Advancements in power conversion components have far outpaced the ability to accurately measure and characterise these designs.
Engineers working on power device designs involving GaN and SiC technologies and other high-speed applications currently have no way to accurately measure differential signals when common mode interference is present. As a result, these signals are essentially hidden, making it difficult for engineers to see what is actually occurring inside circuits, slowing debug and characterisation efforts.
In many cases, designers end up with misleading information on their scope. Because they can’t make the measurement, they turn to alternative methods to try and make sense of what is actually occurring in the circuit.
Common alternatives are extensive simulation and measuring adjacent points and extrapolating the results which in many cases doesn’t reflect what is actually occurring. The consequences can be dire, meaning the circuit could either fail or operate sub-optimally.
In particular, engineers are struggling to evaluate the performance of their gate drivers on both half-bridge or full-bridge designs where a small differential signal is floating at a high common-mode voltage (e.g. 40 V or higher). In cases where the switch node voltage is switching between “ground” and the input supply voltage, the gate-source voltage is impossible to measure without adequate common mode rejection.
For these types of measurements, the key performance parameter of the measurement system is the common-mode rejection ratio (CMRR). This determines a measurement system’s ability to accurately measure the true differential signal by rejecting any signal that is common to both test points (i.e. the common mode signal).
CMRR is defined by:
CMRR = A(diff)/A(cm)
A(diff) = the voltage gain for the difference signal
A(cm) = the voltage gain for common-mode signal
Ideally, A(diff) would be large and A(cm) would be zero, resulting in an infinite CMRR. In practice, a CMRR of at least 80 dB (10,000:1) is considered quite good.
An amplifier such as the LeCroy DA1855A, for example, has a CMRR that exceeds the 80 dB level at low frequencies up to a few MHz. However, the CMRR capability of this amplifier quickly derates and is only 20 dB or 10:1 at 100 MHz. What this means is that a common-mode input signal of 10 volts at 100 MHz will induce a 1 V error signal in the differential measurement.
It should be noted that the plot is for the amplifier only as performance further degrades when using probes with the amplifier.
A new approach
When the best amplifier available fails to deliver repeatable results, it’s clear that traditional probe architectures can no longer keep pace for this application. What is needed is a new measurement system with the ability to completely reject the common mode signal and reveal the true differential signal.
For probing the DUT, IsoVu from Tektronix uses an electro-optic sensor to convert the input signal to an optical signal. This electrically isolates the device-under-test from the oscilloscope. IsoVu incorporates four separate lasers, an optical sensor, five optical fibres, and employs feedback and control techniques.
The probe’s galvanic isolation provides common mode withstand voltages of > 2000 Vpeak across its frequency range with no derating. The electrical limitation for an optically isolated probe is many thousands of volts. Because galvanic isolation is achieved through its fibre optic connection, the only limitation in its common mode voltage rating is due to safety certification standards.
The sensor head, which connects to the test point, has complete electrical isolation and is powered over one of the optical fibres. It is significant to note that there are no batteries required. The sensor head also contains a feedback loop that measures the DUT signal and sends it to the controller for analysis. This allows the system to correct for drift and offset errors.
The connection from the DUT to the sensor head is through tip cables, which have five different attenuation ranges to extend the measurement range. The tip cable connects to the sensor head via an SMA connector and includes readout encoding so the sensor head can communicate the attenuation factor to the scope to display the correct vertical scale factor.
The probe achieves exceptional CMRR over the entire operating range due to its complete galvanic isolation and sensor head architecture.
Engineers working on power device designs involving GaN and SiC technologies and other high-speed applications struggle to accurately measure differential signals when common mode interference is present. To address this challenge, there are advantages to using an optical connection to the oscilloscope, effectively eliminating CMRR as an obstacle, allowing repeatable, accurate measurements in the presence of large common mode interference.
Dean Miles is a senior EMEA Technical Marketing Manager at Tektronix