measuring set up time for a DFF

Dear PaulOkada,

The method you have chosen to characterize set-up time is a bit non-conventional. The method used in the industry commonly is one that measures the propagation delay time and examines its magnitude as the delay between the input data and clock is varied. When the propagation delay increases by a threshold (usually 1% or a few percent) from its value when the clock and data are not in close proximity, the time is classified as  a "setup" or "hold" time. I've illustrated an example from an analysis I did some time ago.

In the method you are planning to study, you will not get a quantitative view as to how close the minimum time you measure is to the actually minimum setup time - if my comment makes any sense.

Nevertheless, to answer your question:

PaulOkada said:

would like to measure the time between each transition of data to the capturing clock edge.  It is not hard to trigger on the crossings for data but I have a hard time to set the stop time as the subsequent clock edge. 

If you locate a given data transition time, I might suggest you use the cross() function on a clipped version of the clock signal to measure the first clock edge after a given data transition. Specifically, the algorithm might be:

1. Locate a data transition, Tdata.

2. Clip the clock signal between Tdata and the end of the simulaiton and define the clipped signal as clock_clipped

3. Locatie the first clock transition of the signal clock_clipped using the cross() function where the crossing incident corresponds to 1 and the direction of the transition corresponds to whether the DFF samples on the clock riisng or falling edge;  define its value as clock_clipped_sampling_transition_1

4. Take the time difference (clock_clipped_sampling_transition_1 - Tdata); this corresponds to the time difference between the data edge Tdata and the next sampling clock edge

I hope I understood your question and this helps you out...

Shawn

Dear PaulOkada,

The method you have chosen to characterize set-up time is a bit non-conventional. The method used in the industry commonly is one that measures the propagation delay time and examines its magnitude as the delay between the input data and clock is varied. When the propagation delay increases by a threshold (usually 1% or a few percent) from its value when the clock and data are not in close proximity, the time is classified as  a "setup" or "hold" time. I've illustrated an example from an analysis I did some time ago.

In the method you are planning to study, you will not get a quantitative view as to how close the minimum time you measure is to the actually minimum setup time - if my comment makes any sense.

Nevertheless, to answer your question:

PaulOkada said:

would like to measure the time between each transition of data to the capturing clock edge.  It is not hard to trigger on the crossings for data but I have a hard time to set the stop time as the subsequent clock edge. 

If you locate a given data transition time, I might suggest you use the cross() function on a clipped version of the clock signal to measure the first clock edge after a given data transition. Specifically, the algorithm might be:

1. Locate a data transition, Tdata.

2. Clip the clock signal between Tdata and the end of the simulaiton and define the clipped signal as clock_clipped

3. Locatie the first clock transition of the signal clock_clipped using the cross() function where the crossing incident corresponds to 1 and the direction of the transition corresponds to whether the DFF samples on the clock riisng or falling edge;  define its value as clock_clipped_sampling_transition_1

4. Take the time difference (clock_clipped_sampling_transition_1 - Tdata); this corresponds to the time difference between the data edge Tdata and the next sampling clock edge

I hope I understood your question and this helps you out...

Shawn