Application Notes-403
 

How to test Noise from the Output of Power Supply.

1. Forewords:
Nowadays, switching power supplies have replaced traditional linear power supply; become the power source of most electronics equipment. It is being used in the desktop computer, monitor, printer, notebook computer, fax machine, photocopy machine and etc. The reason is because the advantage of smaller size, weight less and high efficiency, however, the bad thing of it is the noisy output. This article will discuss about the noise in switching power supply with DC output.

1.1. Definition:
Definition of output noise: the noise looks like a sine wave which overlaps on the DC output, this sine wave contain PARD (Periodic and Random Deviation) noise and it looks like the wave in Figure 1.
Figure 1 switching power supply's output noise

Since switching power supply use high switching frequency (>20kHz), accompany by PWM (Pulse Width Modulation) and output filtering circuit, it can transform the household electricity (AC) into DC voltage which IC circuit need, for example +5V, +12V, -5V, -12V, +3.3V, and so on. The only problem with the switching power supply is some sine wave in the DC output. If this sine wave (will call noise from now on) is large enough, it will cause malfunction in the application circuitry. We use 5V for example, if the noise is 1.0Vp-p, it will exceed the 4.75V-5.25V tolerance (a normal working voltage for logic IC), which can cause malfunction or system shutdown, therefore output noise has a lot of influence. Normally the output noise of switching power supply is control under 1% of the output voltage, for example for +5V, +12V, its' output noise specification should be 50mVp-p and 120mVp-p.

1.2. Controlling Output Noise:
When developing a switching power supply, the output noise has to be set within a certain specification. During production, things like the parts (such as transformer, diode, filtering capacitor and so on) with different material, incorrect assemble, missing parts and so on that cause noise level over the specification. To avoid these problems, checking output noise of each switching power supply is a must during quality control.
 2. Output Noise Testing:
The testing equipment can be either oscilloscope or ripple/noise meter, the differences are describe as follow:

2.1. Oscilloscope:
This is the most popular equipment use for testing, but pay attention to following condition otherwise the test result will not be accurate.
2.1.1. Avoid Ground Return: Differential input oscilloscope should be used, because the BNC negative input in general oscilloscope is connected to the case of the oscilloscope, and the case of the oscilloscope is connected to outside line ground. If use a probe to measure the DC output, it can cause ground return current which will affect the measurement. Because of that, ground return should be avoided. Using a differential oscilloscope or oscilloscope with a external differential amplifier is the correct method for testing connection, see figure 2, INTEL suggested test connection method. Figure 3, HP suggested test connection method. Both INTEL and HP suggestion use differential oscilloscope; HP differential amplifier is shown in figure 4, as Tek differential probe is shown in figure 5.

  Note: 7A13 plug-in is differential amplifier
  Figure 2 PARD differential mode test connection diagram suggested by INTEL

 

 

 

 Figure 3 PARD differential mode test connection diagram suggested by HP
 
  Figure 4 HP 1141A Differential Probe with HP 1142A Power Supply
 
  Figure 5 Tek P6046 Active Differential System
   
 

2.1.2. Test Condition:
When adding capacitor to the terminal during test, the capacitor should be mentioned which include its material, capacitance and etc., like figure 2 which INTEL add 10uF and 0.1uF capacitor in order to simulate the capacitance on the main board. There is no rule set for this test procedure; it is all depend on the condition. The capacitor should be mentioned if it is used during test, like figure A and figure B recommended connecting method, otherwise there will be big difference in the test results.

2.1.3. Frequency Bandwidth:
The frequency bandwidth of the oscilloscope will affect the test result, usually the standard is 20MHz, or 30MHz (INTEL test requirement), and the sampling frequency for digital oscilloscope should be more than twice of the PARD frequency.

2.1.4. Input Resistance:
Usually we use 50 Ohm. For low input resistance like 50 (the oscilloscope can be set to 1M) at the terminal for the purpose of eliminating the signal interference. There is no specific requirement for this part, but the input resistance should be mentioned, otherwise there will be big difference in the test results too.

2.2 Ripple and Noise Meter:
 Due to the reason that most of the oscilloscope have only two channel, it would be difficult to monitor the power supply with many output, addition disadvantage like it need human eye to observe and compare, and high cost. Therefore using Prodigit 4030 (up to 4 channel) with ripple noise meter, it can do the test in one time with upper and lower limit shown, plus the cost is reasonable (much cheaper than oscilloscope), so this is a very good test equipment, 4030 function block diagram is shown in figure 6. Its main specification is described as below:

4030 Ripple/Noise (Peak to Peak) Meter Block Diagram

 
 
  Figure 6 4030 Function Block Diagram
   
 

2.2.1. Input Structure:
Like the oscilloscope description, the connection should avoid ground return, therefore It is necessary to use differential mode, and the input resistance normally is 50W (which can effectively eliminate the input noise).

