An age-old issue of mine was having to jack the volume on my car stereo (2008 BMW 328xi) way up when playing music from my phone (Galaxy Nexus) over the 3.5mm aux connection. I actually tried using a bluetooth receiver from Griffin Technologies to solve this problem but only found that I had to jack the volume up even more and that the whole bass-band disappeared when using it as well.
I found, however, that using my Objective2 headphone amplifier to amplify the aux in would both solve my low volume issue as well as make my music sound better in general. It was clear to me that a basic amplifier typically used for headphones would probably do the job, but I wasn't willing to swap out or recharge batteries every now and then to deal with it. (especially since I'd always accidentally leave the amp on after exiting my car)
The 12V power from a car's cigarette lighter is far from I an ideal power source, and in many vehicles it shares a common ground with everything else in the car. This leads to a lot of common noise and will also end up blowing up amplifiers using a virtual ground scheme such as the cMoy if you try powering them directly off car power. (as the stereo may also share the same common ground)
I ended up making a fairly basic circuit which consists of only 3 ICs. It basically takes in the 12V from your cigarette lighter and produces a regulated, isolated 12V output.
This powers a cMoyBB from JDS Labs which is configured with a gain of approximately 5. Since the cMoyBB board is nice and compact I'm able to cram my circuit in there right next to it.
I'm using pretty much as much heatsinking as possible on the LM2937 as it really does get quite hot. The cMoyBB only draws about 40mA of current even while playing music into headphones but the LM2937 has to deal with almost 300mA coming into it since it's driving the rest of the circuit.
Theory of Operation
The LM2937 IC from Texas Instruments sits right at the "front" of the circuit and takes the nominal 12V in directly from the car's cigarette lighter jack. This is a fixed 5V LDO regulator in a TO-263 form factor which uses the PCB as a heatsink. A more efficient option would've been using a switching power supply such as the TI TL2575HV-05-Q1, but in my case I didn't need to draw a whole lot of current and the LM2937 seemed like an ideal IC which requires very few external components.
The 5V out of this regulator then goes to a CC3-0512SF-E isolated DC-DC converter from TDK-Lambda which can take in anywhere from 4.5V-9V and boost it to 12-15V. (in this case I have it configured for 15V output by shorting the TRIM pin to -VOUT) This series of converters has everything conveniently under a metal hood and requires few external components.
The isolated 15V output from the DC-DC converter then goes to a LM340 linear regulator IC also from Texas Instruments. This is a nicely performing 12V fixed linear regulator also in TO-263 form factor. It provides a super low ripple 12V for the cMoyBB circuitry to do its job.
It might seem somewhat roundabout for me to regulate the 12V from the car down to 5V only to have to boost it back up to 12V again but there a good reason for this. Even though the power out of the cigarette lighter is nominally 12V it can really take on a rather large range of values depending on the circumstances. For example it could be only 5-6V during engine startup or 14.5V while the engine/alternator are running and I couldn't find a DC-DC converter that could live through all of that; however, the LM2937 is "ideally suited for automotive applications" according to the TI datasheet and will "protect itself and any load circuitry from reverse battery connections, two-battery jumps and up to +60V/−50V load dump transients." In other words the LM2937 is an easy way for me to ensure that my circuit and the cMoyBB are protected from any of the extreme transients that could happen in a car.
It might also seem redundant for me to have the DC-DC converter configured for 15V output when I could set it for 12V output and just use that to power the cMoyBB directly but there is good reason for this as well. The CC3-0512SF-E is spec'ed for a maximum ripple & noise of 120mVpp which is far too much for any sort of audio application. The opamp's PSRR would not be nearly enough to prevent that 120mVpp from being very audible in your music, but the LM340 outputs a far more manageable 0.04mVrms of ripple/noise.
I stress tested this circuit for a few hours driving a constant 100mA load with no issues. I recorded temps on the board as high as 50C at a few locations which is perhaps mildly concerning, but never did the circuit shutdown or output anything beyond 12.05V.
I've briefly field tested this in my 1996 Toyota Camry as well as my 2008 BMW 328xi with no issues. The interesting I noticed was that I would hear clipping when turning the amp's volume knob beyond about 1/2 of the way up in the 328xi and I saw the same issue when I was using a 9V battery as well. I had initially thought that perhaps this was due to the NJM4556 opamp not being able to swing high enough but the same thing occurred with other opamps as well. I've also confirmed this issue when using my Objective2 in high gain mode so it appears that my 328xi just doesn't like seeing over a 1Vrms signal or so at the aux jack.
I really hope BMW has done something about the low aux volume in newer generation models, (though perhaps with bluetooth no one cares about aux in anymore) as it is quite annoying for me to have to dial in the volume every time I switch between radio/aux sources. This is actually not as much of an issue in my old 1996 Toyota Camry at all since the cheap $30 stereo my dad and I installed has a setting for the aux source amplification which let's you match the volume levels up fairly closely.