This is the final video in our five part series of common mistakes in flyback power supplies and how to fix them. In this final video, we'll look at audible noise and how to select the bulk input capacitor for a flyback. 

When audible noise is not a primary consideration in the design phase of a power supply, it may rudely show up later during the testing phase. Most controllers are designed to operate well above audible frequencies during full load operation. As we discussed earlier, modern controllers reduce the frequency or enter a burst mode of operation at lighter loads in order to keep the efficiency high and minimize standby power consumption. 

At light loads, you may find your power supply operating well below 20 kilohertz and susceptible to generating audible noise. For example, the 45 watt power supply pictured here generates an annoying buzzing sound when supplying a lighter load of only 10 watts. Have a listen. 

[BUZZING] 

This graph shows what that noise looks like when measured with a microphone. Here are some pictures of the top of the board, the bottom of the board, and a side view. Something in the construction of this power supply is transducing electrical energy into acoustical energy. Can you identify the offending component? 

Transformers and inductors are common sources for audible noise. Electromechanical forces on the wires and magnetic cores can cause movement and displacement of air, resulting in a perceptible sound. The transformer in this power supply has been glued and vacuum varnished. And it is not the source of the noise. 

However, the center drum of the input inductor L1 is unsecured as shown in the zoom of this image and is only held in place by two wires. When excited by pulsating currents, this center drum acts like a speaker and is the source of the annoying buzzing sound. Replacing the inductor with another inductor that has a center drum glued in place eliminates the noise as shown in the new photo here and the new noise measurements. 

Ceramic capacitors are another common source of audible noise due to the piezoelectric effect. Essentially, when the capacitor is excited by pulsating currents, the ceramic material expands and contracts, causing a deformation of the capacitor. The force generated can also cause deformation or flex of the circuit board in the area that the capacitor's mounted to. It is often the flex of the circuit board that generates the most audible noise. 

To prevent singing capacitors, it's best to place capacitors with large ripple current near the edge of the PCB or you can add a slit under the PCB directly underneath the capacitor. Another technique is to mount two parallel capacitors symmetrically on opposite sides of the PCB so that the forces cancel each other. 

For our final common mistake, consider the 12 volt 48 watt two switch flyback shown here. This power converter operates at a nominal input voltage of 115 volts AC. But as the input voltage is reduced to around 90 volts AC, a mysterious ripple shows up on the output voltage. This converter needs to be operated at an input voltage as low as 90 volts AC 60 hertz so this problem needs to be resolved. 

The plot here at the bottom right shows the output ripple voltage and the input voltage after the rectifier labeled V BULK on the schematic. What is the cause of this excessive ripple voltage? A big clue is the frequency of the ripple. Notice it is 120 hertz. And the droops align with the valleys of the rectified AC voltage. When the input voltage sags, the output voltage starts to droop. The problem is due to not enough input capacitance and a limited duty cycle. 

With only 69 microfarads after the bridge rectifier, the input voltage sags below 90 volts in between line cycles, which demands over 50% duty cycle. However the duty cycle of a two switch flyback is limited to a maximum of 50%, which in turn limits the power delivered to the output during the valleys of the rectified AC voltage. 

Increasing the amount of bulk input capacitor reduces the sag on the rectified AC voltage as shown on the left. With a total of 136 microfarads, the supply can operate down to an input voltage of 90 volts AC, 60 hertz, with no noticeable 120 hertz ripple on the output. This keeps the minimum rectified AC voltage above 100 volts. 

Alternatively, the terms ratio of the transformer could also be adjusted to keep the duty cycle below 50% at lower input voltages. The bulk input capacitors are often the largest components in a power supply. It is obviously desirable to minimize the amount of capacitance to reduce the size. The chart shown on the right shows the amount of capacitance needed to keep the rectified voltage from sagging below 80 volts with a 90 volts AC 50 Hertz input. 

A good simple rule to follow is to have at least 1.5 microfarads per watt of bulk capacitance to support your minimum input voltage requirements. Also shown in this plot is the 100 hertz ripple current in the capacitor versus power. You need to be careful not to exceed the ripple current rating of your capacitor. 

The bulk capacitor needs to be sized not just for the ripple at twice the line frequency but also the ripple current created by the power supply at the switching frequency, usually in the hundreds of kilohertz. Aluminum electrolytic capacitors are usually the least reliable component in the power supply. The lifetime of the capacitors typically determines the lifetime of the entire product. 

Capacitor life doubles for every 10 degrees C decrease in temperature. So the stated life, such as 2,000 hours, assumes operating at a rated temperature and ripple current. Designing for low failure rates is possible by using capacitors rated well above their maximum operating temperature and ripple current. 

There is a consistent theme underwriting each of the examples that we've looked at. In each case, the problem can only be solved if the right information is examined. Gathering and evaluating data is critical for fast debugging of any problem. This is true in power supply design just as it is true in life. 

The key points for each common mistake are highlighted here. You can also refer to our associated white paper for this topic for more details and references. Often we gain the most knowledge from our mistakes. With power supply design, there's plenty of opportunity to learn. Hopefully, the information provided here and in the paper will allow you to learn from the mistakes of others.