(Sharing haberdashery) Oscilloscope-based power circuit response test circuit

Designing loop compensation has always been a major challenge in switching power supply industry. Novice engineers always have difficulty learning or debugging a loop design. Even most experienced power supply engineers have only half knowledge of loop compensation.

The main reason is that most engineers have never actually tested response of a switching power supply circuit (including gain curve and phase curve).

Of course, corresponding stability parameters of loop can be calculated using theoretical calculations or software simulations, but since calculation is complex and accuracy is not high, practical application is limited.

Most experienced engineers change loop characteristics based on output performance of product (such as load dynamic response, product noise, etc.) and try various compensation methods when loop stability problems occur. Then evaluate if compensation is working. whether method used is effective or not, depending on whether output performance test results improve.

However, using this method will cause many problems. First, there is no way to take targeted remedial countermeasures based on actual loop gain curve. project cannot be completed on time.

It is also impossible to judge whether phase margin is appropriate by inferring performance figures, so reliability of designed product cannot be guaranteed.

For another example, no actual loop test is performed, and actual changes in gain curve and phase curve cannot be known after each compensation change. Loop compensation design is practiced with little understanding of its basics.

Loop response testing is so important, why don't many electrical companies do a loop response test? The reason is mainly that traditional loop response test equipment (network analyzer) is expensive, and for power supply industry, utilization rate of this equipment is very low.

Based on above issues, today I will introduce you most innovative and cost-effective loop response test solution on market, an oscilloscope-based power loop response test solution.

1. Schematic diagram of test

1. Disconnect voltage divider resistor from output voltage and connect a 5-50 ohm injection resistor in series;

2. Use an oscilloscope signal generator to connect to both ends of injection resistor through an isolation transformer;

3. The two channels of oscilloscope measure respectively voltage from top end of injection resistor to ground (output voltage) and voltage from bottom end of injection resistor to ground (Vin of transfer function).ii). );

4. Use oscilloscope's built-in loop response test software to run an automatic test to measure gain curve and phase of switching power supply.

2. Analysis of test results and basis for judgments

a. Crossover frequency (corresponding frequency at 0 dB gain): 5-20% of switching frequency is recommended

b. Phase margin (corresponding phase when gain is 0 dB): requirement must be greater than 45°, 45°-80° is recommended

c. Crossover Slope (about 0dB): Single-pole crossover required (-20dB per decade of slope crosses 0dB)

d. Gain headroom (corresponding gain at 0° phase): recommended not to exceed -10dB

e. Gain attenuation (gain corresponding to switching frequency): Recommended value is less than -20dB

3. Single frequency point test mode

In addition to simple test, low cost, and high utilization rate, oscilloscope loop response test solution also provides single frequency point test function (time domain waveform observation), and supply voltage setting can be easily estimated by single frequency point test This is reasonable.

Whether or not distortion is caused by excessive injection voltage, this is an advantage that no other loop response test solution has.

Perform a single point test corresponding to crossover frequency. The amplitude of input signal and output signal are equal, gain is 0 dB, signal is not distorted.

If supply voltage is set too high, using an oscilloscope can easily determine if signal is distorted, as shown in figure above. If signal is distorted at same frequency, gain will not be 0 dB. which will lead to incorrect test results.

4. Comparison with network analyzer test results

Near crossover frequency, amplitude of injected and output signals is usually tens of mV, and accuracy of oscilloscope measurement is completely trustworthy. There were practically no differences in test results.