2.2.2. Input Range:
Usually, the specify noise for DC switching power supply output is 1% of its DC output, for example 5 volts DC output, its output noise should be less than 50mVp-p, for 12 volts DC output, the noise should be below 120mVp-p. Prodigit 4030 has 3 input range options to choose from when order, they are 3.0Vp-p, 1.5Vp-p, 0.75Vp-p respectively. Normally for 5V, 12V system voltage, choosing option with 0.75Vp-p input range should be enough, for 24V, 48V system voltage, 1.5Vp-p or 3.0Vp-p option should be chosen.

2.2.3. Frequency Bandwidth:
 Prodigit 4030 Ripple/Noise meter include 3 frequency bandwidth for user to choose from, they are 20Hz~200KHz, 20Hz~2MHz and 20Hz~50MHz. 20Hz~50MHz is mainly for checking output noise, 20Hz~200KHz or 20Hz~2MHz is for checking output ripple or dynamic load over-shoot or under-shoot. Frequency bandwidth range of 20Hz~50MHz means 4030 meter can measure the response of peak to peak frequency range in input signal (noise). It usually means -3dB (0.707 X) frequency range. For example if the input signal is 100mVp-p, at 50 MHz, 4030 will measure 70.7mV or more. 4030 Ripple/Noise meter classic frequency respond is shown in figure 7.
 
 
  Figure 7 4030 ripple noise meter typical frequency respond
   
  2.2.4. Frequency Spectrum of the Power Supply output noise:
Power supply output noise includes many different frequency (like 50Hz, 60Hz ripple on off frequency) and other noise. If output noise has narrow and sharp FET with ON/OFF spike, it will contain many high frequency, we can use Spectrum Analyzer to observe the frequency contain, figure 8, 9 are the power supply output noise wave form and spectrum. When using 4030 Ripple/Noise meter to measure noise, picking different frequency bandwidth will cause its value to change. Normally, high frequency bandwidth (20Hz~50MHz) will result in higher value than using low frequency bandwidth.
 
  Figure 8 typical power supply output noise
   
 
 
  Figure 9 Typical power supply output noise's spectrum
   
 

2.2.5. How to Define Pass/Fail with 4030
4030 has the ability to measure and compare between the upper and lower limit. In every single PARD measure channel, when measure value is within upper and lower limit, it will give a GO or Pass, if measure value is outside the upper and lower limit, it will give a NG or Fail; if all four PARD measure channel pass, green LED that shows PASS on the panel will light up, if there is one fail the red LED that shows FAIL will light up on the panel, to show that it doesn't pass the test condition. Every 4030 PARD measuring channel has its own upper and lower limit value, operator can set its value individually, and its PASS/FAIL comparison function block diagram is shown in figure 10.

Note: 4030 panel will light up green PASS LED when pass, red FAIL LED will light up when fail.

If using logic symbol to represent, the result will be like as follow:

PASS = (PARD1L!OPARD1!OPARD1H) AND (PARD2L!OPARD2!OPARD2H) AND (PARD3L!OPARD3!OPARD3H) AND (PARD4L!OPARD4!OPARD4H)
FAIL = (PARD1L!OPARD1) OR (PARD1!OPARD1H) OR (PARD2L!OPARD2) OR (PARD2!OPARD2H) OR (PARD3L!OPARD3) OR (PARD3!OPARD3H) OR (PARD4L!OPARD4) OR (PARD4!OPARD4H)
 
Figure 10 4030 PASS/FAIL function comparison block diagram
   

2.2.6. Checking Connecting Cable:

4030 includes eight SMB high frequency cable and sixteen SMB connector, providing connection between 4030 Ripple/Noise meter and test power supply, its connecting method can be found in 4030 operating manual. Prodigit also prepare the optional fixture Model 9951 (P/N: 65233005), cost US$10.00 /pc, which is shown in figure 11 for PC power supply, so that the user immediately connect the power supply to the 4030
Figure 11: 9951 PC Fixture with SMB connector
 

3. Conclusion:

3.1. Suitability:
4030 Ripple/Noise meter include four channel of noise measure circuit; it can measure four output from power supply simultaneously. If the output of the power supply is more than four set, two 4030 can provide up to 8 set of noise measurement simultaneously, therefore it will solve insufficient input channel problem.

3.2. Input Structure:
4030/3600A equipment with differential input structure, which can avoid ground return problem, addition advantage like the low cost and ease of use. This can improve the problem of expensive differential probe and hard to use.

3.3. Definition:
4030/3600A equipment with PASS/FAIL define mechanism which improve the difficulty in defining pass and fail using oscilloscope. This is very suitable for quality control and checking on production line.

3.4. System Expansion:
4030 equipment with RS-232C & GPIB interface, and 3600A equipment with RS-232C interface, which can use program to control. It is suitable for manually, computer, or automatic control testing.