audyssey Curve:

20-1khz: Flat (0)
1-3khz: -3db dip
3-10khz: Roll-off to -1.5db
10-20khz: Roll-off -3db (-4.5db total)

* This "popular" curve is the closest I found to flat that some people recommend. I also thought the logic behind the curve was interesting. The roll-off on the top-end is performed as a gated measurement. This eliminates reflections being added to the measured response. Ungated response will be "whatever it is" after reflections are accounted for, but it should be close to flat. This leaves the small dip. It was explained that because of the popularity for 8"/1" and 6.5"/1" speaker combos (both in home and car), we have become accustomed to and prefer the sound of a directivity mismatch, and the dip in power response, and therefore Audyssey "tunes it in" even if it isn't needed. They also flatten the response of the sub up to around 3khz, add the crossover at 80hz, and allow the end-user to adjust sub output if they perfer a boost on the bottom end (similar to all the other graphs)

4.) Audio Control Curve

-20-40hz: +6db
-40-250hz: Transition to +1db
-250-2khz: Transition to 0db
-2-20khz: Roll-off to -5db

* It seems this curve was included in the manual with the old Audio Control RTA's. The manual basically said, "Flatten response as much as possible (you won't like it), and then adjust to something approximating this curve." I found this information on diyaudio.com I believe if you want to dig further.

5.) B&K Curve

-20-160hz: +3db
-160-2khz: Transition to 0db
-2-20khz: Roll-off to -3db

* There is a long PDF about this curve which I've attached. It seems to be well-received in mimimum phase systems (which the car ISN'T), but I wanted to include it because it also represents the people who like a smooth and steady roll-off.

http://www.bksv.com/doc/17-197.pdf

And for those of you who are visual learners...a graphical representation of all the curves and an average of all the curves:

 
See less See more
1
  • Like
Reactions: Truthunter
#67 ·
I'd like to add one to the mix and revive this thread. I'll call it the "bose" curve. I purchased some Bose computer speakers after being astonished at their pleasing SQ after an in-store demo. I measured with my RTA (magenta curve) and mirrored the response in my car with a little more bass (blue curve).
Although I tweaked the bass a bit it has remained as the basis for my favorite sounding curve. It even has a little fletcher-munson built in- notice the valley between 2.5 and 6K?
Also shown is my version of the Andy curve (gold). No offense but this one simply did not last a day. It was too hot and bright. Maybe I need to taper the high end more?
Good stuff these RTA curves.
Reference, uncorrelated pink noise, 6 position mic average at the listening position, body in car.
 

Attachments

#68 ·
I'd like to add one to the mix and revive this thread. I'll call it the "bose" curve. I purchased some Bose computer speakers after being astonished at their pleasing SQ after an in-store demo. I measured with my RTA (magenta curve) and mirrored the response in my car with a little more bass (blue curve).
Although I tweaked the bass a bit it has remained as the basis for my favorite sounding curve. It even has a little fletcher-munson built in- notice the valley between 2.5 and 6K?
Also shown is my version of the Andy curve (gold). No offense but this one simply did not last a day. It was too hot and bright. Maybe I need to taper the high end more?
Good stuff these RTA curves.
Reference, uncorrelated pink noise, 6 position mic average at the listening position, body in car.
Interesting. Ya still gots that big dip from ~60-~80 Hz tho. You really should try out some nice 8-9 inch midbasses in that Rav;)
 
#71 ·
Refined curves shown in the attached pics. the "andy curve" was too bright on the mid-high end. the "bose curve" did not have enough bass given the noise and other cancellations on the road.
My final "hybrid" curve smooths out the bass end by dropping from 50 to 160, not the previous slope from 100 to 200 hz- i.e. a misrepresented andy curve- and merges with the bose curve from there up the register.
I also dropped the boosting below 50 hz in used to flatten out the deep bottom end. It doesn't seem right to boost the low end.
Regardless, the new refined hybrid curve sounds radically different than any curve I've run to date.
A nice solid bass foundation with an "oomph peak" and zero coloration / muddiness to the overall sound, which is clear and open.
 

Attachments

#73 ·
The target curve I published depends a little bit on the mic, but only the high frequency tilt. That's really a matter of preference. MS-8 uses a spatial average and a microphone that goes on your head. High frequency tilt in the target for that technique MUST be greater than for a single mic.

Spectral averaging at high frequencies is almost as effective as the spatial average. In the midrange, it's helpful to look at several mic placements.
 
