2014-2018* Mazda 3 & Mazda 6 w/BOSE Full System Breakdown/Analysis
This is a Mazda 3 focused version of my Mazda 6 BOSE analysis guide. It has been adapted from the original so that color wiring and explanations match exactly how things are in the 2014-2018 Mazda 3. The general progression of how I discovered things is still presented from the perspective of myself exploring my Mazda 6, but this information fully applies to the 3 and as I said I have made the necessary changes to this version after studying the 2014 Mazda Workshop manual. If you want the Mazda 6 version of the guide see here: 2014-2018* Mazda 6 (and 2014-2016 Mazda 3) w/BOSE Full System Breakdown/Analysis
*Applicable Models:
This is a formal documentation of my findings while investigating the BOSE system on my own 2016 Touring 6 out of frustration over the lack of information online that I could be certain applied to the 2016 model.
I simply want to replace the entirety of the audio system in my car while maintaining the factory head unit since I couldn't find any of the ones specifically made for this car that I was a big fan of, and using any of the generic aftermarket units would mean losing tons of functionality (maintenance status, lock preferences, commander knob, etc.). Supposedly, the 3rd gen 6s with the BOSE system supply straight, line-level outputs from their head unit to the BOSE amp that you can simply tap into directly for the connections to your aftermarket amp; however, after countless days of searching, all I could manage to find were similar anecdotal statements of this fact and poorly detailed diagrams with no clear indication of what vehicle they were in reference to. I was able to find some information that was a bit more definitive, but unfortunately it was for the CX-5, and I didn't want to just assume it would be the same in my 2016 6. Even asking others on this forum for more concrete info on where exactly I could tap this signal from lead to dead ends.
On and off I considered looking into the system myself, but with no real starting point, I didn't want to dedicate too much time to something that could prove to be fruitless and that initially I wasn't completely committed to yet. Luckily, I eventually struck gold by finding one single thread on this very forum by someone else that also happened to have a 2016 6 with the BOSE upgrade who shared some details about how he added a sub to his car while utilizing the "original signal" from the head unit. Annoyingly, he was pretty vague with his descriptions, but there was just enough substance to give me a starting point for my own investigation.
I know some people have tapped into the rear speakers that are supposedly full-range (though possibly not quite) and used a hi-low level converter to add an aftermarket amp, but I too would prefer to use the original signal if it truly is as accessible as others have made it sound. So without further ado, the following is a recount of what I ended up doing and what I ended up figuring out.
This post has a companion video on YouTube. The video is more focused on the process and explanations of my investigation, while this post is more focused on the results and details.
Video:
If you only want to see part of the video there are time stamps in the description if you head over to the actual YouTube site. The video is fairly long at about 35 minutes, but was made so that you don't have to just take my word on how the system is setup and can get a better sense of the process required to determine my findings if you need/want to.
Documentation:
Other than the singular post I mentioned, my only reference point was the 2014 Mazda 6 Workshop manual (and in this case the 2014 Mazda 3 Workshop manual) that you can find pretty easily with Google. I won't be posting a link to it, but will be using various images from select sections of the Electrical Diagrams segment. As far as I know, there doesn't exist a newer workshop manual past the 2014 version (at least not that is publicly available), which initially was a large part of my troubles. You can find places to buy the manual that list it as 2014-2016 or 2014-2017, etc, but it's just the 2014 version and has no updates whatsoever. For a large portion of the car this is a non-issue as most of it has remained the same over the years, but there have been some significant changes from 2014 till now (only true for the 6) that obviously are not covered in this dated document and therefore are a major frustration point. The headlights received extra chrome trim in the 2017 model, the 2018 model had an entire front and rear face lift, and most importantly for this topic, the head unit changed starting with the 2016 model. Because of this, there are portions of the electrical diagrams in this workshop manual that are no longer correct (again, all of this is only true for the 6); however, after having inspected my 2016 myself, I can safely say that enough of it still applies (primarily in terms of the connections going to and from the BOSE amp) that every connection you need to hook an aftermarket amp into the BOSE systems "original signal" is still accurately portrayed for models past the 2014.
Starting Point:
There is a fair amount of raw data to cover here, so I will be very brief in my explanations of why I did something or how a result proves the conclusions I made. Again, if you want a bit more detail into my process, watch the above video. If you are confused or want more info about a particular part, just quote me (though please snip this huge post out lol).
OK, so here we go. This is the thread I referred to earlier that got me started: 2016 headunit location
The gist of it is that the OP discovered/confirmed that in the 2016+ models (2014+ for the Mazda 3) the BOSE amp is under the passenger seat and that in one form or another the original signal from the head unit does exist at that point. You'll need to remove two 10mm bolts with a wrench to get to it.
NOTE: If you have a 2014+ Mazda 3 you have a TAU (tuner amplifier unit) behind and to the side of the glovebox that acts as a middle man between the head unit and BOSE amp by handling various things like input selection.
So back to the post... Unfortunately, after he figured out how he wanted to hook up his subwoofer he was very vague with explaining how he did it. Pretty much he just mentioned which of the three cables at the BOSE amp he tapped into and said he used the grey and blue (which happen to be white and red on the Mazda 3) wires present in that cable. No mention at all of exactly what signal was carried over those two wires. Was it a stereo signal before its split into the supposed 4 channels? Was it one of the 4 channels? Which wire was the positive/signal wire and which wire was the negative/reference?
Here are the cable/wires he was talking about and their corresponding section of the workshop manual that shows what they connect to:
![275414]( "275414")
As you can see, the cable he referred to has 10 wires total. First, lets start with the easiest part. If you look at the nearby section of the BOSE diagrams labeled 0920-5e, you'll see that the pins 1G and 1H on cable 0920-322A go to the microphone used for the BOSE AudioPilot feature:
![275415]( "275415")
With those taken care of, that leaves 8 unidentified wires on that cable, including the two mentioned in the other post on Mazda6club, which are pins 1J and 1I (I know this now because in the Mazda 6 these wires are blue and grey respectively). 0920-5c shows these 8 wires grouped together, clearly signifying that they are related in someway, and in addition they are all shielded, as audio cables tend to be. Although, they are completely unlabeled, which nearly drove me insane (how is this information supposed to be useful to Mazda service mechanics if its this mystified?) and required a bit of intuition to figure out. Given that the user in that other post ended up using two of these wires and that he would at least have had to tap into one channel of audio to connect a sub, I surmised this after thinking about it for a bit: With 2 wires per channel, and a total of 8 unknown wires... 8/2 = 4 potential channels. Since I had heard that the head unit does output 4-channel, low-level signals and that for only adding a sub-woofer tapping into just one channel would largely be adequate, I thought that this quite reasonably could be the case. So, I set out to my own car to test out this theory.
