Simplify your power storage design with these Fly-Buck™ power tips


Traditionally, designers have used flyback converters and push-pull drivers with transformers for generating an isolated power supply. The Fly-Buck™ converter (or isolated buck converter) has gained popularity as a low-power isolated bias solution because of its simplicity, ease of implementation and low bill of materials (BOM) cost for uninterruptable power supply (UPS) designers.

The Fly-Buck converter evolved from a synchronous buck converter by adding coupled windings to the inductor for isolated outputs as shown in Figure 1.

Figure 1: Evolution of a synchronous buck converter to a Fly-Buck converter

Fly-Buck converter results in a simpler solution than a flyback converter at lower power levels because of the integrated field-effect transistors (FETs) and the absence of any external feedback loop. The attractiveness of the topology has been mostly at low power levels (< 15W), where tight regulation on the secondary voltage is not required, such as in power supplies for gate drivers.

The Fly-Buck converter also offers the primary non-isolated buck output at no additional cost. Therefore, it results in a simpler design for applications that require both isolated and non-isolated outputs.

However, you cannot configure every buck converter as a Fly-Buck converter.

Choosing a synchronous buck converter

In a Fly-Buck converter, the low-side switch cannot be a diode as it is in a nonsynchronous buck. In a Fly-Buck converter, the primary current flows in the reverse direction during the off time due to the reflected secondary-side current on the primary side and may become negative. The diode in a nonsynchronous buck will block this negative current, which blocks the energy delivery to the secondary. As a result, the isolated output voltage will collapse.

For some synchronous buck-converter integrated circuits (ICs), the low-side FET turns off (for example, in pulse-frequency modulation [PFM] mode) if negative inductor current is detected, thus increasing light-load efficiency. In such cases, the FET emulates the diode behavior, making such buck converters unsuitable for a Fly-Buck converter. However, in some converters you can disable the diode emulation mode so that the converter works in forced pulse-width modulation (PWM) mode. You can use such converters as a Fly-Buck converter. An example of such a converter is TI’s LM5160.

Choosing the right control scheme

Not all control schemes are fit for Fly-Buck converters. Because the primary-side current in the off time is different than a normal buck, current-mode control relying on low-side FET current or valley-current sensing will not be the right choice for a Fly-Buck converter.

Choosing the right current-limit scheme

To incorporate overcurrent/short-circuit protection, a synchronous buck IC that has a current limit on the high-side switch is a good choice, since the current waveform in a Fly-Buck converter during its on time is similar to a normal buck converter. The primary current flows in the reverse during the off time due to the reflected secondary-side current on the primary side and can become negative. For ICs that have only a valley detection control circuit, where the switch current is monitored during the off state, the overcurrent situation may not be detected (see Figure 2). Such a situation may damage the IC.

You should choose the peak current limit based on the design requirements, especially considering a situation where the primary side is not loaded and the secondary side is shorted. Some ICs also have negative overcurrent protection (when the converter sinks current through its low-side FET) that detects voltage across the low-side metal-oxide semiconductor field-effect transistor (MOSFET) and allows the control circuit to turn off the low-side MOSFET if the reverse current exceeds the negative current limit. An example of such a converter is TI’s TPS55010.

The LM25017 is a good fit for a Fly-Buck converter, as it operates in continuous conduction mode (CCM) regardless of the output loading. The device has constant on-time (COT) control, which is not affected by the current waveform, current-limit protection in the high-side (buck switch) FET, and an adaptive current limit off-timer.

 

Figure 2: Overcurrent/short circuit in a Fly-Buck converter

Reference design for an isolated gate driver supply

The Compact, Single-Channel, Isolated SiC and IGBT Gate Driver Reference Design for UPS and Inverters employs a 1.5W Fly-Buck converter configuration for powering isolated silicon carbide (SiC) and isolated gate bipolar transistor (IGBT) gate drivers used in UPSs, DC inverters and electric vehicle charging station applications to drive the power stages. The isolated Fly-Buck converter power supply is realized using the LM25017 in a 33mm-by-23mm form factor. Additionally, this reference design provides a fully tested, robust, independent, easy-to-validate, single-channel driver solution capable of withstanding >100kV/μs common-mode transient immunity (CMTI). The reference design guide has comprehensive details of the Fly-Buck converter design and implementation.

Additional resources:

  • Hi Rami, thank you for the helpful tips. I'm not sure this is the best place to ask you a question, but I continue. Recently I was intrigued by the possibility of using Flybuck design, therefore to try the idea I found an existing sync buck IC (not TI, sorry) but still it seemed to hold some promise for modification to FlyBuck. It is the ST1S10 by ST. Having replaced the existing inductor with a selection of various transformers and even self would transformers. I finally drew a blank. Is this a transformer problem or the IC design? I have carefully analysed your comments here about suitable and not so suitable topologies and I would like to confirm with you before I finally relent. This IC, seems okay in most respects. It does its primary current sensing on the high side FET and runs at very similar frequencies to the ICs you discuss. However it does have a pulse skipping mode and discontinuous modes of operation. It was simple enough to prevent pulse skipping by simply lowering the supply voltage until it stopped. But Discontinuous mode detection was not easy to disable. In fact the IC does not allow this to be forced off and operate in CCM mode. However I wanted to see what happens if the IC is forced into CCM simply by loading the primary sufficiently that it never enters DCM. I did this but still I find that the secondary collapses even when the IC is operating in CCM.

    DCM detection is achieved within the IC using a simple drain source voltage comparator across the low side FET. I assume therefore if the inductor current ever goes negative, it enters DCM. But actually STs internal simplified diagrams show that this negative current situation actually triggers the controller to swap to conduction on the high side FET. It is not clear exactly..

    But to get quickly to my direct question. If DCM can not be disabled, does this totally rule out an IC in which DCM can not be disabled? Will the secondary current still not be available as long as the inductor current stays positive. You state in your blog that primary current flows in the opposite direction due to reflected secondary current, AND MAY BECOME NEGATIVE. However if it does not become negative because the primary is heavily loaded, would this not prevent DCM mode anyway.

    Bottom line my secondary simply collapses as soon as I attempt any reasonable load. Regardless of operating in DCM or CCM mode. Load up the secondary with just a few mA and the voltage falls away, finally holding up at around 1V. Why so?