Optimizing Adapter Power Supply Design

Today, ac adapters are used to supply dc power to a variety of systems, from laptop computers to many computer peripherals. These power supply designs must consider not just functionality requirements, but also their impact on the environment. For example, they must deliver to their target electronic appliance a standby power consumption and an average efficiency that meets the compulsory requirements specified by regulations such as ENERGYSTAR^®, as well as offering safety guarantees. To design an effective ac adapter power supply, it must comply with strict regulations and also survive in harsh application environments.

The first step in the design is to select a circuit topology for the design (typically flyback, forward, dual switch forward, half bridge resonant or a full bridge). Then, determine whether the selected topology meets the cost requirements. The topology should be as cost-effective as possible, while also meeting the specifications to ensure that the final product has a sharp competitive advantage.

A designer can select from the most commonly used circuit topologies, according to the power range of the design that they are working on. For example, for designs less than 100W in power: flyback is the most common; half-bridge resonant or forward for those less than 300W; dual switch forward for those less than 500W, and full-bridge for those greater than 500W. Once the topology has been decided, the designer must then select a controller and a transformer to match.

Controller selection is an important step in power supply design. Most makers categorized their solutions in terms of the different applications that they target. The designer can select the desired solutions according to the product guide provided by the controller maker. In order to meet time-to-market pressures, most designers like to choose solutions that they are familiar with. This does shorten the development cycle, but it can place limitations on the final design. By ignoring new solutions that are significantly improved in terms of performance, functionality, integration and reliability, designers could well ignore more competitive controllers in favor of earlier iterations. Therefore, controller selection can be the key factor in designing for higher performance, lower cost and a sharper competitive advantage.

CORE SELECTION

To select a transformer for the design, we first determine the type of transformer the design needs using the AP (area product) method. There are many types of transformers available. Figure 1 shows the shapes of the magnetic cores commonly used in transformers.

A number of factors may need to be considered when selecting a transformer. For instance, designer needs to find a transformer bobbin category, and try to select a type that matches the design requirements most, in terms of its dimensions, nodes and node pitch.

The selection of magnetic core material depends on the working frequency and the inductance current mode that the power supply uses. For example, if the frequency is around 50KHz and discontinuous current mode (DCM) is used, a ferrite core is usually the best choice; for designs using continuous current mode (CCM) and requiring an inductor with small current ripples and a small hysteretic lag, the magnetic core should be made of material with higher magnetic saturation (such as Kool-mu and powdered iron), as ferrite is limited in magnetic saturation.

The more widely used transformer bobbins are usually less expensive. On the other hand, there are different types of transformer windings. Transformers using triple insulated wire (TIW) are more expensive than those using enameled wires. For example, PQ, RM and POT have smaller winding space than ER and EE. They must deploy TIW wires in the interests of safe spacing.

It should be noted that a larger couple area between the magnetic core and the winding creates low magnetic leakage. For example, PQ, RM and POT bobbins deliver a better performance in magnetic leakage and EMI than either EE or ER bobbins. However, if the design uses resonant topology, it needs to leverage the magnetic leakage effect. This is why ER bobbins are often used in resonant topology. In the example given in the following article, we will discuss how to select a quasi-resonant DC/DC transformer and DCM PFC inductance for a design.

Once the controller is selected, attention can then be turned to circuit design, taking into account such considerations about electromagnetic interference (EMI), protection capability, and regulation factor and sequence if multiple outputs are required. Once the circuit diagram is completed, the designer then moves to the PCB layout. When creating the PCB layout, great attention should be paid to factors such as the spacing between components, the consequent EMI, component heat dissipation, controller feedback loop, large current loop and grounding route. The PCB layout has a direct influence on both design testing and product performance.

A designer needs to determine what specifications (voltage, current, etc.) the devices to be used in the design will require, first by theoretical calculations and then by experimentation, for each of the function block and then for the whole system.