Sweet 4-Channel Glory
I started by pulling 0920-332A and manually connecting only the aforementioned 1J white and 1I red wires (blue and grey in this picture since it was my 6) like so:
![Image]()
Then, I turned on the radio while making sure that Centerpoint and AudioPilot were off, and the Fade, Balance, Treble, and Bass controls were are at their neutral positions. And.... eureka! There was only audio coming from the rear right of the car (other than low bass still coming through the front door speakers since they act as woofers/sub-woofers in this car). I quickly repeated the test for the remaining 3 pairs of wires to see if the rest of my theory was correct, and indeed it was. Each time audio was clearly limited to one corner of the car. Here are the groupings:
Front Left Channel - Black and Red (1O and 1P)
Front Right Channel - White and Green (1M and 1N)
Rear Left Channel - Green and Black (1K and 1L)
Rear Right Channel - Red and White (1I and 1J)
This is great, but still leaves the question of exactly whats coming over each wire in each of those pairs. So, I brought my scope out to my car to find out. I first generated a simple asymmetrical test waveform in Audacity so it would be easy to tell if anything was being read in reverse on the scope, then played it over the system while monitoring one of the audio channels arbitrarily. I believe I started with the front left and had the black 1O wire connected to channel 1 of my scope, while the red 1P wire was connected to channel 2 of my scope. Here are the results:
![Image]()
Left to Right: (1) Source test signal in Audacity snippet. (2) Channel 1 of scope connected to black 1O wire on 0920-322A. (3) Channel 2 of scope connected to red 1P wire on 0920-322A.
These results show that the head unit uses differential signaling, as the signal on the red 1P wire is the exact inverse of the signal on the black 1O wire. The BOSE amp takes the difference of these signals to get the original signal with twice the voltage and any noise common to both lines subtracted out entirely. Again, I'm keeping explanations very short and summarized because while I'd love to relive my university courses (/s), this is already a painfully long post. I will continue hyperlinking relevant pages that give more detailed explanations. Anyway, this means that if you're going to use an amp that accepts differential signals you'll get to make use of the technique's excellent noise rejection, and if not, it isn't too difficult to convert this to a traditional single-ended signal. You can get an idea of how this works from the following scope shot in which I manually subtracted channel 2 from channel 1 (it looks really rough because the resolution of the math functions on this budget scope is poor):
![Image]()
This means that the black 1O wire is the non-inverted signal of the front left channel, while the red 1P wire is the inverted signal of the front left channel. Checking the other channel pairs revealed that this arrangement was consistent for all of them so that all of the non-inverted signals are along the top of the cable and all of the inverted signals are along the bottom of the cable. Nice.
Here is a summary of what we've learned so far:
![275444]( "275444")
But Are They Good Signals?:
We've confirmed that direct access to the head unit's 4-channel, differential audio is available on this car and that you can hook into it without the need of a hi-lo level converter, which by itself is pretty rad. But how usable even are these signals? Well that is what I set to find out next.
First, I wanted to make sure there were no issues in terms of level-balance and time-alignment, and that the left and right channels were truly separate with no weird mixing or other funny business going on. So I fired up audacity again and made a stereo version of the test signal I showed previously, utilizing a square wave so that the left channel of the stereo signal could easily be distinguished from the right. Then, I played it through the head unit while utilizing all 4 channels on my scope to monitor the non-inverted signals of the front left, front right, rear left, and rear right channels simultaneously. Here is what that looked like:
![Image]()
Left to Right: (1) Source test stereo signal in Audacity snippet. (2) FL Channel on scope channel 1, FR Channel on scope channel 2, RL channel on scope channel 3, and RR channel on scope channel 4
The on-screen cursors highlight that all four channels are in perfect alignment in terms of phase so you don't need to worry about adding any delays via a processor for your aftermarket amp. Additionally, all four channels are shown to be balanced in terms of amplitude, particularly by the peak-to-peak measurements at the bottom of the shot, so no volume differences need to be made per channel. The reason the right channels (teal and blue) are measured to be a bit higher in potential than the left channels and are slightly different from each other is due to the overshoot oscillations that you can see occur every time the wave switches from high to low and vice versa, known as "ringing". This distortion is primarily because of what is known as Gibbs effect and is an entirely normal phenomenon that in part occurs when performing digital-to-analog conversion, and doesn't at all indicate an issue with the BOSE head unit. Since the oscillations spike above intended height of the square wave (hence "overshoot") the peak-to-peak voltage is read to be higher than the sine wave equivalents. While performing this test I used the vertical cursors to measure a rough fit of the square waves without the fluctuations and confirmed that they had the same amplitude as the left channel sine waves; except.... I somehow failed and moved them before taking the above screenshot. So this one thing you'll have to take my word on ? . Even so, you can tell at a glance that they're roughly the same and can compare the pixel height of the waves in the screenshot in a photo editor if you really want.
