Designing power supplies is one of the fundamental tasks of an analog electrical engineer. Finding the right part to deliver optimal efficiency, noise, control, and size characteristics can make or break a design. An oscilloscope is one of the best tools for verifying a power supply design, and this article will show how a Tektronics MDO 3000 can test:
- Power supply efficiency
- Switching waveform analysis on the switching FET
- RF emissions from the power supply.
In the video below, I look at TI’s BQ27650EVM , which is the eval board for TI's BQ24650 switch-mode solar battery charge chip. I found that it was very efficient, had a lot of ringing on the switching signal (that I may have created myself with the deviation), and there was a fair bit of RF emissions with lots of harmonics. Check it out!
There are a couple pieces of equipment I used in running these tests that work in concert with the MDO3000:
- Tek TPP1000 Voltage Probe (which comes standard with the MDO3000).
- Tek TCP0030 Current Probe , which is a clamp-type sensor to easily measure current.
- Tek TDP1000 Differential Voltage Probe , which has an isolated reference which can float away from ground, in addition to a number of probe tips perfect for making short connections to the circuit.
- Tek MDO3PWR application module to help run analysis on the switching waveform.
- Beehive 100C, which is a near-field loop antenna for detecting RF signals.
How to Test Power Supply Efficiency
Efficiency is almost always a concern in a power supply design. If battery life isn’t a key design parameter, then heat dissipation likely is. Thankfully, this is also the easiest test to run, because
Therefore all one has to do is measure the voltage and current at both the input and output. Here’s a step-by-step of how I use the MDO3000 to measure efficiency of the BQ24650:
- Turn on the MDO3000, compensate the voltage probe, and degauss the current probe.
- Connect a battery (to act as the load) and a power supply (to act as the energy source) as shown in the wiring diagram.
- Connect the voltage probe across the positive and negative input terminals, and clamp the current probe around one of the input wires (making sure the arrow of the current probe points in the direction of current flow) as shown in the wiring.
- Turn on the power supply.
- Zoom in on the voltage and current signal, which should be easy since they will be DC signals
. - Turn on the MATH signal, set it for MATH = Ch1 * Ch2 using the bezel buttons, and zoom in on the signal (see video for button presses).
- Enable measurements for the ‘mean’ of Ch1, Ch2, and MATH.
- Wait for the mean to settle, and then record all input measurements.
- Move the voltage and current probe to the output of the circuit as shown in the wiring diagram.
- Wait for the mean to settle, and then record all output measurements.
- Compute % efficiency by dividing the output power by the input power. In the video, I measured the efficiency at 2.0W out / 2.4W in, or 83%.
Note that the input and output power need to be measured almost at the same time for this to be valid. In this example, the input and output power can change as the battery charges which would skew the data.
Switching Waveform Analysis:
The switching waveform that shows the drain current through the FET and Vds across the FET can give insight into switching losses, saturation of inductors, peak current levels, and unexpected transients such as ringing. Unfortunately, it is a more complicated measurement to take. As the video shows, wiring in the current probe isn’t easy and will create a large current loop, ruining what could otherwise be a nice layout.
Here’s a step-by-step of how I setup the MDO3000 to look at the switching waveform:
- Power on the MDO3000, compensate the differential voltage probe, and degauss the current probe.
- With the unit not connected to the supply or battery, cut the trace on the PCB that feeds the drain of the MOSFET. Then use a loop of wire large enough to accommodate the current probe clamp to re-connect the drain of the MOSFET (see my setup in the video). Connect the current probe so that there is minimal force being applied to the fragile loop of wire (it is only held on by the solder joints).
- Solder the solderable probe tip for the differential probe across the MOSFET’s drain and source. Connect the diff probe to the probe tip.
- Connect a battery (to act as the load) and a power supply (to act as the energy source) and turn on the power supply.
- Zoom in on the differential voltage and current waveform.
- Note the shape of the current waveform, the voltage drop across Vds, any unexpected transients/ringing, and the ringing frequency as I do in the video.
- To access the MDO3PWR application module for automated analysis of switching losses:
- Press ‘TEST’ near the PAN/ZOOM knob
- Select the ‘POWER ANALYSIS’ application module with the bezel button
- Select ‘Switching Losses’ with the bezel button
- Set the voltage input to the diff probe, and the current input to the current probe using the bezel buttons
- Note the number of periods that are captured, and the resulting data calculated by the scope. My test resulted in ‘Low Resolution’ notifications for some of the data, which was likely caused by all of the ringing on the waveform.
Looking at RF emissions with the Spectrum Analyzer:
Sniffing for RF emissions on the bench won’t get you CE, FCC, or any other certification, however it can be a great ‘oh-sh*t’ test to see if the system will at least stand a chance. As the video shows, when the layout is botched for the current probe clamp, there was about 20dBm more of RF noise emitted from the circuit. Having a proper layout is often the difference between passing certification tests and being slammed with the cost and delay of failure. It is also a handy tool when looking for the source of noise in the circuit – if there is enough noise to show up on the power supply line, it will likely be radiated from the problem circuit as well.
My test used the Beehive probe as the input to the MDO3000. Here’s how I got the data:
- Connect the behive probe to a BNC cable, and the BNC to the input to the spectrum analyzer through an N-to-BNC converter , which may be included with your MDO3000.
- Set the x-axis of the spectrum analyzer by pressing the ‘Freq/Span’ button and then the bezel buttons for the frequency range and center frequency of the display.
- Wave the beehive probe around the different circuit components, ‘sniffing’ for RF emissions.
- If needed, adjust the y-axis settings by pressing the ‘AMPL’ button and using the bezel buttons to set the reference point, and the general purpose knob for setting the division size.
- Once a noise source is found, note the main frequency, its maximum amplitude, and any harmonics.
- Adjust the x-axis to be able to see all harmonics, and note where the harmonics die out.
The above data alone can’t tell you if the design is ‘good enough,’ but it is a great tool for getting a sense of the circuit and what sections need more attention. The MDO3000 and all of the associated probes are great tools for looking closely at a switching power supply. Running these three tests will get a designer a long way towards having the confidence required for a design release!
Here are some additional resources that might be helpful when diving in to switcher characterization:
Fairchild's presentation on power mosfets
Inductor saturation (where the current through the inductor goes through the roof without strengthening the magnetic field)
