Opinion's: Crossover Parts matter?

[Matthew J Poes]

I'm looking to get some discussion and opinions going on this topic. I'm still debating a bit what I want to do with the topic, but have always found it an interesting topic. I'm looking for opinions that run the gamut. Everything from, "I don't think parts make a wit of difference as long as the design is good", through, "I think the quality and design of parts is critical and can hear differences between one brand of film capacitor and another."

I was thinking more about this topic because a friend of mine asked me why a subwoofer (or any woofer) with high inductance can have more distortion and provide worse sound than one with lower inductance. He asked a great question, why don't we talk about this with speaker crossovers, isn't inductance the same here? The answer is yes and no, but I don't want to get into that here. I decided to do a little reading in my text's on just how audible and measurable these issues can be. It took me down a long and lonely road of theory and hard science, but little on actual perceptions. It seems that our perceptions have been left to us for now.

What do you think? Does that iron core inductor in your speaker keep you up at night? What about that electrolytic capacitor in series with your tweeter? Have you played around with different components and noticed a difference? Ever bought an upgraded speaker that had a higher quality crossover and were surprised by the range of differences? What did you hear? Do you think this is all a bunch of non-sense? If so, please share as well.

 

[tesseract]

Better passives can (and should, but not always) have tighter specs, so the outcome should be less production variation.

One can get by saving a bit with cheaper passives. Iron core is OK at low frequencies, electrolytics are fine when new but not really a great choice over the long haul.

 

[Matthew J Poes]

Better passives can (and should, but not always) have tighter specs, so the outcome should be less production variation.

One can get by saving a bit with cheaper passives. Iron core is OK at low frequencies, electrolytics are fine when new but not really a great choice over the long haul.

Both of these parts also suffer from hysteresis and non-linearity which can cause audible distortion. That is why I mentioned them. Electrolytic capacitors aren't too bad if they are bypassed by a small value film capacitor. Iron core inductors in the bass section and sized appropriately are ok, air core is still better if ESR can be kept low enough. A number of folks claim to readily hear the non-linearity in iron core inductors as they saturate and distort. I'll admit I've not played around with them enough to be able to claim audibility for myself or even describe the sound. The distortion is easily measured, especially IMD.

Electrolytic caps have non-linear resistance (and higher ESR overall) than do film caps. Like I said earlier, bypassing them helps avoid this, but....a true film cap is still best overall.

p.s. I'm glad someone finally found this thread! Hopefully more will post.

 

[ddude003]

In my opinion all things in the signal chain from source to delivery matter... Crossovers matter and quality of parts that comprise a crossover matters... Capacitors play a large part here... Just for fun, have you ever seen or heard what Kenrick Sound does with JBL speakers? Check out his external crossovers...

 

[Matthew J Poes]

In my opinion all things in the signal chain from source to delivery matter... Crossovers matter and quality of parts that comprise a crossover matters... Capacitors play a large part here... Just for fun, have you ever seen or heard what Kenrick Sound does with JBL speakers? Check out his external crossovers...

I have not, just took a look. Those crossovers are pretty over the top.

I've been building crossovers and playing around with parts for well over a decade now and have tried a ton of parts. I've never found it worth it to buy some of the crazy over the top capacitors that some like, such as the Teflon Film and Foil, or copper foil inductors, but I do use premium parts. On the other hand, the difference between a standard metalized film cap and an electrolytic is quite noticeable. Cheap parts also tend to fail more often, which is never good.

When I built my speakers I worked with Dr. Geddes to enhance the crossover. Clarity Cap was a newly imported company and had just released the SA series. I was given a ton of SA caps to try. I also obtained a large surplus of MIT Polystyrene caps in a range of values. In my speakers most of the caps are Solen, but in critical series locations, I used the SA cap bypassed by a MIT Polystyrene. I had them laying around and it cost me nothing. I also use a treble tilt adjustment RC filters on the CD and bypass the resistor with an MIT cap. I tried a few other parts in this location and the main cap location even using Teflon metalized at one point. I couldn't hear a difference. However during early testing I had in place a Mylar dipped cap in the bypass locations and I had an electrolytic cap in the main cap location. I'm not totally sure what happened, but a catastrophic failure of one part caused the electrolytic to "give up the smoke", the mylar to burn up, and the amp to develop a hum. Many parts replacements later (including a new coil on the CD) and all was well, but I decided to never test with those kinds of parts again.

 

[AudiocRaver]

I voted for HUGE DIFFERENCE, although it probably falls between Huge and BARELY NOTICEABLE, depending on a multitude of factors.

