Perfect measurement of jitter for oscilloscope settings

Half of the workload for perfect jitter measurement lies in how to set up the oscilloscope. Our goal is to capture and show the true state of the signal in a system environment. Because each lab has a real-time oscilloscope, it's important to know how to operate them. Jitter measurements are particularly sensitive to the environment, so find ways to optimize your test environment for various jitters.

First choose a device with the right bandwidth. If the bandwidth is too narrow, the edge rate tested will be low. A low edge rate translates more amplitude noise into a time domain error. However, if the strip is too large, it will only increase the thermal noise and shot noise in the test to improve the noise floor. In terms of NRZ code stream, an empirical rule is to select the bandwidth to be 1.8 times the code rate.

Next, try to increase the sampling rate to avoid aliasing that occurs due to undersampling. In theory, the sampling rate is at least twice the highest fundamental frequency of the signal; in fact, the analog signal shaping and data conversion during the capture process leaves a margin, so the oscilloscope really needs a sampling rate 2.5 to 3 times the highest fundamental frequency. . Therefore, the oscilloscope's bandwidth sampling rate ratio is about 1 to 3.

It is important to increase the longitudinal resolution of the instrument in order to reduce the ADC quantization error. Adjust the voltage/scale knob until the graph just enters the vertical range of the screen. Excessive will saturate the ADC change, and dissatisfaction will reduce the SNR.

It is also important to measure the time base setting when measuring TIE jitter, as this setting is equivalent to an adjustable high pass filter. The time base sets the minimum TIE frequency at capture (the oscilloscope bandwidth determines the highest jitter frequency).

Similarly, it is determined that the test data pattern contains the correct spectral component range and contains only real spectral components. When using the PRBS pattern, the pattern length should be long enough to capture low frequency components while not exceeding the instrument's memory range.

The delay between the trigger and the first sample point is always reduced. After the signal is triggered, the timing uncertainty is proportional to the length of the time base waiting for the sampled data. Reducing the delay reduces this uncertainty and therefore reduces the measured jitter value.

Avoid oscillating the waveform, select sin(x)/x to interpolate between data points, and use a large fast trigger. Finally, the trigger level setting is made consistent with the actual system receiver threshold level, otherwise it is set to half the waveform value.