Saturday, September 14, 2012
More Interior / Trim Parts
Over the last couple of weeks I picked up more parts for the upcoming interior / cosmetics overhaul. These included:
- Roof ledge moldings. I ordered two of these moldings and they came in last week but my inspection revealed that one molding had been bent at some point in shipping. I wasn't about to pay $90 for a damaged part so I asked the parts guys to order another and that came in today. When I saw the original box the second trim piece came in I realized why they cost $90. BMW has its own logistics of course but shipping the huge box it came in by any traditional carrier via ground service probably would have cost $90 alone. So if you're buying these expensive parts you may take some comfort in that the cost is in the shipping...not the part itself.
- Rear quarter window front trim pieces. At first I thought they were some kind of fiberglass because the paint had worn off my original parts to expose a fiber-like substrate but the new ones are clearly plastic and adhere to the window with double-sided tape.
- OE fabric floormats. I have been looking for good quality aftermarket floor mats that don't look like tacky living room carpet but to no avail so I punted for now and went for another set of OE mats despite the tendency of the backing to fail.
- Side skirt rubber strips that help seal the side skirts to the car body. While getting out of the car one day my foot hooked the strip and tore it. This was not the first time this had happened so the skirt itself has been somewhat loose. Since I need to remove the skirts to fix the seal I bought all new fasteners and sealing grommets too.
- Door sill moldings. My originals are in great shape, except for the rubber end caps which have completely disintegrated and look like shit.
Next month I'll be buying the felt seals for both front doors. After that I'll order new leather and should be ready to start the project. Cold weather is looming, however, so this project may get postponed until next year.
Audio Research Raises More Questions
My original plan was to buy a four channel amplifier that supported balanced inputs and a set of passive crossovers to feed my existing drivers. I ultimately determined that the crossovers I needed for my application ran nearly $450 for all four channels. Always looking to save money, that led me to research how to build my own passive crossovers. At first glance the idea of building a couple of LC filters seems pretty straightforward and from a parts and wiring perspective it more or less is, but the devil, as they say, is in the details.
Passive crossovers consist of electronic components including capacitors, inductors and resistors that react to the power delivered by the amplifier output terminals. While these devices can be leveraged for productive purposes, like filtering out unwanted frequencies for each driver, they also produce many undesirable effects including nonlinear response, power consumption, phase shifts and a surprisingly nasty tendency to interact with one another in ways difficult to predict without a bunch of math.
As a consequence the mantra in crossover design is to minimize the number of components used and that typically means limiting oneself to first order (6db/octave) or worst case, 2nd order (12db/octave) filters. If that seems like an intuitive or completely reasonable idea, consider that some drivers can be damaged if they receive frequency content outside of their design specifications so, depending on how far the crossover frequency is from the edge of the acceptable frequency range, using a first order filter may not be sufficient to protect the driver. Welcome to the world of compromises that is passive crossover design.
Crossover points are selected based on the specific combination of drivers in a given application. The kicker is that these frequencies can only be approximated using the data provided by speaker driver manufacturers. In reality, each driver must be tested individually to verify its frequency response including resonant frequency, impedance (which is frequency dependent) and in the case of woofers in particular, the equivalent voice coil inductance (Le) that also changes with frequency. If you don't compensate for these factors using real world measurements, it's easy to build a crossover that quite honestly sounds like shit.
Proper crossover engineering therefore requires test equipment including a measurement microphone (a calibrated version of the Behringer ECM8000 is a current favorite), microphone preamp, a computer, and some potentially expensive software...though as I discovered some free packages will get the job done at the expense of a steep learning curve. It was around this point that I began to understand why those commercial crossovers cost so much.
I spent the better part of a week of evenings and lunch hours reading about the science and techniques of crossover design before I convinced myself that it would be possible (and probably a lot of fun) to build my own passive crossovers. Late that week, however, I had lunch with a colleague who, in addition to having considerable experience as a professional musician and by extension pro audio, is a EE who has built his share of custom speakers and crossovers. His response to my plans? "Go active and bi-amp. It solves a lot of problems you can't fix properly with any passive crossover. You won't regret the decision."
