Signal Injection Transformers – What makes them special?
In this article we’ll take a look at what makes the signal injection transformer special, why an audio or video transformer is not a suitable alternative for most applications and why the injection transformer is not suitable for measuring input ripple rejection (PSRR) or other measurements that allow DC current to flow through the transformer.
An injection transformer is a special type of transformer that is connected between a network analyzer and a DC?DC converter or voltage regulator in order to both inject a perturbing signal into the control loop and to record the response of the loop over a wide frequency range. The purpose for using a transformer is to measure the phase?gain response of the control loop while minimally impacting the loop performance.
Figure 1: Example schematic showing the test setup connectivity for measuring the Bode response of a voltage regulator. The transformer connects to a network analyzer.
An example of a voltage regulator stability test setup is shown in Figure 1. The output of a network analyzer (Source), an oscillating signal of small amplitude and varying frequency, stimulates the voltage regulator’s control loop across R1, via the transformer. R1 is generally referred to as the injection resistor. It effectively breaks the regulator’s control loop allowing the two channels of the analyzer to compare and generate the magnitude and phase the loop, i.e. Vout/Vin. The circuit response is monitored on either side of R1 via the CH1 and CH2 connections, again, through the transformer. The undistorted transmission of these signals, their levels and the proper setup and connection of the circuit are paramount to a successful outcome.
There is not much written about injection transformers, which is easily verified via a quick internet search. Much of what is published discusses why an expensive transformer is not necessary. Much of what is written, even by manufacturers of the voltage regulators and other power ICs, is remarkably incorrect.
Click here to read the entire article.