[040] Forward Converter Design - Part VII Full Power Cross-Regulation
Initial full-power testing of the forward converter.
Introduction
This article continues the series in which Dr. Ridley documents the processes involved in taking a power supply from the initial design to the full-power prototype. In Part VII, the converter will operate at full power, and cross-regulation of the outputs will be tested.
Full Power Testing
As a reminder from Part I of this series of articles, the power supply specifications are as follows:
1. Output 1 – 35 VDC @ 10A isolated
2. Output 2 – 35 VDC @ 10 A isolated
3. Output 3 – Bias Supply, 12 W fixed load, ground referenced.
4. Maximum power 350 W (only one output fully loaded at a time, application is for audio.)
5. Input – 180 – 265 AC
Regulation on the outputs is not tightly specified. This design was for an audio application to replace an unregulated power supply with a 50/60 Hz transformer. As you can see in Figure 1, the three outputs of the converter are all on one common inductor core. This coupled-inductor approach provides the best regulation of the outputs, and is a very useful technique. With both outputs isolated, the intent was to regulate the converter from the grounded bias output.
Notice that the turns ratios on the inductor must be exactly equal to the secondary turns ratios on the transformer in order for this scheme to work properly.
Prediction of cross-regulation cannot be done theoretically, and the first major task after getting all of the semiconductors protected and prototype flaws fixed is to test the output regulation under all conditions.
Unprotected Diode Failure
The last three parts of this series [1] described how the semiconductors were protected with clamps, snubbers, and proper current limiting to make the converter rugged. This was implemented with low voltage applied to the converter. After this protection was properly designed, the converter was ready for full power testing.
Unfortunately, when the input voltage was raised to 200 VAC with 300 W load applied to the output , there was a failure. The primary switch current climbed rapidly each time the switch was turned on, there were indications of magnetics saturation (limited by the current-limit circuit), and the output diode on the bias winding became extremely hot.
There are a total of six output diodes in this converter, two for each output. On the main outputs, an RCD clamp protected the catch diode, and an RC snubber sufficed to protect the forward diode, as shown in Figure 1.
Figure 1: Three-output forward converter with secondary snubbers and clamps.
The catch diode on the bias output (15 V and 12 W only) had an RC snubber, but the snubber across the forward diode, shown in green in Figure 1, was omitted. It was realized that the RC snubber had not been placed on the board. The low power of the output led to a common trap—not fully testing that particular diode, and assuming that the overrating of the diode was sufficient to protect it.
Figure 2 shows the ringing across the forward diode on the bias winding output with approximately 100 VAC applied to the primary of the converter (maximum voltage rating is 280 VAC). A peak voltage of 160 V is seen on the waveform, more than five times higher than the anticipated square wave voltage of 35 V.
This is a common mistake made in many production designs. The bias windings are often very low power, and high voltage diodes are commonly used, without any snubbers. However, the peak ringing voltages can be extremely high (as seen here), and must be properly suppressed.
Figure 3 shows the diode voltage after an RC snubber was added. Dissipation in the snubber was very low, less than ¼ W, but the snubber was very effective at suppressing the ringing. Once this diode was protected properly, there were no more failures during this phase of testing.
It is interesting to note that the schottky diode fails in short-circuit mode. The current limiting in the primary prevents enough power being delivered to fuse the component. You should always remember that schottkys are not at all tolerant of overvoltage, and there should always be a good margin for a rugged design. Just to reiterate – make sure you test every semiconductor for peak voltage and current stress, regardless of how far within the ratings they may seem to be.
Figure 2: Secondary forward diode voltage waveform, VD, with 100 VAC applied and no snubber