 

[Todd Anderson]

An iron core inductor in a speaker has never kept me up... but I can only assume that quality crossover parts play a huge role in perceived sound quality... active speaker designs with dual amps really make for an amazing show... and if I'm not mistaken, that design eliminates the passive crossover. If that's correct, then color me a giant YES

 

[Matthew J Poes]

An iron core inductor in a speaker has never kept me up... but I can only assume that quality crossover parts play a huge role in perceived sound quality... active speaker designs with dual amps really make for an amazing show... and if I'm not mistaken, that design eliminates the passive crossover. If that's correct, then color me a giant YES

Normally an active system doesn’t have any crossover parts. Sometimes they will include a cap on the tweeter and resistors for protection and amp loading. Some cheaper powered speaker systems use passive crossovers.

My own opinion is that a good bit of the sound quality difference heard is the ability to apply really precise eq. Often the response of active systems is super smooth/flat.

A lot of the claims about improved dynamics are more fluff and less substance. It’s not an absolute truth, more a truth under certain conditions and only to a point.

 

[tesseract]

A passive speaker with DSP room correction added can get pretty close in performance to an active speaker, even when both are corrected.

 

[Matthew J Poes]

A passive speaker with DSP room correction added can get pretty close in performance to an active speaker, even when both are corrected.

It can, but it depends a lot on the speaker. Plenty of speakers have such a poor polar response that Eq can’t really correct the polar response issues and this things like the shift in tonal balance as you move your head or th tonal shift in the reflections isn’t able to be corrected. On the other hand, and active speaker which can eq the drivers individually may be able to do a better job knocking down the cone breakup mode causing an off-axis anomaly without impacting the flatness of the overall response (global eq can’t do that).

In other words a little active eq applies to a really top notch passive speaker is likely just as good as that same speaker if it was active. A crummy speaker may be able to be more readily improved in active form than in passive form.

 

[AudiocRaver]

I am going to modify my vote from Very Much to A Little. I believe it CAN make a big difference, but in most cases it probably only matters a little.

Matt, won't a speaker with poor polar response sound worse in a live room?

 

[Matthew J Poes]

I am going to modify my vote from Very Much to A Little. I believe it CAN make a big difference, but in most cases it probably only matters a little.

Matt, won't a speaker with poor polar response sound worse in a live room?

I think so! My argument is that the tonal balance of the reflections are different. In a lively room those reflections are stronger.

I’m a discussion with James Larson, a great lover of Toole’s thinking, we actually debated this issue a bit. Toole argues that the first reflection points add to the perception of sound in a positive way. It sounds more real. He also further argues that absorbers don’t fully absorb and don’t have a flat absorption curve at the angle of incident they are used, therefor they change the tonal balance of the reflection. That this is bad. I argued that for a speaker with a flat and elevated DI, it probably is desirable to leave the reflections intact. This is a better system overall and a better option in this specific case. However, most speakers do not have a flat DI, and many have a fairly uneven polar response. I figure, in this scenario, the absorber can’t make things any worse, and they might at least absorb more of the reflected sound, this improving the sound of that specific system. We can argue that the system is flawed to begin with (this was what James argued and what Toole probably would argue). I was just taking a practical approach. If you have such speakers and you are intent on using them, they may sound better with the first reflection point absorbed.

Going back to the point at hand, it is preferable to design a speaker that has a smooth change in response off-axis. My own beliefs have been heavily shaped by Geddes, and he argues that a flat raised DI that extends down to below 800hz or so is desirable (ideally more like 200hz if we are dreaming). The DI should be quite elevated, like 6-12dB, and that number simply depends on the dispersion angle. What this does is make it so the speakers response doesn’t change tonally with angle, the response shape stays the same, but it gets quieter. It allows the speaker to naturally do what we would want first reflection absorbers to do but don’t. Ensuring the response shape remains flat with angle requires a few things. First is that the drivers need to be somewhat directional. Call it intentional beaming if you wish, but the response should fall with angle naturally. The other is that at the crossover there must be what I term directivity matching. That is, the directivity of the two drivers must be the same at and around the crossover point. A third issue would be that the midbass needs to not mess with the response of the tweeter. That is, any sound it produces well out of band (past the crossover by an octave or more) needs to be well down. The reason is because the tweeter likely has a much narrower dispersion by this point. Say it’s 3khz. A midbass driver crossed at 1khz to a tweeter will have a much different dispersion at 3khz than the tweeter. As a result, any large spike in the response, such as the breakup mode, can show up as a ridge off axis (so the listening window might be perfect, but the first reflection angle might be a mess). Eq can fix this. You can insert a notch. Similarly, while dispersion might be similar at the crossover, slight response shape differences or toughness can show up more off-axis, but eq can correct this allowing a smoother transition. This won’t show up on axis, but might off axis. Again, typically because there might be slight differences in the dispersion at that frequency.

Why this matters in active vs passive is that you almost always see response shaping in good speakers. Basically there is crossovers in the crossover and there is eq. Passive eq can be accomplished lots of ways, but it’s always lossy. It’s selectively shunting power to ground at certain frequencies, frequently through a resistor. Imagine a speaker with 3 bands or 4 bands or even 10 bands of passive eq. The power wasted adds up. Further, the crossover gets really complicated (10 bands is 30 large extra elements). The accuracy of the eq is also limited. Switch to active and suddenly the number of response shaping bands is practically unlimited with no meaningful waste. This allows for a much flatter response both on-axis and off. We argue how audible this improved flatness is, but it certainly doesn’t hurt to be flatter.

