[sfinxvc]

Hey forum, posting this thread to potentially benefit others.

I recently created this thread to explore my options to improve my base audio system in a 2010 328i.

Now that I've gotten a good understanding of the problem via that thread, and received a lot of helpful information and resources, I'm creating this one to chronicle what I hope will be cheap solutions. I'm still kind of in the midst of solving them though, so this will evolve. But...I'm also far enough along though that this is worth sharing now.

Why this requirement to be cheap? Mainly 'cause I'm a little skeptical about what's possible in such a small enclosed space, audio-wise. I'm somewhat familiar with acoustic problems in typical rooms, and that's quite hard to solve in and of itself (at least IMO). And because I want to be cautious investing in a E9X given its current market value. My car's in great mechanical shape otherwise and if I'm to spend money, I would like it to be for keeping it running smoothly through preventative maintenance, etc. Your priorities and wallet are probably different, and I don't claim to be qualified enough to judge.

Anyway, with all that out of the way...I've tried to employ cheap bastid solution #1 this week, and this is how it's going:

Solution #1 - leverage the stock HU's capability to directly power the speakers and use a cheap DSP to tackle the problems. For this, I'm currently using a MiniDSP 2x4. - Cost: $110 if new, even cheaper if used.

Here are my results for a sweep (1/24 oct smoothing) of the stock unit from driver's listening position (note, I omitted the DB scale since it's not comparable between the two graphs I'm posting, but the horizontal lines are 10db apart):



I then measured each door woofer near-field and generated a 6 band PEQ to smooth the response (in REW). I limited PEQ bands to upper bass and above (more or less). I applied this to the output PEQ for each channel.

I then did another sweep and contemplated both a high pass as well as simply PEQ'ing again to try to fix the bass. The 2x4 can set independent PEQs on inputs and outputs. So I tried to get each to overlap as little as possible. So this input PEQ targeted midbass and below. It wasn't perfect by any means, but it did succeed in flattening out what was a rather large (say 10db) bass shelf. I believe this shelf is a result of the underseat woofer playing full range. The capacitor on the door woofer probably rolls off early and creates that bad transition.

Anyway, after all that, this is what I ended up with (1/24 oct smoothing):



Pros:

• It's quite a bit smoother, and sounded that way as well. Better articulation between notes and an overall better balance between midrange and bass details.
• It's less harsh on the ears.
• It's more versatile genre-to-genre.

Cons:

• This doesn't magically improve the quality of the drivers, they're still very much a limiting factor and it's easy to hear.
• Since PEQ is subtractive, it "eats" into the available amplitude at every frequency, in other words, your overall volume cap is lowered. I did this while parked, so I don't know quite yet how it'll translate to background road noise. In a parked state though, it was loud enough. The limiting factor here is the stock HU. With a different, more powerful HU, like say the Eanon 9465B's 45Ws out, you wouldn't have this problem (but, I haven't splurged on one yet since I'm still trying to do this on the cheap).
• I'm using bluetooth audio to feed the MiniDSP 2x4 (feeds aux), which is my primary source. I never touch the radio. This obviously won't help with any function except however you get audio into your DSP of choice.

Other comments:

• I think the spikes around 35hz and 100hz are the Fs of the underseat woofer and door woofer, respectively. They seemed to show up on every measurement I did, and I did like 70 in the last 24 hours.
• The deep nulls between 300 and 900hz could be modes from the small enclosed space. I'm not 100% sure though. They seemed unfixable via PEQ. If anyone has tackled this problematic area before, let me know what you found or recommend.

Other solutions I'm considering...