The next big check was to see if the BOSE head unit performs any kind of equalization (hopefully not) or if it all occurs at the BOSE amp (hopefully so). I accomplished this using white noise, since it, if reproduced properly, should have a flat frequency response curve. The video goes into this in more detail, but here's the short version: With the help of Audacity, Spek, MATLAB, and the Fourier transform I generated uniform white noise, played it through the headunit from my laptop and then analyzed the output along the headunit's 4-channels. The finite memory of my scope means that I was only able to capture samples that were 24 seconds in length while using all 4 channels, so I compared the results to an input of the same length. So, here is the input I used:
![Image]()
Left to Right: (1) Spectrogram of input uniform white noise stereo signal. (2) Input uniform white noise time signal and its single-sided Fourier transform, left-channel (the right is the same)
The nearly uniform appearance of the white noise spectrogram is due to the same property of white noise that gives it a flat frequency curve: an even distribution of signal power across the frequency spectrum. The snow like effect in the spectrogram is another view of the slight hills and valleys you can see in the frequency curve of the signal in the MATLAB plot. At first you may wonder what good this signal is if it already doesn't have a perfectly flat frequency curve before its even been played through anything that has a chance of distorting it. But if you stop for a moment and look at the Y-axis scale of the MATLAB plot you will see that it is quite small, and so those minor imperfections aren't nearly as significant as they seem at first. Put simply, this is completely normal for white noise that is this short (white noise's frequency curve asymptotically approaches a perfectly flat appearance as more samples, and therefore a longer duration, are/is used in it's generation). So, this input reference point should be considered to have a flat curve within the scope of these tests.
With that out of the way, here are the output results:
![Image]()
Left to Right: (1) Spectrogram of the white-noise headunit output FL and FR combined into a stereo signal. (2) same as 1, but for the RL and RR channels
![Image]()
Left to Right: (1) FL white noise output. (2) FR white noise output. (3) RL white noise output. (4) RR white noise output.
First off, no two generations/reproductions of white noise will look exactly the same. Second, the minor color differences between the spectrograms are simply because I did not record these tests at quite the same exact volume. Unfortunately, it takes over an hour to dump this short data from my scope to a flash drive so I really did not have time to redo these once I realized this. Regardless, its the overall pattern of the output spectrogram and slope of the output frequency curve that are important here. Finally, yes there is a slight droop in the output frequency curves at the very top of the shown frequency range, but if you look closely at the spectrograms where it's most noticeable, 90% of the droop occurs above 20kHz, which is well outside the upper limit of audible frequencies for most people. Additionally, from further experimentation I believe this to be due to limitations of the laptop I used and not of the BOSE headunit. It is also worth noting that because of the memory limitations of my scope I couldn't record the output from all the channels at the same time (since there are 8 wires to monitor and I only have 4 channels), so the samples captured for each channel are not exactly the same, which resulted in further variation between the appearance of the above channel plots. But again, these variations are expected and are of no concern.
Aside from those minor points of note, these are solid results. If there was any deliberate equalization going on it would be very obvious, so yes, again the minor peaks and valleys in the frequency response of the output are not of any concern and the output signals are effectively flat.
If the tiny part of the slight droop that occurs just below 20Hz turns out to not be due to my laptop and is caused by the BOSE head unit and it bothers you, this is easily corrected by a correspondingly slight positive slope in equalization at the location the droop starts in order to cancel it out. Even a very basic processor would be capable of accomplishing this.
I did find one post on a forum for the CX-5 that suggested the headunit may have a bass roll-off curve at the low end of the spectrum based on how their system sounded to them. It would make sense since most stock speakers can't handle loud, lower frequencies without damage, though one would hope that this would occur at the BOSE amp and not the head unit. Well, while I can't speak for the CX-5 and I cannot be certain about 3s or 6s other than the 2016, I can say for sure that this is not the case for the 2016 6 (it is most likely the same for the rest though), as shown in the below frequency spectrum zoom in:
![Image]()
Base range close-up of the uniform white-noise output for the front left channel
The result is the same for the other 3 channels. I've also seen speculation that as the volume of the system is raised, the lower frequencies increase dis-proportionally to the the middle and higher frequencies. This is another realistic possibility, but again if it does occur it seems to happen within the BOSE amp and not the head unit. I say this because the white noise output results you see above were all performed while the head unit was around a volume level of 40 which is fairly high (it caps at 63 I believe), certainly at the point where I'd expect this adjustment to take effect, and as you can clearly see there is no droop in the base frequencies whatsoever.
As a double check, and in part for fun, I repeated this test with a random song from my library and the following are comparisons of the input and output. I arbitrarily chose to show the rear left and right here because the results were the same as the front:
![Image]()
Left to Right: (1) Input 24 segment of Nirvana - In Bloom Stereo. (2) Output 24 segment of Nirvana - In Bloom Rear Stereo
![Image]()
Left to Right: (1) Input 24 segment of Nirvana - In Bloom Left. (2) Output 24 segment of Nirvana - In Bloom Rear Left. (3) Input 24 segment of Nirvana - In Bloom Right. (4) Output 24 segment of Nirvana - In Bloom Rear Right
I synchronized the output with the original input file and made slight adjustments to the volumes of each to try to match them as close as possible (to which I got fairly close, but it's not perfect). They aren't exactly the same, but the only major discrepancy is the slight volume difference at the top end of the shown frequency spectrum, a detail that I already touched on with the white noise. Again, this is most likely more related to the equipment I used rather than issues with the head unit. Otherwise, a large majority of the spectrograms and frequency curves are nearly identical.
To wrap this section up, the following are examples of the white noise and the Nirvana song being observed via the head unit/TAU outputs:
![Image]()
Left to Right: (1) FL Non-Inverted & Inverted Signals, and FR Non-Inverted & Inverted signals while playing white noise. (2) RL Non-Inverted & Inverted Signals, and RR Non-Inverted & Inverted signals while playing In Bloom.
Finally, for the heck of it, here is the same 24 second segment of In Bloom dumped from my scope and formatted into a WAV file: In Bloom - 2016 Mazda 6 BOSE Direct Scope Dump.wav.
There's more?:
At this point we've seen everything we need to for hooking up an aftermarket amp to this system in the best way possible, but there are still a few unknowns about the connections to the stock BOSE amp that I figured I'd look into while I was at it.
