Skip to content

Electronics isolation and multi voltage design

Pekka Nikander edited this page May 28, 2013 · 14 revisions

Multi-voltage domains

Many electronic design blocks have vastly different power and voltage needs. Combining these blocks together in a system creates a need for systematical protection between the blocks. Figure 1, below, illustrates a design with multiple voltage domains.

Illustration of a single-sided multi-voltage system

Figure 1. A single-sided multi-voltage system

Voltage potentials may be single- or double-sided. Single-sided multivoltage systems are most common. Many audio systems use double-sided multi-voltage systems; see Figure 2.

Illustration of a double-sided multivoltage system

Figure 2. A double-sided multi-voltage system.

Level translation

In many cases total galvanic isolation is not necessary. In such a case, the system may share a single common reference potential. For most designs, this shared reference is called as the ground potential. Most commonly, a digital control circuit uses a 3.3V or 5V logic level voltage and the controlled, typically analog circuits use a higher voltage; for example, in the case of Power-over-Ethernet, the supply voltage is nominally 48V and in practise varies between 40 and 56 volts.

For systems sharing a common potential, passing control signals between voltage domains is usually accomplished with simple level translation. Shifting the voltage level from a higher potential to lower one is easily accomplished with a resistor-based voltage divider; see Figure 3.

Illustration of a common ground level down shift

Figure 3. Shifting down a voltage level with a common ground.

Translating from a lower to a higher potential is most easily accomplished with a small FET and a pull up resistor. This is very similar to driving a load using a low side switch; however, in control signal level translation the FET is typically much smaller (and cheaper) than when driving a load.

Illustration of a common ground level up shift

Figure 4. Shifting up a voltage level with a common ground.

Galvanic isolation

In some cases, modules operating at different voltage levels need good protection from each other while allowing for control messages to pass between the protection domain. In the typical case, the protection is accomplished with galvanically isolating the subcircuits and using optical, inductive, or capacitive isolation for control signal communication.

Optical isolation

Signals that have a continuous DC component are usually isolated with optoisolators. An optoisolator has a LED and an optodetector, optically connected within a single package. In the ideal case, the output signal is linearly related to the input signal. Driving and receiving the signals requires tailored but simple circuits. The signal levels on both sides are small. Performance and linearity are quite good while the components are relatively inexpensive. Figure 5 illustrates a typical optoisolator design.

Isolation - optoisolator

Figure 5. A typical optoisolator circuit.

Inductive isolation

Transformers are able to isolate and scale the signal level. They are usually simpler to interface to than optoisolators. On the downside, transformers are not able to pass a continous DC signal and are relatively expensive, especially if a common standard Cots models are not suitable. Linearity may be an issue and the signal bandwidth is typically limited.

Isolation - transformer

Figure 6. A typical transformer isolation circuit.

Capacitive isolation

Capacitors are used in many high frequency isolation needs, as they are very cheap and their performance is easily predictable. They do not pass a continous DC signal.

Isolation - capacitive

Figure 7. A capacitive isolation circuit.

Integrated circuit solutions

All of the optical, inductive and capacitive isolation solutions and some more more exotic methods for passing signals through isolation are readily available as integrated circuits. Some ICs are tailored for a specific task, such as driving a high side FETs of an H bridge. If such a ready-made solution exists, it is usually worth digging into.

Below, Figure 8 illustrates an H bridge driver for a loudspeaker. The problem with driving such a bridge is that the gate drive signal is referenced to the source terminal. For the low-side FETs this is no problem, but the high-side FETs source terminals are floating and may easily be tens of volts above the ground potential. Hence, the driving signal needs to be translated into a potential where the gate voltage exceeds the source voltage.

Special driver example H bridge

Figure 8. An H bridge driver.

Clone this wiki locally