There are a fair number of arguements pro-active that don’t really hold up to scrutiny, or even if they do, are of questionable value. Things like a reduction in IMD, time alignment, higher output, etc. the losses in basic crossovers are often exaggerated. However, the improved precision and flexibility around eq is pretty valuable. Modern dsp let’s you build in all sorts of features that just make a speaker better. I’m all for it, but I get a little ruffled when people argue clear superiority.

I know Wayne, I digressed a bit, but to me this is all related stuff.

 

[tesseract]

It can, but it depends a lot on the speaker. Plenty of speakers have such a poor polar response that Eq can’t really correct the polar response issues and this things like the shift in tonal balance as you move your head or th tonal shift in the reflections isn’t able to be corrected. On the other hand, and active speaker which can eq the drivers individually may be able to do a better job knocking down the cone breakup mode causing an off-axis anomaly without impacting the flatness of the overall response (global eq can’t do that).

In other words a little active eq applies to a really top notch passive speaker is likely just as good as that same speaker if it was active. A crummy speaker may be able to be more readily improved in active form than in passive form.

I considered polar (off-axis) response when making my claim, and am glad you caught onto that. Now we can further the conversation.

How about a DSP-enhanced, low efficiency active speaker using cones and domes vs. a DSP-enhanced, high efficiency passive directivity design?

That is a game changer and not only brings active vs. passive closer to parity, but can give the nod to the more efficient/lower distortion passive design.

 

[Matthew J Poes]

I considered polar (off-axis) response when making my claim, and am glad you caught onto that. Now we can further the conversation.

How about a DSP-enhanced, low efficiency active speaker using cones and domes vs. a DSP-enhanced, high efficiency passive directivity design?

That is a game changer and not only brings active vs. passive closer to parity, but can give the nod to the more efficient/lower distortion passive design.

Well regardless of the use of passive or active crossovers, I happen to think that a controlled directivity design is better. I think what you are really referring to is a speaker whose dispersion pattern is not controlled in any way. I don't like calling these wide angle vs narrow because I don't think that accurately captures the difference. Most speakers using standard small conde drivers and dome tweeters have a dispersion that is out to nearly 180 degrees up to about 3-5khz, then you often see (in the best designs) a narrowing of the dispersion until the tweeter is "beaming" meaning that its dispersion is VERY narrow. Often this is between 10 and 15khz. Where it beams depends on its design and diameter. That means there is a fairly jarring and significant change in directivity with angle that takes place. In the worst of these designs, we see what is sometimes referred to as wastebanding, or a narrowing of the dispersion in the midrange that then returns to normal in the treble frequencies. This comes from poor directivity matching between the midbass and tweeter. While some might argue this effect is unimportant if it is not in the listening window, I think it's a reflection of terrible speaker design, and causes a colored reflection window that impacts the speakers sound. This wastebanding is common, even amongst very expensive speakers, even many loved on forums. While some may love them, you will never see me singing their praises.

With a Controlled directivity design, we actually see a more gradual "controlled" shift in level as you go off axis. The directivity often allows for a more even response at a specific angle. You still see beaming in most designs at some point, often the same 10-15khz or so. In these designs, you do often see (by design) a drastic shift in level as low as 500hz (or below in some specialized designs) as you move across the 90 degrees from center. Within the listening axis the level changes as well (whereas with the above designs, it may be largely unchanged).

When you add the advantages of high efficiency, namely that the speaker can produce much higher output with less distortion or compression, allowing for realistic dynamic reproduction, you get better sound. I think the knock against such designs has been that historically waveguides and horns produced a signature sound that was not to many peoples liking. With the breakthroughs that Geddes made, along with Harman and (I hate to admit this) Klipsch, that sound is either a complete non-issue or a minor issue. The remaining issue was often a less flat response, but if you can eq the individual drivers (regardless of the form of x-over used, passive or active) you get a very flat response and ideal polar response. High-quality speakers using passive crossovers with no form of EQ often have less than 1 dB of loss through them, often MUCH less.

All that to say, I clearly believe that a passive CD speaker that is well designed (like a better SEOS design, Geddes Design, JTR, or JBL M2 with passive crossover) will always sound better than even the best non-CD designs. My requirements for good sound reproduction simply can't be achieved other ways. That being said, since most people don't want something the size of a JBL M2 in their house, and given how difficult it is to control dispersion and have high efficiency in a compact package, I can at least accept smaller compromised speakers, and I believe that active versions have advantages. More so, I believe that modern active technology allows for much better speakers for less money. That is, it is now cheaper to make a really good active speaker than a really good passive speaker. DSP is cheap, crossovers are expensive.