#74 ·
the push towards quieting the vehicle is also the push towards retaining some mid bass coherency, as the 250 hz rumble through the chassis from the tires is going to affect your perceptual "suspension of disbelief" that augments the stereo image.

Notice when you are apt to boost the bass/mid bass at speed, it's usually because you notice that something's not right with the music, and that 'not right' feeling is the dissolution of clarity in the band of frequencies that we are attuned, for survival.

the stereo image collapses the most significantly when those frequencies are affected, I would imagine it's part of the hearing mechanism's designed in traits, that probably also produce a Fletcher-Munson curve's biased or non-flat bearing.

and that's probably why OpSoDis is successful as well, in that there is a greater panorama between speakers that lie 180 degrees to our sides, than at any other point forward in the image.

so bringing all of these keywords and catch phrases together, I propose that it's okay to boost the mid bass when riding down the highway.

Or, quiet the car, that's pretty good too.
 
#75 ·
I agree 100% that mitigating road noise helps lesson need/desire to have different setting for driving and stationary. My stereo isn't perfect but I have pretty well tamed road noise to the point that at my listening levels (which ARE a bit loud), I don't feel the need to have two different settings.

And since I have learned a good bit since starting this thread a year and a half ago, I'd like to update it a bit from my end. First, my personal tune is quite similar to the JBL tune, except the rise starts around 215hz (instead of 160hz) and carries higher (up about 18db a 35hz). What I ultimately ended up tuning for was "ear flat". What this means is that using by using 31 bands of pink noise, I adjusted levels until things were perceived as flat (I interlaced a 1khz tone between each band for "level reference"). Let me also state that this IS NOT easy to achieve. In the three years of building this car what I ended up with was a 12" sub tuned to 18hz (crossed at 60hz), 10" midbass in each footwell (crossed at 60hz and 180hz), 6.5" midranges (crossed at 180hz and 2.5khz), and tweeters covering the remainder. How does that compare to "the norm". Well, I cross a 10" midbass where I see many cross a 6.5" (63hz) and I cross a 6.5" midrange where I see some cross a 3" (180hz). Sub and tweeter are pretty typical. My reason for this is because I feel speakers being played outside their ideal range kills realism more than the system response and getting that increase from 215hz down to 35hz is critical.

Look at the response of most 6.5" speakers and they are rolling off with no filter by 80hz or so. So the speaker's response is going down as your response curve should be going up! I worked around this by crossing high. Even though I "cross" at 180hz between midbass and midrange (the drivers responses cross at this point), I actually cross the midrange around a "typical" 80hz. This is because combining the flatness of the midrange with the rolled off midbass creates a rise in response (just like we want for the increase in the curve). I do the same with the midbass to sub. Basically, I overlap but only with the response of the smaller speaker. It still makes integration tricky, but it keeps subs from playing too high and keeps midbass from playing too high as well.

If you really want to get a good idea of what I'm talking about you need to play around with REW. Set a simple house curve that is +20 at 20hz and -20 at 20khz (you can hone this in later). Now load it up. You can set the speaker to "small speaker" and set the crossover at 80hz. This will show you what you NEED a speaker to do to handle an 80hz crossover with "ear flat" tuning. It is not easy.

If any of this was unclear, I apologize as I quickly typed it on my lunch break. If you need any clarity or help, I'll be glad to do what I can. Also, my experience is not all inclusive and does not mean you can't achieve a great sounding system with a 6.5" midbass or 3" midrange or whatever...I'm just sharing and I hope it helps.
 
  • Like
Reactions: avtexan
#76 ·
What I ultimately ended up tuning for was "ear flat". What this means is that using by using 31 bands of pink noise, I adjusted levels until things were perceived as flat (I interlaced a 1khz tone between each band for "level reference").
Very cool technique. I've often used pink noise for similar purposes and never thought to throw a reference tone in to help overcome the tricks that our minds play with respect to sound. Thanks for the idea!
 
#77 ·
Interesting thread, nice discussion ;)

From lots of experimenting with tuning in different cars I found that the low/high frequency tilt is dependent on cabin size, absorbing of highs and directionality of the drivers. With small widebanders practically beaming above 5kHz I found that they need less tilt in the highs to sound balanced. With small tweeters with more dispersion higher up, I needed to decrease the highs slightly to get it to sound balanced. It would seem like the amount of reflections would alter the perception how loud certain frequencies are. Since limited dispersion is hard to attain lower down in frequency, this would apply to the upper midrange and highs. I found that a continuous rise of the frequency below 100Hz sounds best, 20Hz should be about 10dB higher than 100Hz and then pretty flat between 200-250 to about 1kHz and after that downwards tilt towards 20kHz.

