Quote:
Originally Posted by santacore  JayDee, Could you explain 6 layer circuit boards? I've always wondered what that meant and how it differs from a traditional board. Thanks.
|
Sure Sanoacore; excellent question. This is a very complex topic, and I only have time to give a brief overview. If you still have a Hungry Man appetite after my Happy Meal explanation below, let me know and I'll do my best to satisfy.
In nearly any circuit, the most commonly made connections are for power and ground. Most in the audio world have heard about the importance of power quality, but just as important is the means of delivering that power to the point of load, and then dealing with the return currents. Remember, currents always flow in loops, and what happens to the return current is just as important as how the current got there in the first place. So, you might be in possession of the highest performing regulator of all time, but unless you can get the output of that regulator to the circuits it supplies without harm, you've lost much of the benefit the regulator might inherently possess. And when I say "regulator" here, I could just as easily be talking about a battery, supercapacitor, or whatever power delivery device you can think of.
Most circuit boards, even within the high-end audio world, are either single or double-sided; that is, there are only one or two layers of etched copper. This is so for one reason: cost, both in terms of fabrication as well as simplicity and design time. Single or double-sided PCBs are fine if the circuit laid out on the board is a low-speed, simple circuit. But if the circuit complexity is high, involves high speeds (fast risetime signals), or you want very high performance from the device, then a double sided PCB is inferior in every respect to a multi-layer. BTW, although a double-sided board is literally a multiple layer design, the PCB industry by convention means four or more layers when using the term "multi-layer"
Physically speaking, multi-layer boards are a laminated sandwich of four or more copper layers, each being separated by some dielectric material. The inner board layers can be used to route signals, deliver power, supply a ground reference, or some combination of these. There is infinite variability in this design practice, and though there are some general design rules, this is really one area where there can be a great deal of product differentiation. It's often underappreciated, because so much of it is hidden or non-obvious. I think this is one of the reasons that many in this hobby tend to inordinately fixate on boutique parts, when the effects of the circuit board are often just as important, or even more so. Especially in the case of a multi-layer board, you can't see what's going on in the inner layers, and so they're often left out of discussion or consideration. Out of sight, out of mind.
I think it's difficult to overstate the importance of PCB design in high-end audio. I consider a PCB to be a component - that's right, a component in the same sense that a capacitor or a resistor is a component, because a PCB is both of these and more. There must always be a means of connecting parts to one another, and whether this is done with point-to-point wiring or a circuit board, the "hidden schematic" comes into play. This means that there are
always parasitic components that are implicitly formed by the mere act of connecting everything together, regardless of how it is accomplished. If you understand how and why these components are formed, a circuit board can be a powerful means of controlling them, or even making them work to your advantage. A multi-layer board provides a designer dramatically higher flexibility in making a circuit do what you want it to do. It also costs drastically more and requires more design time and skill - the cost goes up exponentially as you get into the higher layer counts.
The simplest and most obvious advantages of a multi-layer PCB is that you have more flexibility and ease in circuit routing. If you have a very dense circuit, more copper layers yields many more options for interconnections. This is one of the reasons something like a computer motherboard is always a multi-layer design. Some of them might have as many as twelve layers.
The effect of a multi-layer board on the signal path must also be considered. In the Diverter, for instance, because I don't have to make room on the top or bottom sides of the board for power or ground traces (both are addressed with planes), I can make the signal paths much more short and direct. This allows me to avoid the use of jumpers or wiring, and keep everything tight and compact. In fact, there is only one piece of wire in a Diverter - a very short jumper up to the output pin of the BNC connector, and this was unavoidable.
The other, more interesting advantage of a multi-layer board is that, properly executed, it can function as a very sophisticated power delivery system, for these reasons:
1.) When it comes to supplying parts of a circuit with power, it is generally desirable to maximize capacitance and minimize inductance at the point of load. With a multi-layer board, the designer has the option of using power planes to accomplish this goal, meaning that an entire layer (or significant portion thereof) is devoted to nothing but delivering power to parts that are connected directly to it. I use this technique extensively in the Diverter; nearly everything is powered directly from power planes. So instead of a spider network of skinny power traces snaking around the board, I have multiple huge, fat planes running on the inside of the board to deliver my power.
2.) Another benefit of using power planes is that of buried capacitance. Think about the construction of a capacitor: it is simply two electrodes separated by a dielectric material. By placing the power planes on the inner layers and surrounding them with ground planes, I've just created a capacitor that's connected to all of my powered devices in the circuit, and a very fast one at that because it has extraordinarily low inductance. It’s an understatement to say that audiophiles generally appreciate the value of capacitance, but very frequently misapply it in the sense that the inductance is compromised. For an example of what I mean, look around the forums for pictures of mods people have done wherein they mount “bypass” capacitors with long, flying leads. Such a device is anything but a true bypass capacitor, it’s more of an LC tank circuit. Will it change the sound? Very likely, but I doubt in a positive way.
3.) Multi-layer boards provide exceptional shielding for signals or power nets that are routed on internal layers. Solid copper planes shield parts on the back side of the board from those on the front side. By connecting multiple planes together, I can even emulate the construction and shielding of a coaxial cable.
4.) Ground planes are a powerful tool, and the Diverter employs three of them. By contrast, it is often rare for a double-sided design to have even a single, contiguous, unbroken plane unless it is a very simple circuit. In addition to playing a role in shielding and providing buried capacitance, ground planes allow a designer to more easily define where and how return currents flow. Why would that be of interest? Ground is often misunderstood as a sort of huge, zero-impedance sink, and it is anything but. A ground connection always has some finite impedance, and if the impedance is significant, it can result in ugly things like ground bounce. If the current loop area is high because of improper grounding, a circuit can radiate EMI like crazy. A ground plane offers the lowest possible impedance connection, and also allows a designer to prevent higher ground currents from modulating the ground reference (sometimes called "quiet ground") used by low-level circuits.
In summary, a six layer board is a very expensive and elaborate technique. There is a lot more that could be discussed here, like autorouting vs. hand-optimized routing, controlled-impedance boards, and so forth. When I said I spent significantly more on the electronics of the Diverter than I did on the chassis, this is one of the things I had in mind, specifically.
And BTW, any hardware device needs a driver, whether its own or generic OS one. These points are irrelevant to differentiate yourself from others.
Thanks, manaox2; glad you enjoyed it. As a designer who may be unfamiliar to many of you, there is a bit of a "getting to know you" period going on here. As I acknowledged in my first post, I can understand a lot of the reasons for the skepticism. I'm trying to address them methodically.
