General considerations for passive crossover components:

1. POWER   Power is listed as the most important because it is the most likely to degrade the signal or components if not accounted for in a suitable manner. The most obvious consideration is that the continuous power rating of the passive components must be at least as high as the average power of the signal passing through them under maximum volume conditions.

   But more than this is required. Depending on the type of source material (home theater, classical music, pop music, rock, hip-hop, etc.) the signal will contain peaks which far exceed the average power. In highly processed and compressed popular music, this will typically be in the range of 8-10 dB. In classical music and some home theater (HT) applications, this can be as high as 12-20 dB! How the passive components handle large spikes of power is crucial to how well they will perform under actual operating conditions.

Above, I also alluded to the "power passing through them". This is an important distinction which can help you manage the costs of your crossover. There are two aspects to this consideration:

1. POWER SPECTRUM   This is extremely important in multi way loudspeaker systems. Each driver will be passed only frequencies within its operating range. Since most of the power in a variety of typical audio signals is contained with the frequencies between 300 Hz and 4 kHz, signals outside this range passed to drivers will not require the same power rating as signals within this range. The most significant implication of this consideration is that passive components in the signal path for a tweeter section will require power ratings only a fraction of those required for woofers and midrange drivers.

   For non-musical signals, this is a bit more complicated. HT systems will often have to reproduce extremely low frequency sound effects. This isn't as serious a problem in systems which have a separate Low Frequency Effects (LFE, or subwoofer) channel. Such systems will automatically keep the effects out of the main speakers. However, for the rest of us, this can be a problem.

   Both the HT and music issues can be resolved by becoming familiar with the spectrum(s) of the source material to which we listen. As a guide for music, please refer to the musical instrument frequency range chart at Solomon's Music Theory & Composition Resources. Worthy of note is that most music, even bass guitars and drum kits have little power much below 30-40 Hz. For HT, still most source material doesn't go really low unless you're watching an action film with things like helicopter blade or naval engine room noises.

2. CROSSOVER TOPOLOGY   Not all passive components are in the signal path. Components wired electrically in series with the drivers are in the signal path. Components in shunt networks wired in parallel with the drivers will only need to carry whatever portion of the signal is shunted through that parallel network.

   Confused? It's pretty simple… There's no such thing as a perfect or ideal component. In the real world, the shortcomings will consist of either non-linearities or noise. Non-linearity is primarily an issue with inductors, but not a significant issue with good inductors. Noise is a major issue with capacitors and resistors. In electronic network theory, the noise and other problems with capacitors and/or resistors (e.g. poor internal connections which act as diodes) are all modeled as parasitic voltage sources.

   Series circuits work as voltage dividers. Parallel circuits act as current dividers. If you avoid non-linearities by careful component selection, you're left with anomalies which act as voltage sources. Since a loudspeaker is a low impedance system, such components will have more audible effects when in series with the driver than when in parallel.

2. IMPEDANCE   Related to power, impedance is also a crucial factor in passive component selection. It's always critical to remember that since loudspeaker systems are low impedance systems, the signal currents involved will be large and potentially destructive of the passive components.
3. FREQUENCY RANGE   Passive components often exhibit different electrical characteristics at different frequencies. In an ideal world they wouldn't, but we don't live in that particular world. Fortunately, the operating frequency range of audio applications is pretty benign. Low frequencies are rarely a problem, so we should concern ourselves with changes to electrical characteristics at higher frequencies. Still, for CD's, the upper signal limit is 22 kHz, twice that for higher performance CD formats such as SACD, and somewhere between 30-100 kHz for a perfectly recorded virgin LP (once played, the high frequency content is quickly degraded by the mechanical action of the stylus on the grooves).
4. COMPONENT SPECIFICS   The remainder of this appendix will deal with the details of characteristics peculiar to each class of passive components. These relate to materials technology and gross large-signal electrical characteristics.

   But first a final note about tolerances… How much do you really need? Generally, 5% is more than good enough. The tolerances of speaker driver parameters is usually not much better than 10%. Even if you're hand-crafting a system from individually measured and selected matched sets of drivers, the stability of large signal parameters with changes in source material, age, and environmental conditions rarely justifies better than 5% absolute tolerance. What close tolerances buy you is better matching and better worst-case network performance.