The test items include voltage regulation factor, voltage ripple, protection functions, inductor saturation status, maximum stress on main switch transistors and the temperature endurance margin of these transistors.

Continue to next page

The last step is to test and verify the product's safety. This is done by carrying out tasks such as EMI (electromagnetic interference), thermal test (checking whether the maximum temperature of switch transistors, inductors, capacitors are within the specified range), safety testing such as lightning surge, Hi-pot, EFT, EDS, insulation resistance, leakage current test and finally we calculate the product's mean time between failures (MTBF).

DESIGN PROCEDURE

The above discussion presents a brief snapshot of the procedures involved in a power supply design. We will illustrate the discussion using a 120W adapter power supply design. The specifications of the power supply are listed in Table 1.

The circuit topology for this design could be flyback or half-bridge; if the budget is tight, flyback DC/DC topology would be a good choice.

As the design calls for a power factor correction (PFC) function, the design needs a PFC controller. A 120W power supply works well in inductance current critical conduction mode and there are many PFC controllers that support this mode. For this particular design challenge it is worthwhile considering the Fairchild FAN6961.

Delivering a balanced trade-off between performance, EMI and cost, the DC/DC conversion in the design is implemented by a quasi-resonant circuit. The FAN6300 is suitable for quasi-resonant topologies and could be implemented in this design.

FAN6921 is a solution with a higher level of integration that offers all the functions of both the FAN6961 and the FAN6300, as well as some newly-added functions such as input low-voltage protection, overheat protection, input overpower compensation for both high line and low line.

In addition, the FAN6921 performs well in terms of average operating efficiency and standby power consumption, and its total external component count is reduced by 20 compared with that of above-mentioned design solution that considered using the FAN6961 for PFC and the FAN6300 for DC-DC separately. So, it has the advantages in both performance and component cost, Therefore, for this particular design we finally selected Fairchild's FAN6921.

In this design example, we selected RM for the PFC and PQ type for the DC/DC element of the design, (Figure 2 is the Ae and Aw shape of the PQ type,). As the power supply size of the design is relatively (small), the cores that we have selected both have a low magnetic leakage and a larger Ae and Aw value. This delivers advantages in both power efficiency and EMI suppression to the design.

Now, we calculate the specification parameters of the PFC inductor, and select suitable rectifiers and MOSFETs. Before doing the calculation, we have to assume several parameters of the PFC component, as listed in Table 2.

To calculate the minimum inductance:

(See Eq. 1)

To reduce the power loss that the PFC may suffer at low input voltage, the PFC can be implemented in two lever; in the range of 90~150Vac the PFC output voltage is set to 240~260Vdc and 150~264Vac to 390~400Vdc.

Calculating the peak and RMS current:

(See eq. 3)

Calculate the AP :

(See eq. 4)

Continue to next page

Where:

Irms = RMS current of the inductor

Bmax = Maximum magnetic flux density

Ko = Effective coefficient of the inductive window

In this example, we selected a magnetic core of RM10, which has an Ae value of 98mm2, and an Aw value of 69.5mm2.

Calculate the minimum turns:

(See eq. 6)

Calculate the auxiliary winding turns:

(See eq. 7)

(See eq. 8)

The final PFC inductance parameters are shownin Figure 3.

In reference to the maximum current and voltage of the MOSFET to be used in this example, we selected Fairchild's SupreMOS™: FCPF16N60NT, 16A, 600V, RDS(ON) = 0.17Ω, ID=10A (Tac=100°C). The power loss of this MOSFET can be calculated as:

(See eq. 22)

For the rectifier, we selected the FDH08H60S. The power loss of this rectifier can be calculated as:

DC-DC TRANSFORMER DESIGN

In reference to the maximum input voltage (400V) and the output voltage (19V) of the DC-DC, we assume a 650V rated voltage for the MOSFET at the primary side and a 100V rated voltage for the rectifier at the secondary side. We can calculate the VQ_max and VD_max as shown below:

In this example, the voltage headroom values (K) for the MOSFET and the rectifier are set to 0.9. If larger values are required, set them to 0.8 or other proper values. It should be noted that different K values means that different voltage specifications will be required for the MOSFET and rectifier.

