[emailtim]

This is more of a question of passive versus active discussion, than a capacitor Brand X versus Brand Y.

Is the perfect crossover cap a short piece of copper wire, or is there something else involved assuming an active XO handles the point/slope/phase issues ?

The only thing I can think of is protection against amp turn on/off thumps and possible coloration.

With that in mind, what is the perfect XO cap ?

[Speedskater]

Please start over and completely rephrase the question.
There are lots of caps in a passive crossover and lots of different caps in an active crossover and then there are DSP crossovers and of course there are coupling caps.

[emailtim]

I don't quite know how to better rephrase the question other than what would a perfect crossover cap strive to approach (i.e. a short piece of copper wire) ? 

A perfect pre-amp has been described as "straight wire with gain".

A perfect step response would look like the top plot, but in reality, we don't get it.



I don't want to get into the "what-if there are other caps upstream" discussions.  For discussion, assume you are using DSP XO's (of the equivalent quality or better of the recording/mixing software) with a multi-channel DAC, so the closest caps are in the amps directly attached to the drivers.  With a passive XO, you already have caps in the amps and DAC so they are a wash in this discussion.

Other than providing a high pass filter with proper point/slope/phase does the cap strive to do anything else ???

[S Clark]

A perfect crossover cap shape a FR curve without changing the tone... or that's my take.  But the limits of knowledge on the subject are pretty narrow.

[emailtim]

A perfect crossover cap shape a FR curve without changing the tone... or that's my take.  But the limits of knowledge on the subject are pretty narrow.

So basically a perfect cap would approach a dead short through the pass band (while disallowing DC to pass).

[Speedskater]

So basically a perfect cap would approach a dead short through the pass band (while disallowing DC to pass).

Not that's not how caps work.
A perfect cap would be all capacitive reactance, with no inductive reactance, no leakage resistance and an end-to-end resistance of zero Ohms.

[rollo]

  That would be a Duelund Cap in my experience. Great specs fabulous sound. Duelund resistors as well. Parts Connection is having a sale. Not a Duelund dealer.


charles

[HAL]

A perfect capacitor would be open at DC and a short at high frequencies.  It is not a realizable device, but a mathematical function to model a circuit.  A real capacitor has capacitive, resistive and inductive parasitic elements that keep it from being a perfect capacitor.  The same is true for resistors and inductors.

Since a DSP crossover is a mathematical process, it gets much closer to the original function.  There are still mathematical processes in the digital domain that do not let it be a perfect filter, as thing like phase folding occurs.  Luckily with DSP crossovers, the folding process is close to the Fs/2 frequency.  By having higher sampling frequencies, the folding process is now outside the audio band with sample rates at 192KHz as Fs/2 is now 96KHz.  The higher the precision of the math process (more bit length) the lower the artifacts (or noise floor) becomes.

Also with passive components and their value tolerances, matching of parts from channel to channel effects imaging and frequency response variation.  This becomes critical in higher order filters to make them work correctly.

With new DSP's on the market using 64bit floating point math, the errors are vanishingly small with correct filter topologies. 

I gave up on passive filters long ago with new generation DSP processors and better DAC's for crossovers.

[emailtim]

A perfect capacitor would be open at DC and a short at high frequencies.  It is not a realizable device, but a mathematical function to model a circuit.  A real capacitor has capacitive, resistive and inductive parasitic elements that keep it from being a perfect capacitor.  The same is true for resistors and inductors.

Since a DSP crossover is a mathematical process, it gets much closer to the original function.  There are still mathematical processes in the digital domain that do not let it be a perfect filter, as thing like phase folding occurs.  Luckily with DSP crossovers, the folding process is close to the Fs/2 frequency.  By having higher sampling frequencies, the folding process is now outside the audio band with sample rates at 192KHz as Fs/2 is now 96KHz.  The higher the precision of the math process (more bit length) the lower the artifacts (or noise floor) becomes.

Also with passive components and their value tolerances, matching of parts from channel to channel effects imaging and frequency response variation.  This becomes critical in higher order filters to make them work correctly.

With new DSP's on the market using 64bit floating point math, the errors are vanishingly small with correct filter topologies. 

I gave up on passive filters long ago with new generation DSP processors and better DAC's for crossovers.

Great explanation.  Thank you very much.

[Speedskater]

Note that neither of the images in post  reply no#2 have anything to do with capacitors (per se).

[emailtim]

Note that neither of the images in post no#2 have anything to do with capacitors (per se).

Maybe because there are no images in post #2 ???   

If you are referring to post #3, the images were an example of a perfect DIRAC DELTA response and reality of a IR STEP Response which I was seeking an similar analogy for a perfect capacitor.  I did not imply that the 2 plots were that of a cap's response.  Sorry if I did not make that clear.

[Speedskater]

Well duh it's reply not post.
a perfect cap's response is one thing but it doen's look anything like either of those images.

[emailtim]

Well duh it's reply not post.
a perfect cap's response is one thing but it doen's look anything like either of those images.

You need to tone it down a bit.  Your responses are adding nothing to this thread other than being an antagonist.   

You have not answered the original question even though it was repeated as per your request and then you complain about the response to your request. 

HAL was the only one kind enough to give a useful, helpful, factual, educated and thorough response.  Be more like HAL.

[rollo]

A perfect capacitor would be open at DC and a short at high frequencies.  It is not a realizable device, but a mathematical function to model a circuit.  A real capacitor has capacitive, resistive and inductive parasitic elements that keep it from being a perfect capacitor.  The same is true for resistors and inductors.

Since a DSP crossover is a mathematical process, it gets much closer to the original function.  There are still mathematical processes in the digital domain that do not let it be a perfect filter, as thing like phase folding occurs.  Luckily with DSP crossovers, the folding process is close to the Fs/2 frequency.  By having higher sampling frequencies, the folding process is now outside the audio band with sample rates at 192KHz as Fs/2 is now 96KHz.  The higher the precision of the math process (more bit length) the lower the artifacts (or noise floor) becomes.

Also with passive components and their value tolerances, matching of parts from channel to channel effects imaging and frequency response variation.  This becomes critical in higher order filters to make them work correctly.

With new DSP's on the market using 64bit floating point math, the errors are vanishingly small with correct filter topologies. 

I gave up on passive filters long ago with new generation DSP processors and better DAC's for crossovers.

  Excellent explanation Professor. Thanks. I prefer no crossover at all. Simple. Single driver outboard Sub.

charles