HP 54710A User's Reference Manual page 222

Summary of dc Errors
You can minimize the effects of dc errors on time-interval measurements by
observing these guidelines when feasible:
•
Measure between identical edges—same slew rate, amplitude, and offset.
•
Use the same dc level to reference each endpoint of the interval.
•
Reference the measurement to the fastest-slewing portion of each edge.
•
Operate the oscilloscope in a mode that gives the best vertical resolution.
Signal edges having faster slew rates minimize the effects of dc errors on the
time interval measurement.
Dynamic Response Errors
Whenever the input voltage of an oscilloscope changes with respect to time,
there is some transient error in the oscilloscope's reproduction of the input
signal.
The probe also has some non ideal transient response that contributes to the
overall system error. Additionally, the probe and the circuit under test form
a circuit that behaves differently from the circuit without the probe; the
signal at the probe tip is not the same as the signal that is present when the
probe is removed. Probe loading often has a more significant effect on the
measurement than the transient response of the oscilloscope, and must be
included in the analysis of time-interval measurement errors. See "Probe
Capacitive Loading" for more information.
At the very high bandwidth of the HP 54720 oscilloscope, the transient
response of 50 Ω coaxial cables can be significant. You must use good quality
cables and keep them as short as possible for best accuracy.
Measurements
Time-interval measurements
13–37