Application of Dingyang SDS2000 oscilloscope in testing power supply ripple/noise

Abstract
Today's electronic products have faster and faster signal speeds, and the power supply voltage of integrated circuit chips is getting smaller and smaller. In the 1990s, the power supply of chips was usually 5V and 3.3V, but now, the power supply of high-speed ICs is usually 2.5V, 1.8V or 1.5V, etc. The smaller the power supply voltage, the higher the voltage fluctuation requirement. The voltage test for this type of low-voltage DC power supply is referred to as the power supply ripple/noise test. The usual ripple requirement is +/-5%, and some even require +/-2%.

Measurement method
To measure ripple/noise, you must first know the correct measurement method. Many engineers use digital oscilloscopes to measure hundreds of mV of ripple/noise values, which are dozens of times different from the parameter specifications. This is definitely caused by incorrect measurement methods.

The following points will affect the accuracy of power supply ripple/noise measurement:
1. The noise floor and quantization error of the oscilloscope
2. Using a probe with a large attenuation factor to measure a small voltage
3. The distance between the GND and signal detection points of the probe is too large
4. The setting of the oscilloscope channel (50Ω and 1MΩ, DC coupling or AC coupling)
1 Requirements for instruments

Oscilloscope parameter requirements:
Support bandwidth limitation or digital filtering function: Generally, oscilloscopes support 20MHz bandwidth limitation, and also have more advanced full bandwidth limitation function, such as: SDS2000 series, impedance requirements are 50 ohms and 1M ohms optional.

The power supply and ground impedance at the chip end are usually in the milliohm level. After the high-frequency power supply noise is transmitted from the coaxial cable to the oscilloscope channel, when the oscilloscope input impedance is 50 ohms, the characteristic impedance of the coaxial cable of 50 ohms is completely matched with the channel, and there is no reflection; when the channel input impedance is 1M ohms, it is equivalent to high impedance. According to the transmission line theory, the power supply noise is reflected, thus causing the power supply noise tested at 1M ohm input impedance to be higher than that of 50 ohms.

Probe requirements:
To make the ground wire as short as possible, you can configure a short test pin for the probe, or make your own short ground wire: remove the probe ground wire sleeve, and wind a short ground wire with metal wire. Category 5 copper wire is recommended, which has moderate strength. You can also use solder wire, aluminum wire, etc.

Select the 1X non-attenuation position. Generally, when the passive probe is in the 1X position, its bandwidth is limited to 6MHz/10MHz. In this way, the interference of high-frequency ripple/noise can be effectively filtered out at the front end, reducing the impact of ripple/noise measurement. If the 10X position is selected, not only will there be no filtering effect, but the ripple/noise signal will be severely attenuated, which will increase the difficulty of observation.

The following uses Dingyang's SDS2000 as an example to illustrate the correct operating steps for routine measurements:
1. Turn on the bandwidth limit function of the SDS2000 and limit the oscilloscope bandwidth to 20MHz. This is to prevent the high-frequency ripple/noise of the digital circuit from affecting the measurement and to ensure measurement accuracy as much as possible; at the same time, select 50Ω impedance.

2. Set the coupling mode to AC coupling to avoid the oscilloscope not displaying the ripple/noise signal due to the DC signal being too large, which is convenient for measurement (use a smaller gear to carefully observe the ripple/noise, and do not care about the DC level); under DC coupling, when the range is 2mV or even smaller, the bias voltage of some oscilloscopes is not enough. The voltage offset range of Dingyang SDS2000 oscilloscope is as follows,
3. Ensure that the probe grounding is as short as possible. If the probe is far away from the ground, a lot of EMI noise will be radiated into the signal loop of the probe. As shown in Figure 1, the waveform observed by the oscilloscope includes other signal components, resulting in incorrect test results. Therefore, it is necessary to minimize the distance between the probe signal and the ground detection point to reduce the loop area. (The main reason for measuring ripple/noise of hundreds of mV is that the ground wire is too long). The ground wire and shell of the probe can be removed to expose the probe crust, and the ground wire can be wrapped around the probe crust. This can effectively ensure that the length of the ground wire is less than 1cm; (This is the key point. The ripple/noise is originally a small signal. The inductance effect generated by the long ground wire will cause greater ripple/noise interference to the measurement)
 

2 Actual test of a ripple/noise signal:
Measure the ripple/noise of a DC power supply, as shown below:  
 

First, use a high impedance to test the ripple/noise, as shown in Figure 3;
 

Secondly, use 50Ω to test the ripple/noise, as shown in Figure 4;
 

During the test, we found that if a probe with 1x attenuation is used for testing, when the oscilloscope channel input is 1M ohm, the power supply noise measured is usually greater than that of 50 ohm input impedance. This is because after the high-frequency power supply noise is transmitted from the coaxial cable to the oscilloscope channel, when the oscilloscope input impedance is 50 ohms, the impedance of the coaxial cable is 50 ohms and the channel is completely matched without reflection; when the input impedance is 1M ohms, that is, high impedance, the power supply noise will be reflected, which will cause the noise of the 1M ohm test to be greater than that of 50 ohms.

Therefore, when measuring small power supply noise, it is recommended to use a 50 ohm input impedance.