DIYMobileAudio.com Car Stereo Forum banner

A Compartive Study of Loudspeaker Frequency Response in Car Doors

25K views 66 replies 28 participants last post by  Bacovish 
#1 ·
All,
In the spirit of sharing, I'm posting my findings as I dive into the world of in-car measurement w/ a laptop & microphone data acquisition system. I hope others can benefit as I learn the ins and outs of using this tool, and I continue to move forward in that never ending quest for better sound. :D

Background/Purpose
  1. I've always wanted to better understand my car's acoustics.
  2. I already had a laptop, so buying a mic and setting up a measurement system was both inexpensive and easy.
  3. What measurable differences in frequency response will we actually see when speakers (raw drivers) are put in the "harsh" automotive environment?
  4. Does a "car-audio" specific driver really work that much better than a "home" driver in a car door?
  5. What impact does sound deadening and install effort have on the frequency response of a car door loudspeaker?
In the course of this thread, I hope to explore the answer to those questions.

Setup/Procedure
Figure 1: Measurement Setup

  • Acer "Ferrari" Laptop :D
  • M-Audio MobilePRE USB
  • Behringer ECM8000 Measurement Microphone
  • Associated connection cables
  • ARTA Software
  • My 1999 Honda Accord LX 4 Dr., Phoenix Gold XS6600 Amplifier "rear" channels driving the door speakers
  • Seas W15LY-001 loudspeakers
  • Seas Lotus Reference RW165 loudspeakers

Figure 2: Microphone Placement (those are shadows, my car isn't that dirty! :p)


  1. Hardware was set up per manufacturer's recommendations.
  2. Microphone placement per Figure 2.
  3. Sound output was input directly into the car amplifier, bypassing the processor completely. Only one channel measured at a time.
  4. Impulse response measurement (MLS), 1/3 octave smoothing.
  5. First round of measurements were done with two pairs of speakers: Seas Excel W15LY-001 and Seas Lotus Reference RW165. No modification or additions to the car door accept for the attachment of MDF mounting/spacer rings.

Measurement Data
Figure 3: Left Door, W15LY-001


Figure 4: Left Door, RW165


Figure 5: Right Door, W15LY-001


Figure 6: Right Door, RW165


Observations
  • Cone breakup is readily seen (just under 5 KHz) in the RW165 on figures 4 and 6.
  • No surprise: left and right side response are quite different, note the poor bass response on left side.
  • As expected, the magnitude of the swings in frequency response are quite astounding over the speaker's usable range when placed in a car door.

Discussion/Analysis
I exported the data from ARTA and graphed/overlaid the W15 with the RW165 so that a more direct (in door) driver-to-driver comparison could be made. The results are shown below.

Figure 7: Comparison of the the W15LY-001 to RW165, Left Door


Figure 8: Comparison of the the W15LY-001 to RW165, Right Door


Key Takeaways: (however obvious they may be)
  1. The vehicle's transfer function (environment) has far more impact on frequency response than any other variable. Differences between drivers are relatively small in comparison.
  2. Differences in driver performance (frequency response) are quite measurable. These drivers to me sound different, particularly in the midrange, and the differences in response from ~1KHz and up correlate well with my subjective evaluations.
  3. I was surprised that the lower frequency response of the two drivers didn't differ more, but there does seem to be a small measurable difference nonetheless. It's not much though, and error in accurate gain setting and in measurement repeatability renders that point completely moot.
  4. If smooth frequency response is your goal, there's no way of getting there without having separate left/right EQ, AND alot of control via parametric or NUMEROUS graphic bands.

Next Steps:
  1. Deadening/sealing the doors to see how that impacts speaker frequency response. I myself have experienced enormous subjective differences with installs in the past, and I'd now like to see how that measures.
  2. Testing ensolite and/or other treatments behind the speaker to see how they impact frequency response. (How effective they are at killing "midrange echo"?)
  3. If I'm ambitious, I may try to measure differences using clay as a door "decoupler", etc... We'll see. :D
  4. If there's something interesting you want me to look at, throw your ideas my way. :D

Enjoy!
 
See less See more
8
#2 ·
I think if one was to search for it, they will find a tutorial that Ngyen did a while back on in-car response measuring technique. He also posted the results of my 3way setup. I think the plot showed all 3 plus sub in different colors on the same plot and then a summed curve. It was ugly.:(
 
#4 ·
Great study, Great info. Thanks for the forethought and execution. Things like this will def. change the way one thinks of car audio.
 
#5 ·
I don't know if you tried this or not, but check to see if placing the microphone on the passenger side reverses your L/R results.

It would also be interesting to compare the off-axis response of the driver outside of your car with the response in the car to see how much is the door and how much is just the driver.
 
#6 ·
I don't know if you tried this or not, but check to see if placing the microphone on the passenger side reverses your L/R results.
That was something I was going to do, but didn't have time. I'll check things out before I deaden the doors.

It would also be interesting to compare the off-axis response of the driver outside of your car with the response in the car to see how much is the door and how much is just the driver.
This isn't something I'm interested in pursuing. It is more difficult to set up a test like this, and much of the data provided by the manufacturers already covers this, but I do see where you're coming from...

I also thought about taking some measurements with the door open (to just test the speaker "enclosure" and not the "listening room"), but didn't carry through with that either, because in the end, that situation doesn't reflect any kind of reality.