I've already covered everything on connector 0920-322A, so now lets take a look at cable 0920-322B
![275417]( "275417")
The following pins are straightforward connections to speakers (that for once are nicely labeled), which are shown in the summary section of this post, but for now will be skipped since there is no mystery in their purpose: 2O, 2P, 2M, 2K, 2I, 2G, 2E, 2C, 2A, and 2B. This just leaves:
2N - Red wire
2L - White wire
2J - Light Green wire
2H - Yellow wire
2F - Tan wire
2D - Grey wire
Let's start from the top. 2N and 2L are connected to the High Speed CAN Bus of the car, with 2N being the low wire and 2L being the high wire. Nearly everything entertainment related in this (and most) vehicle(s) is connected to the Medium Speed CAN Bus, so what business does the BOSE amp have being a part of this network? The fact the amp is connected to the more essential CAN bus that carries data related to fuel level, oil level, passenger seat occupancy, etc., alone makes some people hesitant to remove it even if they are replacing all of their speakers. While I do have a logic analyzer, this car's High Speed CAN bus is notoriously traffic dense and it would take a ton of time to isolate exactly what the BOSE amp does on that network; however, I honestly don't think such a thing is necessary and instead a little bit of intuition can solve this mystery for us. Here is a segment of an article talking about BOSE's AudioPilot feature:
2J on the other hand is easy if you just look at the workshop manual schematics. Its a connection to chassis ground. Wooh.
2H and 2F are grouped together in the workshop schematics but are also completely unlabeled. Stupid... Fortunately, I was able to figure out their function after playing around on the headunit with my scope probing the wires. It seems to be that their only purpose is to act as a medium for very simple serial communication between the head unit and BOSE amp that is for sharing whether or not Centerpoint and/or Audiopliot are on/off. Nothing else has any effect on the signal these wires carry. Here are shots from the scope:
![Image]()
Left to Right: (1) Just Centerpoint on, 2H & 2F. (2) Just AudioPilot on, 2H & 2F. (3) Both on, 2H & 2F
There was supposed to be fourth picture of the signals while both features were off, but somehow my scope screwed that up and just saved a blank image, that I again, didn't realize until much later. The scope is actually pretty good in most respects, but my god are the USB functions a disaster (supposedly its a known issue and I'm not the only one who's noticed). It was the same as the third image with both features on, except that the order of the signal going high and then low was reversed (so low and then high instead).
Anyway, this seems to just be a simple 2-bit signal that tells the BOSE amp if it needs to employ either or both of these features. The curves of both signals are indicative of an RC circuit charging and discharging, which is quite common for generating signals in serial communication, or even more specifically, when trying to recreate square waveforms. This is clearly the case with the top signal, though the lower signal is classical of an RC differentiator for whatever reason. I'm not sure why the latter signal takes this form, but its spikes always follow the falling and rising edges of the top square-wave like signal. Additionally, the signal isn't produced if the BOSE Amp isn't powered, so its safe to say that the yellow wire on pin 2H is the head unit transmit and BOSE amp receive line, while the tan wire on pin 2F is the BOSE transmit/head unit receive line. Given these two facts, I'm guessing that the lower signal is simply feedback from the BOSE amp to the head unit to confirm it has changed modes. If you've ever sniffed the main serial Tx line of the 2014+ head unit (like when installing Mazda AIO Tweaks via the serial method for example) you'll known that the head unit really likes to log a shit ton of information, so I wouldn't be surprised if this was the case.
The order is a bit weird (the one feature seems to be normally high, while the other seems to be normally low), but it appears the 2-bit value is communicated as follows:
00 - AudioPilot on, Centerpoint off
01 - AudioPilot on, Centerpoint on
10 - AudioPilot off, Centerpoint off
11 - AudioPilot off, Centerpoint on
or individually:
AudioPilot - 0 = on, 1 = off
Centerpoint 0 = off, 1 = on
For connector 0920-322B that just leaves 2D or the grey wire. Lucky for us, this one is actually labeled again. The wire is connected to the "ACC" or Accessory line going to the fuse box, which means that it is most likely a turn-on wire that receives power once the car is in ACC mode or the engine is running. Useful to have for your aftermarket amp if you need it. I checked this out with my multi-meter just to make sure, and indeed this was the case:
![Image]()
Left to Right: (1) Car completely off, (2) Car in ACC mode with engine off and car in ACC with engine started
Now for the last connector, 0920-322C, which has the least number of connections and is the most straightforward:
![275418]( "275418")
Like in the second connector, the following pins are straightforward connections to speakers: 3G, 3H, 3E, 3F, 3C, and 3D. This just leaves:
3A - Green Wire
3B - Black Wire
3A is connected to the 25A BOSE fuse in the Fuse Block, and as is probably obvious is the general 12.5V power line that is available even when the car is off:
![Image]()
As for the black 3B wire... you guessed it! Just another connection to chassis ground.
And there you have it, the whole BOSE system mapped out (at least in terms of audio, i.e., excluding Nav, back-up cam, and other entertainment system features).
Feature Retention:
The last major concern when bypassing any stock equipment is the big question of "will I lose any features?". Well, be concerned no longer! You lose absolutely nothing!... other than the obvious loss of BOSE Centerpoint and Audiopilot, but if you cared about those you wouldn't be replacing the system in the first place lol. But yes, it really is that simple.
The small volume knob near the commander wheel and all of the sound controls in the head unit (fade, balance, treble, and bass) manipulate the signal before it reaches the BOSE Amp. All inputs, including bluetooth, are handled by the head unit/TAU and all call, Google Assistant/Android Audio/Apple Carplay/Mircrophone based features are handled pre-BOSE amp. Lastly, all notification sounds within the car such as the seatbelt chime and the blindspot monitoring system are all on their own dedicated circuits with a dedicated buzzer and were all tested to work with the BOSE amp unplugged entirely.
I had tried to save a clip of me checking my voicemail over bluetooth in both the Mazda software and Android Auto, but my scopes save function failed me AGAIN. Just trust me, it all works.
If you want a bit more concrete proof, just check out the video. I go into this a bit more deeply there.
Summary
In the end, things look pretty great in terms of audio upgrade potential for those that have a 3 with Blowse.
