There are many good reasons for the DIY’er to use active speakers, and we will look at some of those reasons in this article. But along with good reasons for designing active speakers, there have been recent advancements in technology that make it more and more reasonable to ask “why not”. The entry costs have dropped dramatically in the past few years, with amplifier modules, DSP hardware and control processors available at low prices from many sources. And there are some powerful, easy-to-use development tools available to simplify the design process. As a result, active speakers are much more affordable than ever before, and for some designs you don’t have to have an EE or math degree to build them.
This is not intended to be a comprehensive argument for using active speakers, as there are many good resources with very detailed investigations that we can refer to for more details. Also, if you are reading this article you are probably already convinced of some of the benefits, and there is no reason to “preach to the choir”. But a high-level review of the benefits will help us identify the types of DIY designs where an active solution would be preferable, and it may help remove some of the hesitation people might have in jumping into active speaker design.
Active Speakers—our working definition
For the sake of this article, we are going to entertain a very broad definition of active speakers that includes the use of a separate amplifier/DSP enclosure. Sometimes “active speakers” refers to loudspeakers with built-in power amplifiers. But we are going to expand the definition to include speakers for which the electronics are in a separate box or just one speaker of a set. So we aren’t concerned with how the electronics are packaged—just whether the drivers are directly connected to an amplifier and whether the crossover and equalization networks use active circuitry.
Some active speaker benefits
1. Direct connection to the speaker
Directly connecting the speaker to the amplifier, without a crossover network in between, has at least two nice benefits. First, the damping capability of the amplifier is not compromised by the rising impedance of the crossover network at the crossover frequency. Rod Elliot has what is almost a passionate rant about this issue on his website at http://sound.whsites.net/biamp-vs-passive.htm. He points out that directly connecting the speaker to the amplifier results in much better control of the back-EMF voltage from the driver, thereby reducing distortion. He doesn’t provide any measurements that quantify this reduced distortion, and he points out that some “high-end” drivers with good back EMF suppression (such as those with conductive rings in the magnet structure) won’t benefit as much from the direct connection. But even though there are no measurements to back up his claims, the overall argument makes sense, and is a compelling reason to consider active speakers.
The second benefit of direct connection to the driver is that you don’t need impedance measurements to design the crossover. The only driver measurement required is the frequency response, and for some types of crossovers, even that isn’t necessary (see next section).
2. Simpler crossover design with high-order slopes
With modern DSP chips, the designer has the option of using high-order filters for the crossover. So the designer can simply use textbook filters with steep slopes and often get very good results, even without measuring the individual drivers as part of the crossover design process. Although steep crossover slopes aren’t to everyone’s liking, they greatly simplify the crossover design effort and provide a quick way to put a wide range of drivers to use.
By steep slopes we mean 24db or 48db per octave filters in the crossover. A 48-pole 3-way crossover requires 16 digital biquads to implement, and that is easily done in a modern DSP chip. With these steep slopes, the overall response is mostly defined by the electrical response of the filter, and the driver response is “swamped out” by the electrical response of the filter. So as long as the driver is reasonably well behaved in the crossover passband, the overall design will be flat, and from there the designer simply needs to tweak the response with equalization filters. There are many good examples of high quality speakers that use high-order crossover slopes, particularly in professional audio gear, and writers like Linkwitz have extolled the virtues of 24db electronic crossovers for many years.
3. Simpler crossover design without high-order filters
Of course, there are mixed opinions on whether crossovers with steep slopes result in the best audio quality, and for some designers the best crossover will use filters with shallow slopes that work with the driver’s naturally frequency characteristics to result in minimal delay, a smooth phase transition and a large overlap in the response between drivers. But even for active speakers with low-order filters, the active crossover can be much simpler to implement. The designer simply needs to model the driver response and interactively adjust digital filters to achieve the desired crossover slopes and phase response. Or, with the right modeling tools, the filter response can be calculated by comparing the driver response to the target response, and the modeling tools can then determine the optimal filter coefficients and delay and volume levels needed to achieve the target response. Although this type of design can be achieved by an experienced designer with passive crossovers as well, the active filters can be downloaded and tested in real time on the actual hardware without using a soldering iron to swap out passive components. That ability to test out changes almost instantly can save time and simplify the overall crossover design effort.
4. Multiple amplifiers allow using smaller low-cost amps
We’ll again refer to Rod Elliot to help explain this benefit, as he has provided a good analysis of the benefits of using separate amps for the different frequency ranges. He illustrates how two 100 watt amps, separated into woofer and tweeter ranges, can provide greater peak handling capability than a single full-range 200W amp. That’s because the full-range amp may have to deal with the peaks in both frequency ranges, and when the low and high frequency peaks are aligned, it requires twice the voltage from the amplifier. Twice the voltage means that the amplifier will need to provide 4 times the power.
