Forward Converter Design - Part VIII Testing Schedule
Testing and development schedule for the forward converter.
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 VIII, the prototype power supply is connected to a real system load and several issues are encountered.
Power Supply Design Schedule
So far, this series of articles has taken 7 months to present . The actual uninterrupted time spent on the power supply testing and redesign was approximately 160 hours, following the initial completion of the PC board. Many issues have been resolved. Three iterations of the output inductor and four iterations of the power transformer have been built.
There are still issues remaining to be tested, including thermal performance, EMI, and proper control loop design. An estimated 100 hours of testing and data collection will be needed for this phase.
The hours that you will need for your project will vary significantly from this, depending on many factors:
1. How much power is being processed.
2. Input voltage range.
3. Design experience level.
4. Prior experience with the specific technology and topology chosen.
I have observed in the past that a good power supply designer will be able to get to production with perhaps three iterations of the printed circuit board. The first iteration should catch the critical issues that compromise the ruggedness of the power supply. As much data as possible should be collected from each board.
The second iteration of the board should be near production-level, depending on the extent of changes needed. The second board will typically be used for thorough thermal testing, EMI testing, and system testing.
The final board will fix any issues encountered to produce a manufactured product. Only minor changes should be needed on the final iteration.
PC board design time can be very variable. For the best designs, the power supply design engineer is intimately involved in the layout process, physically placing all of the critical components manually, and specifying the critical layout paths, power planes, and spacing.
At this stage of the design, the power supply has been tested at full load, and across the full range of input line. Many changes have been made to the board to address the kind of issues that arise with virtually all switching power supplies running off-line. Most of the issues have been related to the first 200 ns of turn-on of the switch. Controlling overcurrent and overvoltage are paramount to making the power supply rugged. This applies to just about every power supply that is developed.
This is a reasonable point in the project to stop and implement a new PC board design. However, I have always found it very valuable to do preliminary testing with the real system as early as possible. There are always unpredictable issues that can be encountered.
Figure 1 shows the prototype power supply connected to a power amplifier, the intended load. Remember this is a Revision 1 board, designed for stand-alone testing, so there are lots of wires to connect power supply and load. It is not uncommon for the early prototype to look distinctly low-tech, which sometimes alarms engineering managers who lack experience in power. While the setup is not particularly attractive, there is great value in performing this kind of testing early on in the power supply development.
Several issues were encountered in this project:
1. Power supply would not start when connected with low-impedance source and real load.
2. RCD clamp needed redesign.
3. Common-mode EMI interfered with amplifier circuitry, resulting in audible noise.
4. Actual load requirements were significantly lower than original specification.
Continue reading this article?