Determine reflect voltage (VRO) and max. duty cycle (Don_max):

This example uses a quasi-resonant topology, so we have to assume a fall time (Tf =1µs) for the MOSFET, To reduce the turn-on voltage of the MOSFET as much as possible, hence the loss and EMI.

Calculate the inductance, maximum current, RMS current:

(See eq. 27 on page 34)

(See eq. 29 on page 34)

.Calculate the AP to select the magnetic core

(See eq. 30 on page 34)

Based on the AP value calculations, we were able to select a magnetic core of the PQ3225 type, which has an Ae value of 114mm2 and an Aw value of 161mm2.

Calculate the minimum turns of the primary winding:

Determine the turns ratio (n) and the secondary winding turns:

The turns of the primary winding can be recalculated from n and Ns as:

Np=Ns x n = 30Ts > Np_min

Figure 4 shows the final transformer parameters.

Choose the primary MOSFET and secondary rectifier. In reference to the maximum current of the MOSFET that is calculated above for the primary MOSFET, we selected Fairchild's UniFET™ series: FDPF15N65, RDS (ON) = 0.44Ω, ID = 9.5 (Tac=100°C).

For the secondary side rectifier:

(See eq. 37)

To enhance efficiency of the total system. we selected Fairchild's synchronous rectifier controller, the FSR510 (Figure 5, which integrates a synchronous rectifier and a synchronous MOSFET (RDS(ON) = 9mΩ, Vf = 100V), needs only four external components, all while delivering a good performance. The circuit diagram of the Final design is shown in Figure 5.

Continue to next page

PERFORMANCE TESTING

As shown in Figure 7, the average efficiency over the whole voltage range is more than 87%. Regulation rate as shown in Figure 7, the variation between output voltage and input voltage is within 2%. As shown in Figure 8, the PF stands above 0.9 over the whole input voltage range, and the THD is less than 15%, delivering a very good PFC. As shown in Figure 9, the no-load standby power consumption is compliant with Energy Star specifications (50>W>250W, Pin0.5W). As shown in Figure 10, the maximum operating voltages and currents of the primary side MOSFET and secondary side rectifier are all within the specified range.

When setting up the power supply, special attention should be paid to the settings of several pins on the FAN6921, as well as the switch timing control of the synchronous rectified signal. In other words, we have to carefully adjust some settings.

More specifically, apply a proper voltage on PIN3 (INV) through the divider circuit formed by R23, R27, R39 (Figure 6) to hold the minimum output voltage (Vo_minPFC) at 240V. Also, set the maximum output voltage (Vo_maxPFC) to 400V through Pin1 (Range). This could be implemented because when the input voltage rises, the voltage on Pin13 (VIN) also increases; when VIN rises beyond 2.1V, it drags down the voltage on Range, causing the PFC output voltage to rise due to R34 and R39 being in parallel. If the design only requires the 400V output, remove the R34 on the Range and change the value of the remaining R39 to the parallel resistance of the original R34 and R39. The two voltages can be calculated as:

(See eq. 38)

(See eq. 39)

PFC BROWN-OUT TUNING

When the input voltage drops, the voltage on Pin13 (VIN) also drops; when the voltage of the VIN falls below 1V, the FAN6921 will force the VDD to work in hiccup mode. In this case, both the PFC and the PWM will disable until the VDD voltage is re-established and VIN is higher than 1V. This is how the brown out function works. When setting up the design for the first time, remove the PWM MOSFET (Q2), work only on the PFC element of the design and then plug Q2 to set up the PWM element.

Pin10 (DET) consideration

It has three functions:

MOSFET valley voltage detection, which controls low voltage conduction, namely it detects whether DET voltage is below 0.7V. If so, conduction is enabled.