Thanks for the feedback, Chris. I'll report more when I get the chance to take more measurements. :D
 
#7 ·
The open door idea sounds good...and place the mic about 1 meter away outside the car. That should give you a readout of the enclosure as if it were in a vacuum...well almost.

That is just such a huge difference in output below 100Hz between L and R it makes me wonder if that is what you hear when you listen.

Perhaps the seal around the driver wasn't as good on the left?
 
#13 · (Edited)
Next meet, we can definitely do that. Better bring me some beer or something though. :D ;)

Can't wait to get my MobilePre fired up to see how big of a mess mine is too! :(
One thing I am considering is buying a (better) calibrated mic. The ECM seems to work fine, but it doesn't have a calibration curve, so there's some (unknown) absolute measurement error. For this study, it's no problem because I'm looking more for relative results. But if I want to be more serious about this in the future, a better mic is probably in order. Just FYI.

I like the MobilePre, BTW, really nice unit. I got mine off ebay for about $75 shipped, IIRC. Cool unit, and quite flexible with all of its inputs/outputs.
 
#14 ·
Powerful evidence for the need to get mids and highs away from the lower doors!

One question, though: how is the driver mounted in the door? Is it firing straight out, or is it mounted on a metal panel with the outer door panel between? I ask because the 1kHz dip could be some sort of cavity issue. I've found something similar (except ~an octave up) in the NB Miata.

I ask because I've found a Miata's.
 
#16 ·
Wow, big up man! Thanks a bunch for putting together this study. I look forward to the results and your observations.
 
#17 ·
Dumb question but I think a reasonable request considering it might be realistic-how about taking a couple measurements with the windows just to demostrate what happens there. Some of us like to drive with the windows down.:blush: :D
 
#18 ·
Dumb question but I think a reasonable request considering it might be realistic-how about taking a couple measurements with the windows down just to demonstrate what happens there. Some of us like to drive with the windows down.:blush: :D
 
#19 ·
Great info John, this was the fist post that ate up my "break-time" at work ealier today. I look forward for more new and revealing tests, albeit this doesn't so much pertain to installs of my nature, it's still very interesting. Glad I bought a DCX-730, wooooo! Individual Channel EQing! :D
-Fixtion
 
#27 · (Edited)
Great idea to compare the two drivers!!

I used ARTA (with a ECM and MobilePre) for some measurements as well, and the only problem was the inability to save data with trial version.

I was undecided on whether the drivers or car were responsible for the frequency response i was getting, but your results make it pretty clear.

In my tests, there were variations in the midrange depending on microphone placement (+/- 2 inches made a huge difference). The midbass and high end seemed pretty stable.

I was going to try moving the microphone around like npdang had done to average the "left and right ear" response, but I hadn't figured out a good way to do that with a free program yet. I tried moving it between passes, but i'm not sure the results are accurate that way.

I would be interested to know if you observed the same thing with microphone placement considering that you are using the exact same setup i tried.
 
#39 ·
Haha, sounds like me and my current install. There gear is coming in, yet the install is at a stand-still. I mainly wanted other users to check out your findings :] I'd be interested in finding out the thickness of f/b required to simulate a type "x" wood enclosure and compare the various "matting" fabrics for better responses. Seeing as how many of diy go the f/b route or mdf baffle door installs. Density consistency is a big factor in choosing an enclosure. So, I'll stop rambling and await new results. :D
-Fixtion
 
#40 ·
OK, I'm going to attempt to "catch up" a bit on what I've been up to. :D

First, here are measurements I took varying microphone position across the vehicle.

Setup
  1. doors not yet deadened in these measurements
  2. Seas Lotus RW165s
  3. 5 mic. positions, about head high, approximating:
    • driver's left ear
    • driver's right ear
    • center of vehicle
    • passenger's left ear
    • passenger's right ear
  4. everything else as before

Results
Left Speaker Measurements:


Right Speaker Measurements:


Note that, as expected, microphone position dramatically impacts measured response.

Here's what interesting: the car does, for the most part, act quite symmetrically. Below, I've plotted symmetric measurement positions across the vehicle, and you can see that responses are very similar. Of course a car isn't perfectly symmetric, but my car isn't too bad in this regard.





Looking at the responses more carefully, there's lots of interesting things to note. A couple of comments:
  • bass response is worse as you get closer to the speaker
  • cone breakup for this speaker is MUCH more noticeable the more on axis you are.
  • even frequency response in two seats is not an easy accomplishment. :p

much more soon to come....
 
#43 ·
Here's my attempt to "eliminate the room" by measuring door speaker response with the door wide open. Here's what my setup looked like:



I plotted response against right door speaker response measured at driver's ear inside the vehicle. One thing to note is that I significantly increased drive level with the door open so as to reduce the impact of outside ambient noise (I live near a freeway). Keep that in mind when looking at the next plot.



If this isn't proof of Nguyen's quote: "A speaker is only as good as the room you put it in." then I don't know what to say. :D

It's very easy to see how much smoother the response is without the harsh room (car) environment. It is also very obvious that cabin gain has a HUGE impact on bass response. Bear in mind that the door in these measurements has yet to be deadened, so you wouldn't expect the response to be super smooth; the enclosure still sucks rocks.

More coming....
 
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