INTRODUCTION
The stability of a control loop determines the closed loop
performance of most key aspects of the power supply.
An unstable power supply can oscillate, resulting in very large
ripple, poor step load response, and excessive noise and EMI.
A power supply does not have to be fully oscillating to be
inadequate. The control loop stability controls PSRR, dynamic
response and output impedance. Poor stability means that all
these responses will also be poor.
Because the performance is not simply pass or fail, and
changes with part tolerances over the life of the power supply, it
is extremely important to understand the initial performance.
The Bode plot is the generally accepted method for assessing
the stability of a loop. A Frequency Response Analyzer, or FRA,
is used to measure the Control Loop Response, popularly
known as Bode plot, and the Power Supply Rejection Ratio
(PSRR); two key metrics of power supply performance.
The Bode plot is ideally measured with a FRA but now, due to
the addition of specialized software, can also be obtained using
oscilloscopes, along with a built-in or external function
generator.
The control loop response computes and plots gain as 20 log
(Vout / Vin). Phase margin is measured at the gain crossover
frequency, which occurs at the frequency where the gain plot
crosses 0 db. The corresponding point on the phase plot gives
the phase margin. The Gain Margin is calculated using the
Phase Margin.
In this app note the Picotest VRTS1.5 demo board (adjustable
regulator board) is used to demonstrate this test.
An injection transformer is used for connecting the FRA to the circuit
being tested for control loop stability measurements. The goal of the
transformer is to inject a signal into the control loop being
measured, without impacting the performance of the loop. In order
to accomplish this to a reasonable degree, the transformer is
isolated and, therefore, is capable of floating on a high voltage, such
as a Power Factor Corrector (PFC), which is often close to 400VDC.
The usable bandwidth of an injection transformer is significantly
greater than the 3dB frequency limits. Because the transformer
bandwidth itself is outside of the measurement, it leads many to
incorrectly believe that the transformer is a non-critical element.
The bandwidth of the transformer is strongly related to the
terminating impedance. Most network analyzers provide a 50Ohm
oscillator source impedance. Assuming this impedance, the optimal
termination resistor is 5 Ohms. This significantly attenuates the
injection signal, which is beneficial, as a common error in Bode
measurements is using a signal which is too large, and therefore not
a small signal measurement. An added benefit of this low value is
that it can always be left in circuit, simplifying the connection to the
network analyzer or scope without appreciably impacting the output
voltage of the circuit being tested.
Picotest supports the oscilloscope stability testing with three
injection transformer options for different circuit applications as
discussed below.