Overvoltage protection, which locks the IC once DET voltage is higher than 2.5V. The locked IC can only restart after it is powered off. It is recommended that the normal DET voltage is set to around 2V.

Overpower compensation, which detects the current (IDET) going through R15. When the input voltage rises, IDET also increases, and this drags down the overpower limit voltage. So, by regulating the bias resistance (R15) on the DET, the overpower protect point can be held unchanged when the PWM input voltage is between 240V and 400V.

OVERTEMPERATURE PROTECTION

Pin12 (RT) implements overheat protection in two ways:

By using a thermal resistor as a temperature sensor. In this case, the controller is locked only after the RT voltage remains below 0.8V for a time greater than 10 ms.

By using an external component (for example, an optically coupled circuit) to control Pin12. In this case, the controller will be immediately locked once the RT voltage drops below 0.5V for 100µs.

When setting up the DC/DC, it is better to replace the synchronous rectifier controller with a Schottky diode. This avoids damage to the DC/DC controller and the MOSFET that can be the result of improper settings. Set the synchronous rectifier after the PWM has been set up. In reference to Figure 11, disconnect Pin1 (DET) and apply a 5V DC voltage on the pin. We do this to check whether the synchronous rectifier is working normally. If the rectifier is abnormal, regulate the values of R1, R2, R3 and R4, with reference to the DET and MOSFET waveforms, to ensure that the synchronous MOSFET turns on at the same time as the primary MOSFET and turns off about 2µs ahead of it.

Continue to next page

TURNS RATIO (N)

Finally, we discuss how the design is influenced when the turns ratio (n) of the flyback DC/DC transformer is changed. It can be seen from the VRO calculation that when the turn ratio (n) rises, the reflected voltage VRO will rise, the primary MOSFET maximum voltage Vds_max also rises, but the secondary rectifier maximum voltage Vsd_max drops. Therefore, the turn ratio (n) must be a proper value that is trade-off between Vsd_max and Vds_max. As with the calculations made earlier in this design, we must first determine the primary/secondary side voltage, then determine the turn ratio (n), and finally select the proper current specifications of the MOSFET and rectifier based on the currents that we have calculated.

Figure 12 shows the performance and cost comparisons among several MOSFETs (sourced from Fairchild's website for reference only). It can be seen from this figure that the 13A/600V and 16A/600V MOSFETS (Parts FCPF13N65NT and FCPF16N65NT respectively) are very similar in their performance and cost. While the 13A/800V MOSFET (Part FQA13N80) is more expensive than the 12A/600V MOSFET (Part FQPF12N60NC), it exhibits poorer RDS(ON) performance. So, it is better to calculate the MOSFET voltage specification first, and then follow by calculating the current. Of course, the voltage of the secondary side rectifier is related to the output voltage of the power supply; when the output voltage is very high, a secondary side rectifier with a higher voltage is desirable.

Of course, the example design still requires modification before it can be turned into a highly reliable and safe power supply product. The power adapter must be able to survive the extreme temperature, humidity, and vibration of harsh application environments, as well as be able to meet strict performance requirements and the stringent government regulations surrounding safety and energy-efficiency.

REFERENCES

Fairchild Semiconductor: Design Guidelines for Quasi-Resonant Converter using FSCQ-series Fairchild Power Switch (FPS),AN-4146,Fairchid semiconductor application note, 2005.
Fairchild Semiconductor: FAN6300-Highly Integrated Quasi-Resonant PWM Controller, AN-6300,Fairchid semiconductor application note, 2008.
Fairchild Semiconductor:Design of Power Factor Correction Circuit using FAN7529, AN6026, Fairchild Semiconductor Application note, 2008.
Fairchild Semiconductor: Secondary Synchronous Rectifier for Quasi-Resonant Controllers, AN-6085, Fairchild Semiconductor application note, 2009.
Lloyd H.Dixon,“Magnetics design for switching power supplies”,2000