20230507 TI Engineer It - How to test power supplies - Measuring Noise

Hi. I'm Bob Hanrahan application engineering at Texas Instruments.
This is a series on measuring performance of power supplies .we will be measuring noise on a power supply,
as many types of noise, but specifically will be working with a switch mode Power supply,
i'll be looking at two types of noise, i'll be looking at switching, ripple noise.
Now,that's the noise that occurs at the PWM frequency occurs from the charge and discharge of the LC filter in the circuit.
And we'll be looking at transient switch noise. That's the noise that occurs at the switch time at a much higher frequency.
And it's a frequency of the function of the inherent tanks circuit.
The LC tank that inherent to the circuit itself, before we go to the bench and actually do some testing.
I like to talk for a moment about proper, oscilloscope probing techniques. Now, let's go to the graphic.
I have over here and show you a few of the techniques that are used.
Now, on the left is a typical passive probe with a ground wire on it.
And we use this all the time for low frequency measurements, but it's not recommended for noise measuring at all.
It will give you extremely high readings. That are not valid In the middle,
we have the same probe but we using a short ground wire off of the barrel of the probe or a ground spring.
That I'll show you in just a moment. On the right side is the technique I like to use.
It's called the tip and barrel method. I like it because it's quick.
You can literally touch the tip to a positive.
A stub on the positive output of your supply and the barrel to the negative output of your power supply.
So with that, let's go over and perform some bench testing will be using the LM5117 evaluation board.
So board that I described on an earlier video, it is configured with 15 volts coming in and 12 volts going out.
We're driving 8 amps out of this power of supply and we'll start by measuring ripple noise.
Switching, ripple noise, and we'll use this ground spring.
Now, this slips on the outside around the barrel of the probe and gives you basically a second tip that will be using to touch the output of the power supply.
Now, really the ideal place to probe is right across the output capacitor shown down here.
If there's a higher frequency capacitor, I mean a lower value. You'd want to go right across that capacitor.
So, we touch the ground side.
On the negative, we touch the pin on the positive, and then we will bring your attention up to the oscilloscope.
Now, you'll notice a lot of noise on the oscilloscope right now.
So we with this test, we can put our bandwidth limiting on. I recommend you come down below, 100 megahertz.
We'll use 20 megahertz band with limiting, will stop the sweep, so we can come in and make some measurements.
Now, in this case, I'll use a cursor or two cursors, to measure the peak to peak ripple noise.
Now, the ripple noise is the triangle waveform and that's the actual charging and discharging of the output LC filter.
and we'll bring our markers to roughly the top and the bottom of that waveform.
And then we'll come down to the lower left of the screen and we'll make,
we'll look at the measurement that the scope made the delta between the cursors is 9.2 millivolts.
So that is your ripple amplitude. 9.2 millivolts feet to peak, you could log that capture the screen shot so you could take keep it for later analysis.
Now let's move to the transient noise measurement.
Now on this one I'm going to change this scope setting back to full bandwidth because we want to see all of the noise and we will for this one I'll use the tip and barrel method.
So, I'll pull off the ground spring.
Now, we come down to the evaluation board and we carefully touch.
The tip to the positive and the barrel of the scope probe to the negative.
And then we come back to our oscilloscope, and we adjust the amplitude down.
Obviously, it's going to be a higher amplitude signal.
And for this one, I think I'll use a feature of the scope where it will make the measurements for us.
So, in this case, I'll turn the cursors off. And you'll notice there's a maximum and a minimum amplitude.
And if we look down here, the maximum amplitude is 381, millivolts minimum 299,
and the frequency of this transient noise is 141 megahertz. Now keep in mind that at this free high frequency simple,
decoupling caps at your load, will reduce that amplitude significantly.
Even the inductance of a trace across a PC board will filter it out significantly.
So and also keep in mind that even the small ground that you ground connection we use,
here is still enough to pick up noise that is not really existing on the signal itself.
So in summary, do not use long ground wires when making your noise tests and good.
Probing technique is important.
Visit the following web addresses for more information.