2021 10-speaker Bose CX-5 Frequency Response Curves

So, i brought out my laptop with REW and measured some frequency response curves using my UMIK-1. The microphone sat next to the driver's headrest. Response smoothing set to variable smooth.

First impressions: I noticed hardly any difference with the subwoofer connected or disconnected. I, also saw no difference between standard and linear... All measurements taken with the Centerpoint at 1, Audiopilot at 0.

Standard, Centerpoint=1, Audiopilot=0, Bass=0, Treble=0
Subwoofer Connected = Green Graph
Subwoofer Disconnected = Red Graph
1626714098819.png


I was able to get the most flat response by setting the treble all the way up and setting the bass to -3

Standard, Centerpoint=1, Audiopilot=0, Bass=-3, Treble=+9, Subwoofer connected
1626714255723.png



I was curious to see what crossover point the subwoofer is set at. I thought i would've seen something more dramatic when disconnecting, but that wasn't the case. From the graph... maybe it's set at 80 or 100Hz... but obviously is doing nothing in the 20 to 70Hz range.

I'm thinking of getting a Rockford Fosgate P300-10 powered subwoofer and tapping into the subwoofer speaker cable to feed the sub's high-level input. Ideally, it would be best to tap off the signal while it is still full range (that way i could try to get the suck out at around 200Hz)... but that would mean having to tear apart a 2 week old car... something I'm not ready to do. To power the sub, I can run the dc power under the car.

Combined Curves to see the effects of the Bass and Trouble settings.
Treble: Green=0, Blue=+3, Brown=+6, Purple/Blue=+9
Bass: Green=0, Blue=-3, Purple=-6
1626714716123.png


Last Graph: Standard vs. Linear. Both set to Centerpoint=1…. It appears Standard and Linear only apply if Centerpoint is off.
1626715059999.png


Comments and opinions welcome.

Matt
 
Last edited:
Excellent work!!!

And Mazda is improving the Blose system, if slowly. My '15 was +/- 14 dB from 40 Hz to about 14k(Can't find my graph), with shelf 3500 Hz and above and another from 40 to 200 Hz. I detest a smile shaped response curve.

My '16 was similar but the shelving was only about 6 dB. The bass hump can be eq'd out by setting the bass control to -6 dB. I'm getting good response down to 40 Hz and a very steep slope below that. To get the highs right, I changed the dash twiddlers and removed the center twiddler. Still need to replace the rear door speakers to kill the screech they have, too. I'm getting a pretty good response curve with -6 dB on the bass and 0 on the treble.
 
interesting graph. Thanks for that. By some perverse luck, I have three Bose systems at the moment. The one in my 2014 Buick is the worst, the one in my 2015 Chevy Colorado is better, and the one in my CX-5 is better yet. In all those Bose systems, I have to roll off the bass as it lacks definition and just muddies things. In the CX-5, I set the treble at +3 and the bass at -3, Center point =1, Audiopilot = 0. I also center the focus slightly rearward. That seems to provide the best imaging. With Centerpoint =0, you might as well be listening to mono since the imaging is terrible….possibly because of the center speaker (?). In any event, my hearing is so bad anymore, I can’t hear much above 8k so take my comments with a grain of salt. Oh, and of course your source material has a lot to do with sound quality. I use mp3’s sampled at the highest frequency…384kbs or something like that.
 
That's actually not horrific. Yes, you definitely need a real subwoofer. You've got to do that for sure. A DSP would do wonders as well. I use a Dayton Audio DSP-408 in my wife's van, as they are very inexpensive (only about $150!) and have 10 bands of full parametric eq per channel.

I would wire the Dayton to the inputs of the Bose amp (which I am almost certain is 4 channel, LF, RF, LR, RR), and then also use that to feed the subwoofer amplifier instead of using the speaker level output from the Bose.
 
The only way to get a *reasonably* representative car system measurement using an RTA, is to pan hard front-left, front-right, rear-left, rear-right and sum the measurements visually (save them and display concurrently) or in software. If you just fire up all speakers and put a mic somewhere you'll get the sum of all the speaker arrivals, but given the different distances from speaker to mic, they create a complex comb filter at the mic capsule, most dramatically shown in the severe +/- swings in the response below 1kHz,. The response isn't *really* that ragged. The *measured* and ear-confirmed response above 1KHz gets increasingly more accurate as the comb filter influence is very dense and tends to be averaged decently in software.