Being able to use smaller amplifiers for each frequency band is very important because of the recent availability of many low-cost class D amplifiers at reasonable prices. This cost benefit is addressed in the section below on class D amplifiers.
Benefits versus Cost & Complexity
Not so long ago, active speakers were costly, difficult to design and took a long time to build. As a result, they were usually off-limits for the DIY’ers who didn’t suffer through an engineering program. But the electronics and design tools available for the hobbyist have changed dramatically, especially in the last 5 years, and now it is much easier to build active speakers. We’ll take a quick look at some of these technology changes in the sections that follow, but we’ll save the details for some articles later in the series.
1) Class D amplifiers:
Class D amplifiers simplify active speaker design in two important ways. First, they generate much less heat due to their high efficiency. The common class AB amplifiers have a theoretical efficiency of 78%, but that efficiency is only achieved at full volume, outputting a square wave. Typically, class AB amps are running in the 40% – 45% efficiency range. That means that most of your power goes into warming the heat sink. Getting rid of that heat can be a difficult challenge for active speakers, as it requires large plates with fins or other heavy, complicated cooling solutions. Class D amps have a theoretical efficiency that is usually over 90%, and at typical volume levels the efficiency is in the 80-90% range. That’s a huge difference in the amount of heat that needs to get removed, and the heatsinking solution can be much simpler and lower cost. Of course, the designer still needs to make sure that heat is removed, but it is a much easier problem to solve.
The second reason class D amplifiers simplify active speaker design is that most of the class D amps use a single supply voltage. This single supply voltage reduces complexity and allows using readily available off-the-shelf products. And since you aren’t having to waste power warming the heat sink, the power supply can be about half the size of a comparable class AB power supply. As a result, it is often possible to get by with a plug-in AC adapter type of power supply, like those used for laptop computers. These power supplies tend to be very low cost and easy to use, and by keeping the power supply outside the cabinet, it is much easier to dissipate the heat.
And as already noted, the medium-power class D amplifier boards are available at a very low cost, benefiting from low R & D investment (they are mostly reference designs provided by the chip manufacturers) and large overseas production runs. This large selection of quality amplifiers that are ideal for active speakers is a great benefit for the DIY community.
2) DSP chips
The third article in this series provides a short history of DSP evolution and a comparison of a number of DSP chips, but for this overview we’ll just look at the high level capabilities of a single device, the ADAU1701. The ADAU1701 was introduced in 2008, which is actually a long time ago in “semiconductor time”. But it is still a competitive chip, and it provides a lot of capability and is (somewhat) easy to work with.
The ADAU1701 executes 1024 instructions for every audio sample. A lot of functions such as volume control and signal routing are completed in one or two clock cycles. A biquad filter, which is discussed in more detail in the second article, takes 5 clock cycles. So if you are processing 48KHz audio, the ADAU1701 can calculate the results of about 200 biquad filters for every sample. That’s a lot of DSP horsepower for a chip that sells for around $6. In addition to the processor core, the ADAU1701 has 2 analog inputs and 4 analog outputs, and 4 more digital output streams. It also has some general-purpose inputs and outputs and some ports for reading volume controls or other analog signals.
With this capability and low cost, the entry price for active speakers has dropped rather dramatically. The miniDSP was the first to offer a commercial product with the ADAU1701, but recently Sure Electronics started selling a $20 ADAU1701 DSP board. And you can download board designs to build your own version of the miniDSP 2×4 from Audiodevelopers. Ahh, what times we live in… 🙂
3) Microprocessor evolution and Arduino Software
This topic also has its own article, so we will simply point out here that designing with microprocessors has gotten a lot easier. The Arduino boards that evolved from the Maker community have finally matured to where it makes sense to use them for controlling audio DSP. That wasn’t the case several years ago, as the boards were a bit underpowered, the development tools were too simplistic, and there weren’t enough good libraries to properly support audio DSP designs. But things have changed. An especially beneficial development has been the growth of the Arduino support community, as there are many tutorials and examples that will lower the learning curve and help you find code that you can integrate into your projects. So now it makes good sense to embed an Arduino board in active speakers, and if you haven’t played with microprocessors in the past, it’s a great time to learn about them.
4) Design tools
The Active Speaker Designer (ASD) application allows importing driver measurement data and interactively designing crossovers and EQ with real time control of the hardware. The details of ASD are in article 13. I don’t know of any other tools that are comparable, so that’s the only design tool that we will consider in these articles. Unfortunately, it is one of those programs that is always under construction and always needs fixes that take more time than what I can give it, so it is a bit “rough”. And since the bulk of the user interface was done several years ago using Windows forms, it could use a makeover to improve the look and feel. The ASD executable code can be downloaded from the Audiodevelopers web site.