For your convenience, here is a pin-out of all the connections to the BOSE amp that I covered throughout the thread:
![275419]( "275419")
CONTINUED IN NEXT POST
This is a Mazda 3 focused version of my Mazda 6 BOSE analysis guide. It has been adapted from the original so that color wiring and explanations match exactly how things are in the 2014-2018 Mazda 3. The general progression of how I discovered things is still presented from the perspective of myself exploring my Mazda 6, but this information fully applies to the 3 and as I said I have made the necessary changes to this version after studying the 2014 Mazda Workshop manual. If you want the Mazda 6 version of the guide see here: 2014-2018* Mazda 6 (and 2014-2016 Mazda 3) w/BOSE Full System Breakdown/Analysis
*Applicable Models:
- 2014-2016 Mazda 3: 100% confident
- 2014-2016 Mazda 6: 100% confident
- 2017-2018 Mazda 3: 95% confident
- 2017-2018 Mazda 6: 95% confident
- 2014-2018 Mazda CX-3,5 and 9 may have some differences
- 2019+ Anything: Unsure
This is a formal documentation of my findings while investigating the BOSE system on my own 2016 Touring 6 out of frustration over the lack of information online that I could be certain applied to the 2016 model.
I simply want to replace the entirety of the audio system in my car while maintaining the factory head unit since I couldn't find any of the ones specifically made for this car that I was a big fan of, and using any of the generic aftermarket units would mean losing tons of functionality (maintenance status, lock preferences, commander knob, etc.). Supposedly, the 3rd gen 6s with the BOSE system supply straight, line-level outputs from their head unit to the BOSE amp that you can simply tap into directly for the connections to your aftermarket amp; however, after countless days of searching, all I could manage to find were similar anecdotal statements of this fact and poorly detailed diagrams with no clear indication of what vehicle they were in reference to. I was able to find some information that was a bit more definitive, but unfortunately it was for the CX-5, and I didn't want to just assume it would be the same in my 2016 6. Even asking others on this forum for more concrete info on where exactly I could tap this signal from lead to dead ends.
On and off I considered looking into the system myself, but with no real starting point, I didn't want to dedicate too much time to something that could prove to be fruitless and that initially I wasn't completely committed to yet. Luckily, I eventually struck gold by finding one single thread on this very forum by someone else that also happened to have a 2016 6 with the BOSE upgrade who shared some details about how he added a sub to his car while utilizing the "original signal" from the head unit. Annoyingly, he was pretty vague with his descriptions, but there was just enough substance to give me a starting point for my own investigation.
I know some people have tapped into the rear speakers that are supposedly full-range (though possibly not quite) and used a hi-low level converter to add an aftermarket amp, but I too would prefer to use the original signal if it truly is as accessible as others have made it sound. So without further ado, the following is a recount of what I ended up doing and what I ended up figuring out.
This post has a companion video on YouTube. The video is more focused on the process and explanations of my investigation, while this post is more focused on the results and details.
Video:
If you only want to see part of the video there are time stamps in the description if you head over to the actual YouTube site. The video is fairly long at about 35 minutes, but was made so that you don't have to just take my word on how the system is setup and can get a better sense of the process required to determine my findings if you need/want to.
Documentation:
Other than the singular post I mentioned, my only reference point was the 2014 Mazda 6 Workshop manual (and in this case the 2014 Mazda 3 Workshop manual) that you can find pretty easily with Google. I won't be posting a link to it, but will be using various images from select sections of the Electrical Diagrams segment. As far as I know, there doesn't exist a newer workshop manual past the 2014 version (at least not that is publicly available), which initially was a large part of my troubles. You can find places to buy the manual that list it as 2014-2016 or 2014-2017, etc, but it's just the 2014 version and has no updates whatsoever. For a large portion of the car this is a non-issue as most of it has remained the same over the years, but there have been some significant changes from 2014 till now (only true for the 6) that obviously are not covered in this dated document and therefore are a major frustration point. The headlights received extra chrome trim in the 2017 model, the 2018 model had an entire front and rear face lift, and most importantly for this topic, the head unit changed starting with the 2016 model. Because of this, there are portions of the electrical diagrams in this workshop manual that are no longer correct (again, all of this is only true for the 6); however, after having inspected my 2016 myself, I can safely say that enough of it still applies (primarily in terms of the connections going to and from the BOSE amp) that every connection you need to hook an aftermarket amp into the BOSE systems "original signal" is still accurately portrayed for models past the 2014.
Starting Point:
There is a fair amount of raw data to cover here, so I will be very brief in my explanations of why I did something or how a result proves the conclusions I made. Again, if you want a bit more detail into my process, watch the above video. If you are confused or want more info about a particular part, just quote me (though please snip this huge post out lol).
OK, so here we go. This is the thread I referred to earlier that got me started: 2016 headunit location
The gist of it is that the OP discovered/confirmed that in the 2016+ models (2014+ for the Mazda 3) the BOSE amp is under the passenger seat and that in one form or another the original signal from the head unit does exist at that point. You'll need to remove two 10mm bolts with a wrench to get to it.
NOTE: If you have a 2014+ Mazda 3 you have a TAU (tuner amplifier unit) behind and to the side of the glovebox that acts as a middle man between the head unit and BOSE amp by handling various things like input selection.
So back to the post... Unfortunately, after he figured out how he wanted to hook up his subwoofer he was very vague with explaining how he did it. Pretty much he just mentioned which of the three cables at the BOSE amp he tapped into and said he used the grey and blue (which happen to be white and red on the Mazda 3) wires present in that cable. No mention at all of exactly what signal was carried over those two wires. Was it a stereo signal before its split into the supposed 4 channels? Was it one of the 4 channels? Which wire was the positive/signal wire and which wire was the negative/reference?
Here are the cable/wires he was talking about and their corresponding section of the workshop manual that shows what they connect to:
As you can see, the cable he referred to has 10 wires total. First, lets start with the easiest part. If you look at the nearby section of the BOSE diagrams labeled 0920-5e, you'll see that the pins 1G and 1H on cable 0920-322A go to the microphone used for the BOSE AudioPilot feature:
With those taken care of, that leaves 8 unidentified wires on that cable, including the two mentioned in the other post on Mazda6club, which are pins 1J and 1I (I know this now because in the Mazda 6 these wires are blue and grey respectively). 0920-5c shows these 8 wires grouped together, clearly signifying that they are related in someway, and in addition they are all shielded, as audio cables tend to be. Although, they are completely unlabeled, which nearly drove me insane (how is this information supposed to be useful to Mazda service mechanics if its this mystified?) and required a bit of intuition to figure out. Given that the user in that other post ended up using two of these wires and that he would at least have had to tap into one channel of audio to connect a sub, I surmised this after thinking about it for a bit: With 2 wires per channel, and a total of 8 unknown wires... 8/2 = 4 potential channels. Since I had heard that the head unit does output 4-channel, low-level signals and that for only adding a sub-woofer tapping into just one channel would largely be adequate, I thought that this quite reasonably could be the case. So, I set out to my own car to test out this theory.