Some people mention they tune after the Equal-Loudness-Curves. Imo, that's not a good way of doing it. It won't sound correct from 5kHz and up for starters and often 2-4kHz end up too recessed. To perceive the lowest octave equal at all volume would require some kind of volume dependent bass control (this is the range that our ears sensitivity varies the most in vs volume, i.e "phon" vs frequency relationship.
 
#81 ·
That sounds exactly right.

I've noticed if the HF is closer like on dash and midbass in doors or kicks the highs need be tilted a lot more. Seems the closer they are the more of a shelf is needed. A rta or sweep may show flat but sounds too bright
 
#78 ·
Good point on level. I personally use 85-90db for my "ear flat" tuning because I feel it nets the best compromise for competition.

Previously I only used my miniDSP for tuning which meant I had to break out a laptop to change presets. I now have a P99 as my source so I plan to use the 31-band presets to set up minor changes for higher and lower volumes...eventually. :)
 
#79 ·
Yes, I know this is an oldie but goodie thread. There seems to be more and more talk about processing and EQ lately and figured target curves would be relevant information for some of those getting in to tuning.

I also wonder if some people's views on these curves have changed over time?
 
#80 ·
I'm finding, as a newb, I'm having better results leaving a 'curve' idea as for the end, but tuning to general good tonality and balance first, leaving the curve for last. So while I have some idea of the curve I'm looking for, it's only after knocking down individual peaks, bringing sides into balance, centering up the image throughout the bandwidths, and only then being overly concerned about conformity to a curve. If done well, a good balanced curve for the individual car may present itself to you.. Trust ears, and verify by graph.
 
#84 ·
Sounds possible. Could also be boosting frequencies around a very narrow dip that doesn't show because of the measurement technique or settings. However, that's derailing the topic just like happened originally. Has anyone else had success with variations of these or combinations of these?
 
#86 ·
Seeing the different target curves just comes to show there really isn't no wrong type of curve. It's all listener preference like the pre chosen jazz, rock, live presets. One guy will like one and hate the other.

I know this is a older thread but the info is great and should help some guys out. You don't have to have a true flat curve going end to end, it just needs to be smoothed to the listeners preference and the setup that ends up sounding the best. Looking at some of those popular curves I can tell I would hate some of them. I do want to see more graphs of other people's target curves to see the differences
 
#88 ·
You can use the tool in my signature to export a JBL curve or one of a few others into a text file for use with REW. You can also use my tool for a few other neat tricks too.
 
#89 · (Edited)
The target curve is not only vehicle dependent but also speaker location dependent. I remember tuning a car that had Focal tweets and 6-1/2's in the kicks. They were aimed up at the center of the car, so mostly on axis. That car sounded wonderful flat between 100 and 10k with a 12 dB rise below 100 and a 12 dB fall above 10k. That car eventually had the kicks changed so that everything was off axis and that same curve didn't work anymore. I could never get the sound quality back to where it was even though the rta looked good. The stage was lower and fell on the sides. It's tough to say that any of the above curves would work well but from experience I can say a few things which should make finding the right curve easier. In the 400 hz and below range, aim doesn't matter. Only path length matters. The closer to equal the better. On axis is not as important. From 400-4000, pathlength and aim matter. On axis is better than off. Above 4000, pathlength doesn't matter. Only aim does. On axis being highly preferred. If you can abide by those rules, those general curves will work.
 
#90 ·
Novice question:

So if I set my gains for a 3 way active system with all EQ settings and levels flat/0db for each network, and using a -5db test tone @ 4khz for tweets, 400hz or 1khz for mids & 40hz for the sub, then start to dial in (for example) the JBL target curve with a +9db between 20-60hz, isn't that much of a boost going to cause a clipped signal?
 
#91 ·
Novice question:



So if I set my gains for a 3 way active system with all EQ settings and levels flat/0db for each network, and using a -5db test tone @ 4khz for tweets, 400hz or 1khz for mids & 40hz for the sub, then start to dial in (for example) the JBL target curve with a +9db between 20-60hz, isn't that much of a boost going to cause a clipped signal?
Maybe. Just subtract 9 db from everything, or set your gains with a 0 db test tone, then subtract 4 db from everything. (Everything in the target curve)
 
Top