Three Probing Techniques To Improve Power Conversion Testing With Oscilloscopes

punjolife(60)

Published in

#steemstem

Words

1423

Reading

7 min

Listen

Play

Today's power designers and test engineers are struggling to find very small, incremental improvements to improve power conversion efficiency or reduce design losses. This requires the ability to accurately evaluate and measure very small performance improvements.

Measuring Instrument Oscilloscope
Image credit

There is a source of losses in almost every part of the power converter, which often includes switching semiconductors, magnetic components, and rectifiers. Even a few percent improvement in performance, or even less than a hundred percent, could be of great significance. In order to accurately evaluate and measure such a small increase in performance, abnormal accurate measurement is crucial.

Most oscilloscopes come with a 10X attenuated passive probe because the probe is suitable for a wide range of applications. The nominal bandwidth of these transducers is typically DC ~ 500 MHz and typically measures voltages up to several hundred volts. Of course it is possible to use general purpose probes for power measurements, but it is not possible to provide the required accuracy to these transducers designed for power applications to drive improved power conversion performance.

Signal sensitivity

Let's take a look at the example of a generic probe that has a short board. A common challenge in power supply design and measurement is to isolate noise from ripple voltage. In this example, we are going to use a generic 10X probe to probe the 3.3 V supply. The problem is that the 10X probe does not provide enough sensitivity to trigger the periodic noise present in the waveform. These probes are ideal for many general-purpose electronic measurements because they increase the oscilloscope's voltage range and provide a relatively high bandwidth.

However, to measure small tens of millivolts, a 1: 1 (1X) probe would be a better choice because it caused less signal attenuation and did not push the signal down to the oscilloscope's noise floor. Unfortunately, this sensitivity advantage is offset by its bandwidth disadvantage, which is typically only around 15 MHz. If this measurement of bandwidth is not enough, then it is best to use a passive 2X probe.

Proven in this application, 2X probe is the right choice. Take a look at the waveforms in Figure 1. The yellow track is a 10X probe that adjusts to the lowest vertical setting of 10 mV per cell; the blue waveform is a 2X probe. The 2X probe can be adjusted to a minimum vertical setting of 2 mV per cell. Since the power output produces a 3 mV ripple signal, it is clear that a 10X attenuated probe is not well suited for this measurement.

Figure 1. Measure the 3.3 V power supply using a 2X probe (blue trace) and a 10X probe (yellow trace).
Image credit

Differential measurement

The ripple measurement discussed above is just one of many applications that can safely and efficiently use single-ended (reference ground level) probes in power supply design and debug. However, many functional conversion measurements are to be made in a floating environment where the ground level can not be referenced.

Figure 2 shows a variety of common power conversion measurements that are not tied to ground and require differential measurement techniques:

The drain-to-source voltage (VDS) on the MOSFET
Diode voltage on freewheeling diode
Inductance and transformer voltage
Voltage drop in ungrounded resistor

Figure 2. Part of the differential measurement point on the push / pull power converter.
Image credit

Differential measurements can be performed in a number of ways, including:

Use two single-ended probes to calculate the voltage difference
Use an oscilloscope with a specially designed floating input
Select the differential probe that best matches the measurement

Use two single-ended probes

One commonly used technique is to use two single-ended probes, each probe grounded to ground and pointed on either side of the device under test, as shown in Figure 3. Then set the oscilloscope to display the difference between channel 1 and channel 2. This is sometimes referred to as "A-B," which uses math in the oscilloscope to show the voltage difference between the two channels. Engineers sometimes use this technique when differential measurements are required, but without the proper test equipment.

Figure 3. Quasi-differential measurement using two single-ended probes.
Image credit

There are several problems with this method. This method yields good measurements only when the probe and oscilloscope channels are well matched (including gain, offset, delay, and frequency response). This method does not provide very good common mode rejection (clearing any AC or DC part of the two input common signals). In addition, if the two signals are not properly scaled, the oscilloscope input overload may occur, resulting in erroneous measurements.

Use floating input

We can also use "floating" oscilloscopes. Each of these oscilloscope's input channels is electrically isolated from the chassis ground and the oscilloscope is battery-powered. The parasitic capacitance of the oscilloscope chassis to ground is also very low. These isolation features of a floating oscilloscope allow the use of an isolated, passive probe for differential measurements. These instruments are very convenient, easy to use and effective. However, the differential voltage probe has a low capacitance and requires a high degree of balance.

Matching differential probe

For the best measurement accuracy, using a differential probe whose specifications match the measurement tasks is usually the best choice. Differential probes are active devices. They have a specially designed differential amplifier at the probe tip that measures only the voltage across two test points regardless of the potential between any test point and ground, which greatly simplifies the probing tasks and eliminates some of the potential The source of error. In addition, since they measure only differential voltages, they can also ignore and eliminate any common mode AC swing or DC bias voltage that may be present.

Since the measurement of different parts of the device under test (DUT) can have quite different requirements, the probe must be chosen carefully. In the example shown in Figure 4, the task at hand is to measure the turn-on, turn-off, and conduction losses of the power MOSFET switch under test. Figure 4 is a simplified schematic of a MOSFET with measurement points TP1 and TP2.

Figure 4. Simplified schematic of a MOSFET with multiple test points.
Image credit

The device under test is a "universal" power supply that is designed to supply voltage from AC lines (or "mains") all over the world. Only this one, give the engineer's test requirements and test equipment made a number of requirements:

The nominal input voltage for this device is typically 80 VAC to 250 VAC or wider. To characterize performance at various input voltages worldwide, not only one measurement but also a series of measurements at various input voltages is performed. This applies to every performance parameter being tested. Switching characteristics (and corresponding losses) are expected to vary across each input voltage and may not change linearly. This not only increases the total number of measurements to be performed, but it also requires repeatability between the measurements.
Due to the input supply voltage up to 250 VAC, the voltage between the drain and the source of the switching MOSFET is expected to reach 354 V or higher. Probing solutions must have enough versatility to measure these voltages and, in some tests, be able to measure much lower voltages.

The tested power supply has a switching speed of 250 kHz. According to the commonly used 5-fold rule of measurement bandwidth, this equates to a measurement bandwidth of 1.25 MHz. But this is a simplified version of the real-world signal speed because the actual rise time of a switching device is expected to exceed it by an order of magnitude. Again, spikes, transients, and other noises may be explored. If you want to measure a rise time of tens of nanoseconds, then the probe's rise time should be a few nanoseconds. For accurate measurements in this application example, the measurement system bandwidth should be 350 MHz or higher.
summary

The best choice of probe is closely related to the application, so it is important to understand the measurement requirements of your application to ensure that the probe is fully adapted to your job. For many power electronic measurements, the differential probe is a clear choice, especially without reference to ground level measurements. For reference level measurements, single-ended probes are a good choice, but be careful not to use 10X probes to avoid over-attenuating small signals. For low-voltage signals, such as ripple, the best use of 1X probe or 2X probe.

==========

References used:

Support @steemstem and the #steemstem
project - curating and supporting quality STEM
related content on Steemit