In an earlier post, I talked about how I entered the world of low-voltage audio and my commitment to delivering the best possible performance subject to that constraint. In this post I’d like to consider some strategies for generating power in your LV application.
Power supplies have a big impact on the rest of the circuit, so getting grounded in some approaches seems like mandatory prep. Below are four strategies you might consider for doing high-performance audio using a limited DC voltage power source.
Keep it low
For a lot of people, low-voltage audio is synonymous with using a rail directly provided by a USB source or LiPo cell with some filtering. This means you end up with a single positive supply in the range of 3.3 to 5VDC. From a perspective of circuit complexity and cost this is a very enticing option. But from an audio performance perspective, this is the most challenging because it limits both the range of devices you can use as well as the available headroom in the system.
Generate a negative rail
If you’re stuck using a USB or LiPo cell power source, you can generate an additional negative rail from that source using a charge pump or similar technology. This opens up the range of devices you can use, potentially increases system headroom, and lets you use bipolar design techniques. On the downside, generating the additional rail will introduce inefficiency and potentially noise.
Go high
If you’re not bound to using a USB or LiPo cell power source, you can use a unipolar DC power source to give you additional headroom and/or more working space. This approach brings with it all the advantages of generating a negative rail (above), but with potentially even more headroom, a wider range of potential devices, cleaner supply lines, and/or a simpler circuit. One of the things you might want to do with the added working space is employ a phantom ground. You could also generate a negative rail as discussed above.
Switch it up (and down)
If your current and assembly budgets can handle it, a conceptually easy but circuit-wise complex solution is to use a combination of buck/boost/charge-pump technologies to generate the bi- or unipolar rails you already love from whatever source you’re stuck using. If you go this route, don’t automatically assume you need to generate a negative rail: it might be better, easier and/or cleaner to generate a phantom ground instead.
A drawback of this approach beyond the inefficiency, cost, and complexity it brings with it is the potential for added HF noise from one or two additional switching supplies. But if you just need to have +/-15 rails, this is a valid approach.
Managing constraints
The strategies above that are open to you are determined by a few potentially conflicting constraints.
One of these is circuit complexity and cost. Some of the strategies above come with a significant complexity as well as BOM and assembly cost. Many low-voltage audio applications are also cost-sensitive — a reality that would often be convenient to ignore.
Another constraint is the current budget you have available. While the switching technologies used to step up rails or generate negative rails can be quite efficient, it’s still a lossy process. Also, keep in mind that the total power consumed by an opamp or other amplifier tends to grow more or less linearly with its supply rails. This is particularly important for battery-powered applications.
Maximum signal level is yet another constraint. If you need to swing 24Vpp at some point in the circuit, you’re not going to be able to "keep it low." On the other hand, if you only need to swing a maximum of 2.8Vpp (typical of lots of DACs and ADCs), then all options are open.
A final and potentially most important constraint is the required audio performance. It may or may not be possible to wrangle the required performance from a device running off 5VDC in your application. This is typically both an objective and a subjective determination. When you’re making making any subjective evaluations, I can’t emphasize enough the importance of freeing yourself from both biases and wishful thinking.
In the next installment of this series, I hope to discuss some of the terminology surrounding rail-to-rail opamps and the circuit architectures they use.