September 2011

Charge Coupled Crossovers
Dramatically improve the sound of inexpensive capacitors.
Article By Jeff Poth

Difficulty Level

   

  I'm going to make an assumption here- your system is compromised by your capacitors. Yep, it's true! Some people go to great extremes to avoid caps in their rig, but they're darned near unavoidable. A typical hi-fi will have as many as a dozen of these reactive little weasels in the “signal path”, wreaking havoc with your wonderful sound. In efforts to minimize the damage, audiophiles have been inserting better/different capacitors into their equipment for decades. There are capacitors out there (some very nice sounding) that cost literally thousands of dollars. See Duelund and Audio Note if you doubt me.

What exactly is a capacitor, and what does it do?

A capacitor has two primary functions in audio equipment. One is as energy storage. Capacitors can be thought of as rechargeable batteries, being able to store significant amounts of energy at whatever required voltage you need- assuming the capacitor is designed for that voltage rating. Batteries are limited chemically by the operation of the cells, where capacitors can operate over a large range of voltages. Additionally, a capacitor is a high pass filter and increases it's impedance at low frequencies, with the specifics depending upon the values of the capacitor and circuit details. This is desirable within the power supply application as they provide a path for unwanted Alternating Current (AC) to shunt to ground while passing the Direct Current (DC) to the circuit and providing energy storage for said DC. Below is a basic power supply schematic showing a typical reservoir/filtering capacitor used for both these purposes.

What we're concerned with in this article is the behavior in use of high-pass filters. A high pass filter is as it sounds- high frequencies pass through. This is what a capacitor allows, so one or several capacitors are usually present in series with midranges and tweeters, even SOMETIMES on woofers. Depending on the size of the capacitor (Typically measured in microfarads, or uF), frequencies below some corner frequency will be attenuated- so the bass is cut out. They are also used in parallel rather than series, to create a low-pass function. This requires careful tuning and a series inductor as if someone were to simply parallel a capacitor with a woofer, to try to cut out the treble, they'd be shorting their amplifier terminals above some corner frequency (again, determined by the value of the cap).

What do capacitors do wrong?

Capacitors are not perfect devices by any stretch. They are prone to dielectric absorption and mechanical resonance, as well as other difficulties like leakage and ESR. Dielectric absorption is managed by using thinner or lower dielectric constant materials, Teflon is used in some of the most expensive and well-regarded capacitors for the same reason it's popular in cables – it is a good high-voltage insulator, and very low absorption. Mechanical resonances are dealt with in a number of ways-tighter windings, mixed materials and other ways. To some extent, the efforts to minimizing these effects in high-end capacitors are effective, but really good caps cost serious money (Cardas caps are a personal fave for non-charged operation). Dielectric absorption and mechanical resonance are what we're going to try to deal with by updating our capacitors. 

This is not my idea...

Our good friends at JBL didn't like using electrolytic capacitors (They're much worse than film capacitors in most ways), but didn't have a lot of other options for sufficient capacitance in small packages. So, they came up with the concept of "Charge Coupling". This is a DC bias inserted into the midpoint of two series capacitors. This could be achieved by using a normal power supply, but JBL, and I, prefer batteries. There's no significant need for current, so the batteries essentially last as long as their shelf life. That means years.

The execution is rather simple from a circuitry standpoint; each terminal uses a resistor of highish value to one spot on either side of a capacitor plus a second capacitor is in series with the first. The capacitance (uF) values of these capacitors are twice that of a non-charge-coupled single, unless you use dissimilar values, in which case you'll need to use (1/C1)+1/(C2)=1/C, where C1 and C2 are the capacitors used and C is the total capacitor value. So take for example a 5uF capacitor. You could replace it with two 10uF in series and still have 5uF, or you could have a 50uF in series with a 5.5uF capacitor and also wind up with approximately 5uF.

The resistor values are high, as the only action the battery is performing is to provide a charge to the capacitors, which will only require trickle current to maintain, once they're charged. The resistors help ensure that you don't wind up accidentally sticking DC on your driver (a very bad thing). JBL used one resistor, I'm paranoid so I used two. Though one is sufficient and that can go between either terminal and either side of the capacitor- the critical part is that in the middle of the two caps, you have a voltage (9V is most common).

I can build that but why would I want to? 

