[34] Forward Converter Design - Part I Introduction

Introduction to the series on how to design a forward converter.



This article begins a new series in which Dr. Ridley shows the multiple testing and revision steps needed to get a power supply from concept to a fully-working prototype.


Power Supply Development Time

Whenever I look at doing an off-line power supply development, I have a formula for calculating how long it will take. I lay out each step of the process, and estimate the maximum amount of time that each step might take. Then I double the result. After this, I am normally hard-pressed to finish on time.

Why does this happen? It’s because there are always unforeseen events that can drastically slow down the development process. After many years of working with switching power supplies, you learn that parts seldom behave exactly as you expect them to, and when designing for a rugged environment of line variations, surges, short-circuits, and temperature variations, you can never be sure how long things will take.

It is hard to explain to those new to power supply design exactly what can go wrong. It is also hard to explain to management why the project schedule estimate is so long, and why we might need three or more board turns to get such a “simple” circuit right. Part of the reason for this difficulty is that when solving the problems of a power supply, we are often in a hurry and rarely have time to fully document the steps involved in fixing problems. Later on, we forget exactly what happened to take so long.

I am guilty of following the same process. When working on a consulting design, or on a design for my own projects, the main objective is to be done as quickly as possible. Recently, however, I had a unique opportunity to take the time to fully document the process steps in going from a power supply that was already laid out and ready for testing, to getting the supply up to full power. In following this process, I encountered many events I had seen in the past, and never documented.


Power Supply Requirements

The specification for the power supply was as follows:

1. Output 1 – 35 VDC @ 10A isolated

2. Output 2 – 35 VDC @ 10 A isolated

3. Maximum power 350 W (only one output fully loaded at a time, application is for audio.)

4. Input – 180 – 265 AC

5. Power Topology: Two-switch forward

6. Controller: Digital controller from TI

The project was launched for several purposes. Firstly, I needed a bulk supply for an audio product which is currently running from a linear supply. Secondly, I had a digital designer who was eager to work for me for four months to try and learn some real-world power supply design. I have watched the promise of digital for over 20 years now, and felt is was finally time to see whether it could really work in a real-world product. And finally, I have always recognized the two-switch forward as the most rugged topology available, but never actually built one from scratch. It was time to learn the details of the topology first hand.

The optimistic goal of the project was to finish in plenty of time so that a second converter with the same specifications could be built to run from the one digital controller. I didn’t really expect this to happen in the given time frame, but it was kept as a goal. At the end of the project, I hoped to have a working prototype, learn a lot about digital controllers and their quirks, and to impart some knowledge on analog power design.

As is often the case in the power supply world, things didn’t quite work out as planned. At the end of the four months, I found that I still couldn’t program a digital controller (although I learned a lot), and a digital designer couldn’t possible learn enough about high-voltage analog design in such a short time. However, I learned a tremendous amount about our analog design process and why it remains a challenge after so many years. The digital designer returned to the world of low-power offline work, and he learned more about his digital algorithms but not, unfortunately, what makes the analog world go round.   



Separation of Power and Control

The first major lesson when working with a digital controller is that the layout needs are totally different from the needs of the analog power stage. The digital part of the board requires fine-pitch traces, and at least a four-layer board. On the other hand, the power stage requires heavy copper for high-current traces, large trace separation for voltage breakdown requirements, and pleny of allowance for thermal management, and bulky magnetics.  And, as we shall see later in this series, you must be able to rework the power board manually to arrive at a well-designed converter in the shortest amount of time.

I always work with just two layers for power supplie. I like to be able to see all of the traces on the top and bottom sides to ensure proper spacing for all components. I also like to have the ability to move a trace if necessary, and relocate critical components without having to layout a new circuit board.

In early development, the most practical thing to do is separate the controller and power stage into two boards with a connector between them. Digital control forces you to this approach. This is also an approach I have always taken for analog power development. A controller from one project is often very applicable for your next project, and can be taken as a partial pc board and interfaced with a new power stage. This can save a tremendous amount of time, and is easy to do if you have a clear delineation between power and control.

article34 01

Figure 1: Separation of Power Circuit and Control Circuit with Interface Signals

During the development of most power converter, the control and power stage delineation becomes blurred with more and more communication from one to the other. The addition of ground planes confuses the line even more. This can lead to multiple problems with noise.

Ground and other PCB planes should be completely separate between the control board and power board. This is naturally achieved when the two boards are built separately, as is recommended for the early stages of digital control. This proper separation of the power and control boards will be discussed in more detail in the next article of this design series.



Full Schematic and Layout

The full schematic of the forward converter will be presented in the next article of this series. The power layout took about 1 month to complete, including the entry of all component footprints and placement of parts and traces. When working with existing parts databases, I usually expect this process to take me about two weeks. Less experienced designers can easily spend eight weeks on this process if starting from scratch.

The layout alone is a topic worthy of a complete series of articles. I won’t go over that in full detail in this series, and will concentrate instead on the issues involved in completing the design after the first layout was complete.



Continue reading this article?



contact us

Our worldwide HQ
601 E Daily Drive
Suite 112
Camarillo, CA 93010 USA

Phone US: +1 805 504 2212
Phone UK: + 44 (0)1509 276 245
Fax +1 888 339 6419
Email: info@ridleyengineering.com

Contact Us


Get Advice