Though there can be overlap between the two and they can be implemented with the same on-chip resources, DCO (digitally controlled oscillator) mode and FOTF (frequency on the fly) typically have different uses and fit different applications. Here is a brief description of the two:
DCO (digitally controlled oscillator) mode
Often implemented as a closed loop external to the DSPLL, DCO mode usually involves small changes to the output frequency. For example, if the nominal output frequency is 122.88 MHz, the step size for each frequency increment or decrement may well be less than 1 ppm. There is a CBPro window with the name "DCO (digitally controlled oscillator)" that describes and implements this feature.
FOTF (frequency on the fly)
FOTF usually involves large changes to an output clock frequency due to some kind of an application mode change. For example, a line card is currently provisioned for a 125 MHz output, but needs to changed to a 156.25 MHz output. FOTF can make this transition as seamless and clean as possible. For multiple DSPLL devices like the Si5347, an important aspect of FOTF is that one DSPLL can undergo an FOTF change without in any way affecting the three other DSPLLs. This can usually, but not always, be achieved. CBPro provides a tool that will help this effort. To open this tool, at the "Design Dashboard" opening window of CBPro, click on "Export" and then the "Multi-Project Register/Setting" tab.
For details about implementing either DCO mode or FOTF, see the Reference Manual and CBPro for the specific device of interest.The following application notes deal with FOTF and DCO mode for various different members of the Si534x family: AN922, AN858, AN909, AN959, AN1178
A square wave with a frequency of FREF can be represented by its Fourier series, which consists of a sum of sine waves at frequencies that are integer multiples of FREF (also called harmonics). A perfect square wave with 50% duty cycle has only odd harmonics with their relative amplitudes decreasing linearly with frequency, as shown in the figure attached. The higher harmonics contribute to the sharp rising and falling edges of the square wave. A scope that has a limited bandwidth will attenuate those higher harmonics and the signal will be distorted. For more information: http://mathworld.wolfram.com/FourierSeriesSquareWave.html
There are several ways to express jitter in time units.
As an example: Given a 156.25 MHz output signal with a total of 10ps of total jitter:
One period = 1/156.25MHz = 1 UI = 6.4ns = 6400ps = 360 degrees = 2 * 3.14159 radians
1. Total Jitter: 10ps
2. Jitter in % UI = (10ps/6400ps) * 100% = 0.15625% UI
3. a) Jitter in degrees = (10ps/6400ps) * 360° = 0.5625°
3. b) Jitter in radians = 10ps/6400ps * 2 * 3.14159 radians = 0.00981748 radians
RMS jitter specifications are typically used in communications applications. Peak to peak jitter can be specified with relation to the RMS jitter value. RMS jitter is the one sigma standard deviation value. Typically to convert from RMS jitter to pk to pk jitter the 6 sigma value is used which specifies that 99.999% of the jitter would be accountable. The 6 sigma pk to pk Jitter = 6*RMS jitter value. Therefore if one multiplies the RMS value by 6 this gives the expected pk to pk jitter.
A signal is properly terminated when the output impedance of the driver and load resistance perfectly match to the transmission line characteristic impedance.
It is important to take care and use proper measurement techniques when measuring the signal to not introduce parasitic impedance onto the trace. Make sure to measure the etch as close to the end of the run as possible with a high bandwidth probe. Also be sure to keep the scope ground as short as possible. This is often done by using a very short ground clip. It is important that the probe doesn't add impedance onto the etch while measuring. This can easily happen if a long ground wire is used.
When there is an impedance mismatch this can cause overshoot, undershoot and ringing on the rising and falling edges of the signal or it can look like a step on the rising and falling edges. Assuming this is not due to the probe measuring technique, the signal the waveform can usually be improved by adjusting the series termination resistor from the input and/or the output termination. Attached is a generic image of a 50 ohm transmission line.