As for the sub, that steep roll off at 70 Hz is likely your source signal, or how it is getting into the head unit, or something in the mic signal chain, which is masking the sub from the measurement. In other words, if you can *hear* the sub coming and going but it doesn't show in your measurement, then the measurement is faulty. And if you can't hear the sub coming and going, then your source signal/path is faulty.
 
Last edited:
While I haven't measured my Bose system yet, your measurements above 1Khz seem to agree with my ears; a gradual HF rolloff, perceived as pleasant but lacking "air".

Did you make any measurements while cruising (say 60mph) to see if the head unit is making any volume or response tweaks to compensate for road and wind noise?
 
the measurement is faulty.
While that is certainly possible, I highly doubt it. MiniDSP UMIK-1, while certainly not up the standards of something like a Bruel and Kjaer, is actually a pretty damned good piece of kit for what tiny bit of cash it costs. Each individual mic comes with its own calibration file that REW uses to make sure that a baseline measurement is indeed flat.

You seem to have a good general knowledge of what's going on here, I want to encourage you to pick one up and play around with it. I think you'll see it really is pretty good.

As far as measuring sweeps, how I do mine is, only play one side at a time. For the left side, I place the mic at my left ear, and do three replications. Then I play one replication at my right ear. Sum those 4, and there's my left side graph.

Rinse and repeat on the right: 3 at the right ear, 1 at the left, sum.
 
but given the different distances from speaker to mic, they create a complex comb filter at the mic capsule
That's why before you even start your first electrical measurement, you break out the measuring tape and set your signal delay. And no, it's NOT time alignment, it's signal delay. "Time Alignment", as described by the typical car audio enthusiast, literally doesn't exist. Time is the 4th dimension. Unless you have the mass of a Star or Black Hole, you cannot align time. You cannot change time. You cannot shift time. Time is constant. Time just is. What you can do is delay signals so that the sound from each speaker arrives at a given point in space simultaneously. This is easily done with a simple tape measure.

Once you have the signal delay dialed in, then and only then do you take your measurements.
 
Yes, I am a design professional in electroacoustics. The root issue is that there is multiple arrivals at the mic for drivers covering the same part of the spectrum (e.g. two mid-low drivers arriving at different times). This is compounded for vehicle cabins which have relativity small dimensions, where the flight times of drivers operating in the same spectrum are close enough, and thus the FFT can't be windowed narrow enough to remove the additional speaker sources, without impacting FFT resolution.

What caught my eye was the huge suckout at about 180Hz; unless defective, this isn't (probably) the actual response of the speaker, nor the perceived response of a human ear, but 99% of the time an artifact of measurement; mic location, FFT windowing, using a time-ignorant RTA mode.

The sag at 700Hz or so is a suspect artifact. You could close-mic measure the door and pillar drivers to determine if the driver really is that ragged, or the suck out is an artifact of mic location and/or FFT "hearing" and summing multiple driver arrivals.

If you move your mic 6" one way or the other and the response radically changes, then the measurement where the dramatic changes occur is FALSE. Your measurement is lying in that part of the spectrum.

I wasn't dissing your measurement equipment, but rather suggesting *something* in the source or measurement chain is causing the steep LF rolloff, where your ears hear the sub is present, but the measurement doesn't show it. Thus the measurement is is false in that part of the spectrum. Can be a measurement software setting, a filter mic on a mic or mic interface, a pink noise source that was accidentally processed with a LF cutoff, etc.

Last; try my suggested measurement approach to minimize the number of drivers operating in a given spectrum; e.g. just one mid-LF driver and one HF driver per measurement. OVERLAY those curves; wherever on the curves they closely follow each other, THAT is likely the real response, and BTW, the ONLY part of the spectrum you should attempt to EQ.
 
Last edited:
I’ll be making additional measurements this weekend as suggested. For comparison, I’ll also test my other car that has an aftermarket system to verify REW’s sinusoid sweep function.

