[033] Testing Off-Line Power Supplies

How to safely test your switching power supplies. 

Introduction

In this article, Dr. Ridley shows how to safely test off-line power supplies, without the use of expensive (and sometimes hazardous) high-voltage dc power sources.

Off-Line Power Supply Design and Testing

Off-line power supplies can be dangerous circuits to test and debug, especially at high power levels. Standard off-line designs generate dc input voltages up to 424 VDC after the rectifier – a voltage level that can be hazardous. Proper testing of these power supplies is essential to ensure a reliable product, and this testing must be thorough and safe.

Figure 1: Offline power supply with full-bridge rectifier on input.

Figure 1 shows the typical configuration of an offline power supply, omitting the input filter for simplicity. The ac line is rectified through a full bridge, producing a dc input voltage equal to the peak of the ac input. For a 240 VAC nominal input specification, experienced designers will test their power supplies at up to 300 VAC to ensure proper design margins to cope with real-world variations.

With the circuit configured as shown, there is a problem with testing the power supply. Much of the control circuitry is referenced to the return of the input rail. However, this circuit node is connected to the neutral of the input ac supply for half of the input line cycle, and to the live of the input ac supply for the other half of the line cycle. There is no ground reference for the circuit, making waveforms difficult to measure. 

Floating the Oscilloscope

One solution to the problem is to put a “cheater” plug on the oscilloscope, and allow it to “float” at whatever voltage the connection requires. You can certainly do this, albeit very carefully. The oscilloscope will become a part of your system, and will place a complex (high-impedance) load at the connected node of the circuit. Don’t ever connect it to a high-frequency node, such as the drain of the FET of the flyback circuit shown. This will have a drastic effect on circuit waveforms, leading to unreliable measurements or possible failures.

It is acceptable to connect the ground reference of the scope to the negative rail. First, however, you must be aware that the entire body of the scope, and the probe grounds of the other input channels, are then referenced to  hazardous voltages. I have done this in the past, but it is always an uncomfortable experience touching the scope carefully and avoiding any metal parts of the instrument.

Differential probes can also be used, but these can be inaccurate when the common-mode voltage is moving by hundreds of volts over a cycle. Failures of semiconductors can be due to events which last for 100 ns or less, and these can be inaccurately observed with ungrounded systems.

Using a High-Voltage DC Lab Supply

Another solution for power supply testing is to use a dc input source. You can purchase lab supplies which have an output voltage range of 500 to 600 VDC. My past experiences have made me wary of this type of supply. There are not many vendors that like to produce such products since the outputs are very dangerous. Also, if you buy a 1 kW supply at 600 V, it can only produce 1.7 A. For low-line testing, this will be inadequate, giving you only 250 W capability at 150 VDC input, the range where many power supply designs will try to get started. Inrush requirements may require that you have a 5 kW dc lab source, just to test a 1 kW power supply design. Such bench supplies are expensive if you can find them.