Sweet 4-Channel Glory
I started by pulling 0920-332A and manually connecting only the aforementioned 1J white and 1I red wires (blue and grey in this picture since it was my 6) like so:
Then, I turned on the radio while making sure that Centerpoint and AudioPilot were off, and the Fade, Balance, Treble, and Bass controls were are at their neutral positions. And.... eureka! There was only audio coming from the rear right of the car (other than low bass still coming through the front door speakers since they act as woofers/sub-woofers in this car). I quickly repeated the test for the remaining 3 pairs of wires to see if the rest of my theory was correct, and indeed it was. Each time audio was clearly limited to one corner of the car. Here are the groupings:
Front Left Channel - Black and Red (1O and 1P)
Front Right Channel - White and Green (1M and 1N)
Rear Left Channel - Green and Black (1K and 1L)
Rear Right Channel - Red and White (1I and 1J)
This is great, but still leaves the question of exactly whats coming over each wire in each of those pairs. So, I brought my scope out to my car to find out. I first generated a simple asymmetrical test waveform in Audacity so it would be easy to tell if anything was being read in reverse on the scope, then played it over the system while monitoring one of the audio channels arbitrarily. I believe I started with the front left and had the black 1O wire connected to channel 1 of my scope, while the red 1P wire was connected to channel 2 of my scope. Here are the results:
Left to Right: (1) Source test signal in Audacity snippet. (2) Channel 1 of scope connected to black 1O wire on 0920-322A. (3) Channel 2 of scope connected to red 1P wire on 0920-322A.
These results show that the head unit uses differential signaling, as the signal on the red 1P wire is the exact inverse of the signal on the black 1O wire. The BOSE amp takes the difference of these signals to get the original signal with twice the voltage and any noise common to both lines subtracted out entirely. Again, I'm keeping explanations very short and summarized because while I'd love to relive my university courses (/s), this is already a painfully long post. I will continue hyperlinking relevant pages that give more detailed explanations. Anyway, this means that if you're going to use an amp that accepts differential signals you'll get to make use of the technique's excellent noise rejection, and if not, it isn't too difficult to convert this to a traditional single-ended signal. You can get an idea of how this works from the following scope shot in which I manually subtracted channel 2 from channel 1 (it looks really rough because the resolution of the math functions on this budget scope is poor):
This means that the black 1O wire is the non-inverted signal of the front left channel, while the red 1P wire is the inverted signal of the front left channel. Checking the other channel pairs revealed that this arrangement was consistent for all of them so that all of the non-inverted signals are along the top of the cable and all of the inverted signals are along the bottom of the cable. Nice.
Here is a summary of what we've learned so far:
But Are They Good Signals?:
We've confirmed that direct access to the head unit's 4-channel, differential audio is available on this car and that you can hook into it without the need of a hi-lo level converter, which by itself is pretty rad. But how usable even are these signals? Well that is what I set to find out next.
First, I wanted to make sure there were no issues in terms of level-balance and time-alignment, and that the left and right channels were truly separate with no weird mixing or other funny business going on. So I fired up audacity again and made a stereo version of the test signal I showed previously, utilizing a square wave so that the left channel of the stereo signal could easily be distinguished from the right. Then, I played it through the head unit while utilizing all 4 channels on my scope to monitor the non-inverted signals of the front left, front right, rear left, and rear right channels simultaneously. Here is what that looked like:
Left to Right: (1) Source test stereo signal in Audacity snippet. (2) FL Channel on scope channel 1, FR Channel on scope channel 2, RL channel on scope channel 3, and RR channel on scope channel 4
The on-screen cursors highlight that all four channels are in perfect alignment in terms of phase so you don't need to worry about adding any delays via a processor for your aftermarket amp. Additionally, all four channels are shown to be balanced in terms of amplitude, particularly by the peak-to-peak measurements at the bottom of the shot, so no volume differences need to be made per channel. The reason the right channels (teal and blue) are measured to be a bit higher in potential than the left channels and are slightly different from each other is due to the overshoot oscillations that you can see occur every time the wave switches from high to low and vice versa, known as "ringing". This distortion is primarily because of what is known as Gibbs effect and is an entirely normal phenomenon that in part occurs when performing digital-to-analog conversion, and doesn't at all indicate an issue with the BOSE head unit. Since the oscillations spike above intended height of the square wave (hence "overshoot") the peak-to-peak voltage is read to be higher than the sine wave equivalents. While performing this test I used the vertical cursors to measure a rough fit of the square waves without the fluctuations and confirmed that they had the same amplitude as the left channel sine waves; except.... I somehow failed and moved them before taking the above screenshot. So this one thing you'll have to take my word on ? . Even so, you can tell at a glance that they're roughly the same and can compare the pixel height of the waves in the screenshot in a photo editor if you really want.
The next big check was to see if the BOSE head unit performs any kind of equalization (hopefully not) or if it all occurs at the BOSE amp (hopefully so). I accomplished this using white noise, since it, if reproduced properly, should have a flat frequency response curve. The video goes into this in more detail, but here's the short version: With the help of Audacity, Spek, MATLAB, and the Fourier transform I generated uniform white noise, played it through the headunit from my laptop and then analyzed the output along the headunit's 4-channels. The finite memory of my scope means that I was only able to capture samples that were 24 seconds in length while using all 4 channels, so I compared the results to an input of the same length. So, here is the input I used:
Left to Right: (1) Spectrogram of input uniform white noise stereo signal. (2) Input uniform white noise time signal and its single-sided Fourier transform, left-channel (the right is the same)
The nearly uniform appearance of the white noise spectrogram is due to the same property of white noise that gives it a flat frequency curve: an even distribution of signal power across the frequency spectrum. The snow like effect in the spectrogram is another view of the slight hills and valleys you can see in the frequency curve of the signal in the MATLAB plot. At first you may wonder what good this signal is if it already doesn't have a perfectly flat frequency curve before its even been played through anything that has a chance of distorting it. But if you stop for a moment and look at the Y-axis scale of the MATLAB plot you will see that it is quite small, and so those minor imperfections aren't nearly as significant as they seem at first. Put simply, this is completely normal for white noise that is this short (white noise's frequency curve asymptotically approaches a perfectly flat appearance as more samples, and therefore a longer duration, are/is used in it's generation). So, this input reference point should be considered to have a flat curve within the scope of these tests.