Thanks,

Matt
 
And no, it's NOT time alignment, it's signal delay. "Time Alignment",

That's a colloquial phrase meaning the adjustment of delay amongst a group of sources, in reference to the closest source in order to achieve the same (in phase) acoustic time arrival at a particular location. It is properly called time alignment (of sources).
 
It is properly called time alignment (of sources).
Nope, it's not proper at all, and I will argue that point until my dying breath.

Another car audio term that is completely wrong is "subsonic filter". Nope. Incorrect. It's an infrasonic filter.

Subsonic means slower than the speed of sound. As in, you can buy subsonic rounds for your pistol, so that the bullet is traveling at less than the speed of sound, and therefore the report is diminished. Or, a fighter jet can travel with supersonic speed (faster than the speed of sound) or subsonic speed (slower than the speed of sound).

winchester_subsonic_22lr_42grn_ammunition_1.jpg

The correct terminology is infrasonic. Infrasonic is a frequency lower than what normally can be detected. On the opposite side, ultrasonic is a frequency higher that what normally can be detected.
 
Also, I'm super glad you brought up phase cancellation by multiple drivers playing the same frequency. Because of phase interactions, that's why I'll never do that.

But, a cool story: back in the 90's, before DSP really became a thing, Precision Power came out with a component called the PSC-211. It was an analog phase shifter that was continuously variable from 0-360 degrees.

There was a lady named Alma Gates who was going for the World Record in SPL in her Ford Bronco. She was using PPI components, and they sent over a case of their PSC-221's. They hooked them up, and dialed in each individual speaker to give exactly the same phase response at 40Hz where the measurement mic was placed. Getting those speakers in phase was the sole reason she was able to both set the world record in SPL!

That piece was so easy to use. Just turn two knobs while you're watching the SPL Meter. When the numbers increase, you're getting more and more in phase. When they go back down, back off, and move to the next. Easy piezy.

DSP's do this for us now.

104_0182.jpg
 
This is way too complex for me to understand, but I do appreciate the conversation. I am surprised the sub seems to make no difference and that there is a huge "hole" in the 100-200 Hz range
 
This is way too complex for me to understand, but I do appreciate the conversation. I am surprised the sub seems to make no difference and that there is a huge "hole" in the 100-200 Hz range
Agreed. But I still enjoyed reading the discussion!

My guess about the useless on factory Bose sub-woofer is that its 5” speaker / driver is simply too small for a woofer.
 
Agreed. But I still enjoyed reading the discussion!

My guess about the useless on factory Bose sub-woofer is that its 5” speaker / driver is simply too small for a woofer.
Correct. A 5" driver simply doesn't have the ability to have a very low resonant frequency (Fs). Very generally speaking, a given driver can only play down to the Fs. So, a 12" driver with an Fs of 28Hz can play down to 28Hz. Your average 5" driver, even if it advertised as a "subwoofer", will have an Fs in the 80-100 range.

Again, this is way, WAY over generalizing the way a driver behaves, and there are a multitude of ways to get even a small driver to play low, like transmission lines or whatnot. But that's far outside the scope of this discussion.
 
There is something definitely wrong with my measurements and the work Dell laptop I was using... i measured my home system which has a very good subwoofer (SVS SB-2000 Pro) and I saw a similar result as I did in the car... no low end frequencies. so I brought out my old Toshiba laptop... see the difference below:

Dell laptop (I tried both Bluetooth and a direct analog connection.. same results as below)
Capture.JPG


Toshiba laptop
Capture2.PNG


Now, the Toshiba does not have Bluetooth so the above measurements were both using an analog connection between the laptop and my BlueSound Node 2i streaming music player.

Since the 2021 Mazda does not have an AUX input, I'll have to buy a Bluetooth device in order to get my Toshiba laptop to take measurements.... or I could try a pink noise source which I might try later on.
 
Avoid Bluetooth for anything that you actually care about sounding good. It's even worse than mp3, if you can believe that. And mp3 via BT? OOF. Please, just no.

Focal has a free test CD on their website that you can download, and it includes both pink noise and sweeps. If you car doesn't have a CD player, just turn it into a flac or WAV file and run it that way via USB.
 
Back