Creating a 150V non-synchronous buck solution with a lower rated input voltage controller

In many applications today, the required input voltage rating exceeds the VIN maximum rating of many existing DC/DC controllers. Traditional solutions include using expensive front-end protection or implementing low-side gate drive devices. This means using isolated topologies such as flyback converters. Isolated topologies often require custom magnetics and increase design complexity and cost compared to non-isolated approaches.

There is another solution that can solve the problem by using a simple buck controller with VIN max (maximum input voltage) less than the system input voltage. How is this achieved?

Buck controllers are usually biased from a reference potential (0V) (Figure 1a). The bias supply is derived from the input voltage; therefore, the device is subject to the full VIN potential. However, because the gate drive voltage required to turn on a P-channel metal oxide semiconductor field effect transistor (MOSFET) is lower than VIN at VGS, a P-channel buck controller has a gate drive supply referenced to VIN (Figure 1b). Turning off the P-channel MOSFET is simply a matter of changing the gate voltage to VIN (0V VGS) (Figure 2).

Figure 1: VCC bias generation for N-channel (a); and P-channel controller (b)

Figure 2: Gate drive of a P-channel controller

A non-synchronous P-channel controller derives its bias supply to drive the P-channel gate, which can bring huge benefits and may provide a virtual ground floating above the 0V potential. For an N-channel high-side MOSFET, the voltage comes from a ground referenced supply. This is charge pumped using a boost capacitor and diode to provide a gate voltage above the VIN source potential. Using a P-channel high-side MOSFET can greatly simplify this problem. To turn on the P-channel MOSFET, the gate potential needs to be lower than the source potential of VIN. Therefore, the power supply is only referenced to VIN, rather than VIN and ground as mentioned above.

Floating grounding

How do you create a floating ground for the controller? This is easy and can be accomplished by using an emitter follower. Figure 3 shows this scheme in basic practice. The potential of the PNP emitter is at Vbe (~0.7V), which is lower than the Zener diode voltage potential (Vz). Essentially, you float the controller to VIN and adjust the controller's reference to limit the voltage between VIN and the device ground.

Figure 3: Creating a virtual ground using a simple emitter follower scheme

Output voltage conversion

There is a challenge to overcome here. Since the controller is at virtual ground (Vz-Vbe) and produces a buck output voltage referenced to ground (0V) potential, how can the output voltage signal be converted to a feedback voltage that is above virtual ground (usually between 0.8V and 1.25V)? Figure 4 illustrates the challenge.

Figure 4: Schematic showing the voltage potential difference between VOUT (referenced to 0V ground) and the controller’s feedback voltage (referenced to virtual ground).

To close the loop, you can use a matched pair of transistors to implement the circuit shown in Figure 5. One matched pair sends the feedback signal to VIN; the other matched pair generates current from VIN to a potential above virtual ground.

Figure 5: High-level schematic of a nonsynchronous controller and feeding practice using paired transistors

In summary

The LM5085 is an ideal choice for the application I described because it is a P-channel non-synchronous controller with its VCC bias supply referenced to VIN. In traditional applications, the LM5085 can withstand input voltages up to 75VIN. For applications with input transients much higher than 75V, consider the solution proposed here, which has a 12V output.

Starting with a controller feedback voltage of 1.25V and using the current to set the feedback (Ifb) to 1mA, calculate the Rfb value using Equation 1:

Where Rfb = 1.25k.

Rfb1 sets the reference current for the current mirror. Again, using 1mA as the reference current and using Equation 2, calculate Rfb1 to set the output voltage:

Where VOUT = 12V, Rfb1 = 11.3k, and Vbe is ~0.7V.

When 1mA flows into Rfb2 and the emitter current is roughly equal to the collector current (Ie~Ic), the reference current Iref2 is set. The loop is closed and the voltage will be regulated to the set voltage as described.

Output Voltage Regulation

This idea is applicable when the transient voltage is significantly higher than the absolute maximum of the LM5085 . The LM5085 is a constant on-time (COT) controller; therefore, its on-time (Ton) is inversely proportional to VIN. However, when VIN is clamped to the LM5085 , Ton will no longer adjust as VIN (to the power stage) increases because the device will have a fixed voltage set by the Zener diode, while VIN (to the power stage) will continue to increase. This will cause the frequency to drop because the increase in the power stage input voltage exceeds the clamping voltage of the LM5085 ; therefore, the regulation voltage may start to increase slightly. Therefore, to ensure the ripple injection voltage is sized with the Type 1 ripple injection standard. Ultimately, ensure that the ripple is sized within an acceptable range to maintain stability and minimize output errors as the ripple increases.

Example Schematic

Figure 6 shows a schematic diagram of a 48V power supply with an absolute maximum VIN rating of 150V. This example can provide 12VOUT at 3A.

Figure 6: 24V to 150VIN (max) / 12VOUT at 3A design using LM5085 

Figure 7 shows the efficiency graph obtained from the prototype board. The two main parameters in the graph are efficiency (%) and load current (A).

Figure 7: Efficiency (%) vs. load current (A) at different input voltages

Figure 8 shows the switch node voltage and inductor ripple current at 150VIN.

Figure 8: Channel 1 switch node voltage, Channel 4 inductor ripple current

in conclusion

You can use a P-channel nonsynchronous buck controller in applications where the system input voltage is higher than the device's maximum input voltage rating. The advantage of this application is the use of a lower cost controller and minimal component count. For design guidance on the buck converter power stage, see the Application Information in the LM5085 datasheet .