Research into that topic confirmed much of what I already knew, but my new-found knowledge of passive crossovers helped to put things in perspective. Indeed, the best place to do filtering is in the small signal domain, where simple operational amplifiers in conjunction with physically small and inexpensive passive components can be combined to create the necessary filters without regard to the interactions of these components with the speakers. Hint: they don't call a voice coil a "coil" for nothing: from the perspective of a passive filter network the voice coil is just another inductor with all the nasty nonlinear attributes pertaining thereto. It should now be clear why crossovers must be tuned to the specific drivers in use, and why using any off-the-shelf crossover (regardless of price or quality) with the wrong drivers will likely sound worse than a bad DIY hack.
Active crossovers avoid many of the pitfalls of passive crossovers and the consequences of working in the large signal domain. Specifically they:
- Eliminate the need to build potentially bulky and heavy passive crossovers due to the requirement to specify inductors and other components with low equivalent series resistance (ESR) and sufficient power handling. This is particularly an issue when specifying inductors for a low pass crossover below 500 Hz. Do the math sometime. It's enlightening.
- Increase the overall efficiency of the amplifier, since it does not send current out a speaker terminal only to be dissipated as heat in the various components of the crossover. Each channel amplifies only the specific frequency spectrum intended for a particular driver and the amp delivers that current directly to the driver.
- Allow quick and easy tweaking of the filter characteristics (crossover point and slope) in response to live measurements. A passive crossover would need to be brought back to the bench each time, and changing the slope could require a complete redesign.
- Preserve the amplifier's high damping factor, particularly near driver resonance, so it can more tightly control movement of the driver motor for improved sound quality.
- Prevent back-EMF (generated by the mid-bass / woofer in particular) from inducing unwanted signals in the other drivers that would otherwise be interconnected through a passive crossover network.
So if active crossovers are so great, why isn't everyone bi-amping? Simple. Cost and space. The problem is worse with three way systems, assuming you want to separately power the midrange and tweeter. Most people, however, just build a simple passive low pass filter for the midrange (thus completing a band pass filter) and a high pass for the tweeter because of diminishing returns. The calculations ultimately reveal it is possible to use simpler low-order filters with fewer and smaller components at the higher frequencies. Should I decide to bi-amp I plan to do exactly this.
While the thought of dropping serious coin on a couple SE4100's is enough to turn me off the idea entirely, my more immediate problem is space. I'm limited to about 32" between the speakers on the rear deck, and to only about 10" front to back because the rear deck and related structure is curved so I can't efficiently use the space. The SE4100 is 16x8" but like all the other Arc Amps unfortunately terminates its connections on the ends of the chassis so they won't fit end to end in the space available. Regardless of cost, the SE4100 is therefore out of the running in a bi-amp configuration simply because it won't fit. For this reason I went in search of other amps that support balanced inputs.
I quickly found the Soundstream Reference REF4.400. At 16x10 it's not exactly small but it terminates connections on the rear of the unit so I can fit them end to end. The problem is that while Soundstream was once known as a high end amp manufacturer of similar caliber to Zapco and Arc, changes to the design and ownership since that time make me wonder if the new reference amps are indeed as good as the old. I believe the old amps were Class AB like the SE4100 while the new versions are Class G, but that's not necessarily a bad thing because at moderate listening levels (what I do most of the time), Class G is more efficient. On paper they look like good quality amps, and if the price ($280 each) is to be believed, a steal too. To their credit, they actually responded to a couple questions via email in about 4 hours. They clearly punted with one of their answers but at least they responded. I emailed Arc days earlier and I'm still waiting for their response.
If I had to make the call right now I'd probably go with the bi-amp Soundstream solution, but I know I can make a passive crossover solution work and if it allows me to gain additional knowledge in audio engineering -- despite its obvious shortcomings -- so much the better.
Many thanks to Rod Elliott for his excellent explanations of passive crossover design techniques and his comparision of active vs. passive crossovers. The articles assume some knowledge of electronics but are still an easy read.
Mileage: 225075, Parts: $550