With that out of the way, here are the output results:
Left to Right: (1) Spectrogram of the white-noise headunit output FL and FR combined into a stereo signal. (2) same as 1, but for the RL and RR channels
Left to Right: (1) FL white noise output. (2) FR white noise output. (3) RL white noise output. (4) RR white noise output.
First off, no two generations/reproductions of white noise will look exactly the same. Second, the minor color differences between the spectrograms are simply because I did not record these tests at quite the same exact volume. Unfortunately, it takes over an hour to dump this short data from my scope to a flash drive so I really did not have time to redo these once I realized this. Regardless, its the overall pattern of the output spectrogram and slope of the output frequency curve that are important here. Finally, yes there is a slight droop in the output frequency curves at the very top of the shown frequency range, but if you look closely at the spectrograms where it's most noticeable, 90% of the droop occurs above 20kHz, which is well outside the upper limit of audible frequencies for most people. Additionally, from further experimentation I believe this to be due to limitations of the laptop I used and not of the BOSE headunit. It is also worth noting that because of the memory limitations of my scope I couldn't record the output from all the channels at the same time (since there are 8 wires to monitor and I only have 4 channels), so the samples captured for each channel are not exactly the same, which resulted in further variation between the appearance of the above channel plots. But again, these variations are expected and are of no concern.
Aside from those minor points of note, these are solid results. If there was any deliberate equalization going on it would be very obvious, so yes, again the minor peaks and valleys in the frequency response of the output are not of any concern and the output signals are effectively flat.
If the tiny part of the slight droop that occurs just below 20Hz turns out to not be due to my laptop and is caused by the BOSE head unit and it bothers you, this is easily corrected by a correspondingly slight positive slope in equalization at the location the droop starts in order to cancel it out. Even a very basic processor would be capable of accomplishing this.
I did find one post on a forum for the CX-5 that suggested the headunit may have a bass roll-off curve at the low end of the spectrum based on how their system sounded to them. It would make sense since most stock speakers can't handle loud, lower frequencies without damage, though one would hope that this would occur at the BOSE amp and not the head unit. Well, while I can't speak for the CX-5 and I cannot be certain about 3s or 6s other than the 2016, I can say for sure that this is not the case for the 2016 6 (it is most likely the same for the rest though), as shown in the below frequency spectrum zoom in:
Base range close-up of the uniform white-noise output for the front left channel
The result is the same for the other 3 channels. I've also seen speculation that as the volume of the system is raised, the lower frequencies increase dis-proportionally to the the middle and higher frequencies. This is another realistic possibility, but again if it does occur it seems to happen within the BOSE amp and not the head unit. I say this because the white noise output results you see above were all performed while the head unit was around a volume level of 40 which is fairly high (it caps at 63 I believe), certainly at the point where I'd expect this adjustment to take effect, and as you can clearly see there is no droop in the base frequencies whatsoever.
As a double check, and in part for fun, I repeated this test with a random song from my library and the following are comparisons of the input and output. I arbitrarily chose to show the rear left and right here because the results were the same as the front:
Left to Right: (1) Input 24 segment of Nirvana - In Bloom Stereo. (2) Output 24 segment of Nirvana - In Bloom Rear Stereo
Left to Right: (1) Input 24 segment of Nirvana - In Bloom Left. (2) Output 24 segment of Nirvana - In Bloom Rear Left. (3) Input 24 segment of Nirvana - In Bloom Right. (4) Output 24 segment of Nirvana - In Bloom Rear Right
I synchronized the output with the original input file and made slight adjustments to the volumes of each to try to match them as close as possible (to which I got fairly close, but it's not perfect). They aren't exactly the same, but the only major discrepancy is the slight volume difference at the top end of the shown frequency spectrum, a detail that I already touched on with the white noise. Again, this is most likely more related to the equipment I used rather than issues with the head unit. Otherwise, a large majority of the spectrograms and frequency curves are nearly identical.
To wrap this section up, the following are examples of the white noise and the Nirvana song being observed via the head unit/TAU outputs:
Left to Right: (1) FL Non-Inverted & Inverted Signals, and FR Non-Inverted & Inverted signals while playing white noise. (2) RL Non-Inverted & Inverted Signals, and RR Non-Inverted & Inverted signals while playing In Bloom.
Finally, for the heck of it, here is the same 24 second segment of In Bloom dumped from my scope and formatted into a WAV file: In Bloom - 2016 Mazda 6 BOSE Direct Scope Dump.wav.
There's more?:
At this point we've seen everything we need to for hooking up an aftermarket amp to this system in the best way possible, but there are still a few unknowns about the connections to the stock BOSE amp that I figured I'd look into while I was at it.
I've already covered everything on connector 0920-322A, so now lets take a look at cable 0920-322B
The following pins are straightforward connections to speakers (that for once are nicely labeled), which are shown in the summary section of this post, but for now will be skipped since there is no mystery in their purpose: 2O, 2P, 2M, 2K, 2I, 2G, 2E, 2C, 2A, and 2B. This just leaves:
2N - Red wire
2L - White wire
2J - Light Green wire
2H - Yellow wire
2F - Tan wire
2D - Grey wire
Let's start from the top. 2N and 2L are connected to the High Speed CAN Bus of the car, with 2N being the low wire and 2L being the high wire. Nearly everything entertainment related in this (and most) vehicle(s) is connected to the Medium Speed CAN Bus, so what business does the BOSE amp have being a part of this network? The fact the amp is connected to the more essential CAN bus that carries data related to fuel level, oil level, passenger seat occupancy, etc., alone makes some people hesitant to remove it even if they are replacing all of their speakers. While I do have a logic analyzer, this car's High Speed CAN bus is notoriously traffic dense and it would take a ton of time to isolate exactly what the BOSE amp does on that network; however, I honestly don't think such a thing is necessary and instead a little bit of intuition can solve this mystery for us. Here is a segment of an article talking about BOSE's AudioPilot feature:
For me, that right there says it all. The only reason the Bose AMP is connected to the Hi Speed CAN Bus is so that it can access information broadcast by the vehicle's speedometer. Additionally, as a test I've driven around for quite a while with the Bose AMP completely disconnected and noticed no issues/oddities whatsoever other than the obvious lack of sound.
