About oscilloscope setup for perfect jitter measurement

　　Half the battle for a perfect jitter measurement is setting up the oscilloscope. The goal is to capture and display the signal as it truly appears in the system environment. Since every lab has a real-time oscilloscope, it is necessary to know how to operate it. Jitter measurements are particularly sensitive to the environment, so find ways to optimize the test environment for various jitters.

　　First, select a device with the appropriate bandwidth. If the bandwidth is too narrow, the edge rate of the test will be very low. The low edge rate will convert more amplitude noise into time domain errors. However, if the bandwidth is too large, it will only increase the thermal noise and shot noise in the test and thus increase the noise floor. For NRZ bit streams, a rule of thumb is to select a bandwidth that is 1.8 times the bit rate.

　　Next, try to increase the sampling rate to avoid aliasing effects caused by undersampling. In theory, the sampling rate should be at least twice the highest fundamental frequency of the signal; in practice, the analog signal shaping and data conversion during the capture process will leave margin, so the actual sampling rate required by the oscilloscope is 2.5 to 3 times the highest fundamental frequency. Therefore, the bandwidth-to-sampling ratio of the oscilloscope is about 1 to 3.

　　Increasing the vertical resolution of the instrument is important to reduce ADC quantization error. Adjust the voltage/scale knob until the graph fits into the vertical range of the screen. Oversaturating will saturate the ADC, undersaturating will reduce SNR.

　　The timebase setting is also important when measuring TIE jitter, as it acts as an adjustable high-pass filter. The timebase sets the minimum TIE frequency that can be captured (the oscilloscope bandwidth determines the highest jitter frequency).

　　Likewise, make sure the test data pattern contains the correct spectrum component range and only contains real spectrum components. When using PRBS pattern, the pattern length must be long enough to capture low-frequency components while not exceeding the instrument's storage range.

　　Always minimize the delay between the trigger and the first sample point. After the signal is triggered, the timing uncertainty is proportional to how long the time base waits for the sample data. Reducing the delay reduces this uncertainty, thereby reducing the measured jitter value.

　　Avoid oscilloscope averaging of the waveform, choose sin(x)/x to interpolate between data points, and use a fast trigger with a large amplitude. Finally, set the trigger level to match the actual system receiver threshold level if known, or to half the waveform value if not.