In SMPS, maintaining precise output regulation while preserving galvanic isolation presents a fascinating engineering challenge. In Part 4 of our series on galvanic splitting techniques, we’ll explore feedback without direct connection. It’s time to see how the output voltage affects the switching system's regulation. That feedback from Vout to Vin must also be made without a direct (galvanic) link.
Figure 1: Isolation in the Feedback Loop on a DC-to-DC Converter
We will focus on four main methods:
- The optocoupler (old classic way, but still in use)
- The 3D winding
- The direct primary side sensing
- The intrinsic built-in iso-buck
Optocoupler
In this approach, light serves as an intermediate agent in the feedback loop. The Vout signal is converted into light via an LED. Facing the LED is a photo-sensitive element (e.g., a photoresistor, photodiode, or phototransistor) that converts the light information back into an electrical signal. Here, light serves as an isolator.
To avoid ambient light interference and perturbations, the LED and phototransistor are encapsulated in the same package: the optocoupler.
Figure 2: Optocoupler Devices: Structure, Symbol, and a Real Example
Figure 3: Optocoupler Action in the Isolated Feedback Loop
Unfortunately, an optocoupler cannot always be integrated because it consumes high current and is bulky. In addition, its characteristics vary with external conditions and time.
Third Winding on the Transformer
Figure 4: Third Winding on the Same Power Transformer Body
The third winding (here, NAUX) senses Vout, with the proportion determined by the ratio NAUX/Ns during the MOSFET switch's off state. NAUX is oriented as the secondary coil. The control block rectifies and filters the snubbed voltage Vw, which is used to quantify and adjust the switching duty cycle.
Direct Primary Side Sensing
Figure 5: Direct Sensing from the Transformer Primary Side
Primary sensing is a smart way to measure Vout without a third winding or an optocoupler. It simply takes the existing primary winding (used to transfer power) as a sensing element. The result is a simpler transformer, reduced board area, easier implementation, and cost-effectiveness.
Intrinsic Built-in Iso-Buck
Figure 6: Iso-Buck Intrinsic Feedback
The feedback loop in the iso-buck is intrinsic and embedded in its structure. In effect, on the standard buck (non-isolated) configuration, the output capacitor, Cout, serves as the energy storage: it is charged by the primary winding during Ton and discharged to Cout during Toff. Vout (or a portion of it) is feedback to FB.
In the iso-buck, the feedback info is still taken from the storage capacitor, Vpri. But this also carries Vout: when the secondary winding is discharging to Cout, Vpri sees that during Toff.
Figure 7: Typical Implementation of an Iso-Buck Intrinsic Feedback with Signals
The iso-buck converter here shows the simplicity of its implementation: the bill of materials (BOM) is dramatically reduced. It is essentially (on the input side) a synchronous buck converter driving a gapped transformer rather than an inductor. When T1 is ‘ON,’ the voltage across the primary winding is negative (equal to -VFB), and the voltage across the secondary winding is positive, allowing the energy to be transferred to the output. The opposite is true during the T1 ‘OFF’ time. The control loop is closed at the primary side, eliminating the need for additional elements, such as an optocoupler or a third winding, while still remaining isolated.
Isolation and Integration
We have seen that isolation techniques, both on the direct and feedback paths, require additional elements, often bulky, such as a transformer, a third winding, an extra inductor, an optocoupler, diodes, etc. The added complexity means an extra hardware challenge for the power supply designers. This is why many semiconductor companies, including Analog Devices, propose isolated DC-to-DC regulators in modules that integrate (nearly) all the necessary components.
Figure 8: Fully Integrated Isolated DC-to-DC Regulator as µModule (based on the LTM8058)
Figure 9: Typical Application of Compact Isolated µModule (LTM8047)
Conclusion
This final part of the series is dedicated to isolation techniques in DC-to-DC converters and discusses how regulation loops can be implemented without direct contact between Vout and Vin. Whether you choose the proven path of optocouplers, leverage the transformer’s inherent capability with auxiliary windings, embrace the cost-effectiveness of primary-side sensing, or explore the integrated elegance of iso-buck topologies, each method offers unique advantages for different applications. The key is understanding that isolation doesn’t mean isolation from control. These techniques prove that, with clever engineering, we can maintain precise regulation while keeping input and output safely separated. The choice of a particular feedback loop depends on the transfer scheme (e.g., flyback, iso-buck, forward) and on the desired performance in terms of regulation quality, speed, noise, size, and cost. The right choice could be the difference between a good design and a great one.
Read all the blogs in the SMPS Galvanic Isolation series.