2J on the other hand is easy if you just look at the workshop manual schematics. Its a connection to chassis ground. Wooh.
2H and 2F are grouped together in the workshop schematics but are also completely unlabeled. Stupid... Fortunately, I was able to figure out their function after playing around on the headunit with my scope probing the wires. It seems to be that their only purpose is to act as a medium for very simple serial communication between the head unit and BOSE amp that is for sharing whether or not Centerpoint and/or Audiopliot are on/off. Nothing else has any effect on the signal these wires carry. Here are shots from the scope:
Left to Right: (1) Just Centerpoint on, 2H & 2F. (2) Just AudioPilot on, 2H & 2F. (3) Both on, 2H & 2F
There was supposed to be fourth picture of the signals while both features were off, but somehow my scope screwed that up and just saved a blank image, that I again, didn't realize until much later. The scope is actually pretty good in most respects, but my god are the USB functions a disaster (supposedly its a known issue and I'm not the only one who's noticed). It was the same as the third image with both features on, except that the order of the signal going high and then low was reversed (so low and then high instead).
Anyway, this seems to just be a simple 2-bit signal that tells the BOSE amp if it needs to employ either or both of these features. The curves of both signals are indicative of an RC circuit charging and discharging, which is quite common for generating signals in serial communication, or even more specifically, when trying to recreate square waveforms. This is clearly the case with the top signal, though the lower signal is classical of an RC differentiator for whatever reason. I'm not sure why the latter signal takes this form, but its spikes always follow the falling and rising edges of the top square-wave like signal. Additionally, the signal isn't produced if the BOSE Amp isn't powered, so its safe to say that the yellow wire on pin 2H is the head unit transmit and BOSE amp receive line, while the tan wire on pin 2F is the BOSE transmit/head unit receive line. Given these two facts, I'm guessing that the lower signal is simply feedback from the BOSE amp to the head unit to confirm it has changed modes. If you've ever sniffed the main serial Tx line of the 2014+ head unit (like when installing Mazda AIO Tweaks via the serial method for example) you'll known that the head unit really likes to log a shit ton of information, so I wouldn't be surprised if this was the case.
The order is a bit weird (the one feature seems to be normally high, while the other seems to be normally low), but it appears the 2-bit value is communicated as follows:
00 - AudioPilot on, Centerpoint off
01 - AudioPilot on, Centerpoint on
10 - AudioPilot off, Centerpoint off
11 - AudioPilot off, Centerpoint on
or individually:
AudioPilot - 0 = on, 1 = off
Centerpoint 0 = off, 1 = on
For connector 0920-322B that just leaves 2D or the grey wire. Lucky for us, this one is actually labeled again. The wire is connected to the "ACC" or Accessory line going to the fuse box, which means that it is most likely a turn-on wire that receives power once the car is in ACC mode or the engine is running. Useful to have for your aftermarket amp if you need it. I checked this out with my multi-meter just to make sure, and indeed this was the case:
Left to Right: (1) Car completely off, (2) Car in ACC mode with engine off and car in ACC with engine started
Now for the last connector, 0920-322C, which has the least number of connections and is the most straightforward:
Like in the second connector, the following pins are straightforward connections to speakers: 3G, 3H, 3E, 3F, 3C, and 3D. This just leaves:
3A - Green Wire
3B - Black Wire
3A is connected to the 25A BOSE fuse in the Fuse Block, and as is probably obvious is the general 12.5V power line that is available even when the car is off:
As for the black 3B wire... you guessed it! Just another connection to chassis ground.
And there you have it, the whole BOSE system mapped out (at least in terms of audio, i.e., excluding Nav, back-up cam, and other entertainment system features).
Feature Retention:
The last major concern when bypassing any stock equipment is the big question of "will I lose any features?". Well, be concerned no longer! You lose absolutely nothing!... other than the obvious loss of BOSE Centerpoint and Audiopilot, but if you cared about those you wouldn't be replacing the system in the first place lol. But yes, it really is that simple.
The small volume knob near the commander wheel and all of the sound controls in the head unit (fade, balance, treble, and bass) manipulate the signal before it reaches the BOSE Amp. All inputs, including bluetooth, are handled by the head unit/TAU and all call, Google Assistant/Android Audio/Apple Carplay/Mircrophone based features are handled pre-BOSE amp. Lastly, all notification sounds within the car such as the seatbelt chime and the blindspot monitoring system are all on their own dedicated circuits with a dedicated buzzer and were all tested to work with the BOSE amp unplugged entirely.
I had tried to save a clip of me checking my voicemail over bluetooth in both the Mazda software and Android Auto, but my scopes save function failed me AGAIN. Just trust me, it all works.
If you want a bit more concrete proof, just check out the video. I go into this a bit more deeply there.
Summary
In the end, things look pretty great in terms of audio upgrade potential for those that have a 3 with Blowse.
- Easy access to relatively clean, 4-channel audio directly from the head unit/TAU
- All channels are transmitted via differential signaling for great noise rejection if your amp supports it
- All channels are in-phase and balanced with the head unit at default settings
- No equalization performed on the signal pre-BOSE amp
- No features lost by tapping pre-BOSE amp (other than the BOSE processing obviously)
- 12.5v turn on wire available if you don't want to run your own. You most likely can't make use of the 12.5V power wire though because the fuse will blow if the line exceeds a mere 312.5W (25A fuse at 12.5V)
- You can safely remove the BOSE amp in its entirety if you are replacing all of your speakers or connecting some of them to the new amp (100% certain).
For your convenience, here is a pin-out of all the connections to the BOSE amp that I covered throughout the thread:
CONTINUED IN NEXT POST