• Add tweeters. As you can see, the output drops pretty drastically after about 10K. My only concern with this is how the stock HU will react to such a low impedance load. I read that the capacitor is in series with the door woofer not only to act as a filter, but also to reduce how hard the HU has to work to drive both it and the underseats in parallel.
• Get a HU that's more powerful. This has some benefits in that it doesn't require ripping apart the car to add wiring and what not and is pretty much a self install. However, things could get expensive, especially if I avoid the Eanons for their reputation of not working well (I don't know if this is well founded or not).
• Maybe the "nulls" aren't actually caused by modes and more by the idiosyncrasies of the stock door woofers. In that case buying some cheap widebanders could be a solution.

[ctuna]

explain to me how you put the dsp between the output stage on the Head unit
and the speakers?
So are you running it in through the aux?
This would be the same as running a DAC through the Aux.

This does not overcome the incredibly weak output of the Head Unit in any way.
Base audio

The "base" audio system has been around since the early days, although it only made it to the US in 2010.

It uses four channels of deck power (yes, that's right, good old 50Wx4 when lightning strikes, 18Wx4 if you're charitable, and 12Wx4 if you do the actual 12V math into 4 ohms and 1% THD).

[sfinxvc]

Yeah, it goes through aux.

"This does not overcome the incredibly weak output of the Head Unit in any way."

Correct.

[sfinxvc]

Follow up...

I managed to solve a few of the problems:

a) road noise and the volume cap were indeed issues. It turns out just plugging in the MiniDSP causes about a 10db loss in the midrange and treble, and about 5db in the bass. Just to give myself a little more gain to work with, I turned the stock HU's bass and treble tone controls to the maximum of +10 (this is fine since "shaping" of the response will happen via the DSP).

b) those nulls in the midrange actually weren't modes, I was able to get at them a bit with some targeted work around 350-450hz.

The curve now looks like this:


Loudness is good and tonal smoothness is decent (better than stock), however all this shaping also meant noise floor increased (verified with measurements). I was getting less noise when my MiniDSP was plugged into my laptop though, so I think if I can isolate USB power from that noisy 12V cigarette lighter, it'll help a bit.

Treble is now good enough too, as you can see in the graph I now get good extension to 12.5khz (compared to the pitiful 3khz of my previous attempt). I actually ended up applying a 1db/oct roll-off after 2Khz from what the blue line shows in the graph to tone it down a bit more. Although the stock woofer extends that high, it's not exactly as smooth as a dedicated tweeter.

The bass to midrange/treble level ratio is also good now too, bass is set to about 5-7db higher.

This time I optimized for both driver and passenger positions. I have a little bit of a dip in the upper bass on the passenger side, but overall it looks the same.

Anyhow, in sum, I'm super happy with the results here. It changed things from 'awful' to 'passable'. And it's hard to argue with the cost.

I might look for a USB power isolation solution though.

If anyone else goes this route and needs filter coefficient files for easy upload, I'm happy to share, just PM.

[kaigoss69]

I think a JBL MS-2 works similarly.

[andrewbr]

This is great work, I was skeptical reading your other thread but looks like you made it happen. I have a spare set of L7 tweeters (4 ohm w/inline capacitor) from the tweeter pods I bought for my own audio project that I'll send you for the cost of shipping to contribute to this adventure, PM me if you want them. You'll still need to find empty pods to install them in though.

[sfinxvc]

Figured another way around the volume issue. Basically REW's PEQ generator doesn't apply any filters with additive gain, this is because it assumes the measurement you're giving it is already at maximum capabilities of the amp + speaker combo, when in reality it's limited by whatever volume we can tolerate while sitting in car and running the sweep. I guess I could dodge out of the car, but I also don't want to push the HU that hard.

Solution? Just apply a +10db offset after taking the measurement. I've yet to do this but realized it'll work.

Originally Posted by andrewbr

This is great work, I was skeptical reading your other thread but looks like you made it happen. I have a spare set of L7 tweeters (4 ohm w/inline capacitor) from the tweeter pods I bought for my own audio project that I'll send you for the cost of shipping to contribute to this adventure, PM me if you want them. You'll still need to find empty pods to install them in though.

Thanks! I'm in. You have PM.