You'd want to because it is claimed to address the two largest problems of capacitors in audio circuits -- dielectric absorption and mechanical resonance/microphony. Dielectric absorption is exactly what it sounds like. A capacitor in simplest form is comprised of conductive plates with an insulator between them. The field created by AC can pass between the plates as the electrical field exists in the space around the conductor, but the DC cannot pass over the insulator, as it does not create the field to pass across the insulation barrier. The capacitance value is determined by the area and distance between these plates. A thicker, more voltage resistant capacitor usually relies upon a thicker layer, so higher voltage capacitors tend to grow rather quickly in size compared to their lower voltage counterparts.

Dielectric absorption is when this insulation layer absorbs some of the AC energy passing across it (some DC can also be absorbed), and can later re-release it. This phenomenon shows itself most at the "Zero crossing", where the voltage across the capacitor passes through 0V on its way from positive voltage to negative, or vice versa. Music has this happen all the time, but a static DC bias pushes the voltage across the plates (conductors) of the capacitor away from this zero crossing. In the 9V case, you would then need a +/- 9V audio signal to ever encounter the zero crossing (that's 18V). That's EXTREMELY loud with almost any loudspeaker, if it's frequently encountering that. If you were running up to 18V often, you could always add more batteries and thus voltage. Additionally, we're very sensitive to zero-crossing phenomena, like "crossover distortion" in amplifiers, or stored energy in a loudspeaker suspension.

Now, you needn't have 18V RMS to encounter the zero crossing, music can be highly dynamic and give peaks far above the average level. But, fortunately for us, minor distortions at these peaks are masked, not only by the loudness of the sound, but also by other distortions associated with large peaks and displacements. If you're concerned about high peak outputs encountering the zero crossing of the capacitor, you can up the voltage with multiple batteries in series (or a higher voltage supply).

By removing the zero crossing, we get a capacitor that mimics one with a much thinner or higher quality dielectric. This leads to a lack of listening fatigue, or at least that's what my ears tell me about charge-coupling. I'm going to assume that the "smearing" effect I've noted is more related to mechanical resonance.

Mechanical resonance is an issue, if you change the spacing of the plates, the characteristics of a signal passing between them would also change, providing modulation of the capacitor value and overlaying the characteristic resonances onto the signal. Applying a voltage is claimed to create a tensioning effect on the windings of a capacitor, holding it in place and suppressing vibration behavior- just as a capacitor with better quality windings, and heavier-duty construction might.

Isn't it a lot of effort for a little thing?! 

It's a simple project, and there's a significant gain to be had- You can build a higher-end capacitor than you can afford by simply using quality inexpensive caps and charge coupling them. Commercial Mylar and poly caps are available very inexpensively from surplus houses like Apexjr and Madisound seems to have an endless supply of handy little 10uF yellow caps on their clearance page. Sure you need to buy twice as much, but if you shop carefully, you can come in quite a bit less expensive than even the cheapest "audio grade" cap from Solen, Erse, Clarity Caps, or whomever. So not only can you make a capacitor that will significantly outperform most models, you can do it in values that aren't necessarily available and would be extremely expensive. At the time of this writing, a single, not terribly good sounding, Solen Fastcap of 40uF is going for $16.28 at Parts Express. That's what it cost me to build a pair of superior audio capacitors with charge-coupling. So a 50% savings and better sound. Not too shabby!

There are a lot of locations for capacitors within a speaker crossover. Other locations in most systems are less appropriate as they're usually already biased with DC and the components are using them as DC blocking capacitors rather than AC filters, so in those cases they're effectively already charge-coupled. Below is a typical, simple crossover three way speaker – there are four caps in there! You can share a single supply/battery between all of them, since they're isolated by resistors.

So, that's it. I've found charge-coupling to dramatically improve the sound of inexpensive capacitors, and it's not very difficult or expensive to implement, so I can firmly recommend charge-coupling as a tweak for our people who know which end of a soldering iron is hot. Depending on the speaker and system it may require relocating crossovers outboard, but that's on a case-by-case basis. There are significant improvements in clarity, imaging, and listenability to be had, get to it kiddies!

Fan Mail
ADDED: Going 'back' to the circuit source whereas I am only familiar with the "tapping between the capacitors and going to battery terminal through the high M-ohm resistor. Image at this link is the perfect example by someone very skilled in its implementation. I would hate to see your readers not given the proper design to use themselves. This design also gives information as to how to select the values of your capacitors from your circuit being customized. (double the original value and pair them). 

I hope this information is in line with your article.

Kind Regards,
Steve Griffin