Got the Time? An Introduction to Time Measurements

 When measuring time, we don’t actually observe the passage of time itself, but rather its indirect analog. Figure 1 shows a block diagram of a basic time-interval measurement instrument. The instrument comprises a precision frequency source (a clock oscillator, or timebase), a gate, a counter and display, and control logic.

The instrument measures the duration of an event by using the event to open and close a gate, thus turning on and off a stream of uniformly timed clock pulses. The longer the event, the more pulses pass through the gate. The counter accumulates clock pulses and transfers the counts to an appropriate display. By selecting a convenient clock frequency, you can view the event’s duration directly in standard time units—for example, microseconds, milliseconds, and so on.

In addition, a counter can measure the frequency of an incoming stream of pulses. The instrument does this by interchanging the gate and input sources and applying a gating event of known, fixed duration. Thus, the instrument counts pulses for a set period. Indeed, most of today’s general-purpose universal counters use the same basic architecture to offer several basic measurements:

• frequency—events per unit time;
• period—time between repetitive events
• interval—time between two events;
• transition time—elapsed time between two voltage levels on the same event; and
• total number of events.

Circuits Challenge Designers
The block labeled “Signal-Conditioning Circuit’’ in Figure 1 presents interesting design challenges. Operating as a comparator or a 1-bit analog-to-digital converter (ADC), this circuit must accept a slowly changing input and produce a clean logic-level transition to drive the digital counting logic. It must also discriminate against high-frequency noise riding on a legitimate signal.

Figure 2 shows how hysteresis—the difference between 0-to-1 and 1-to-0 voltage-transition levels—can assist in rejection of high-frequency noise superimposed on a signal. Adjustable threshold voltages in an instrument let users set optimum trigger levels for the specific signals they want to measure.

Amplifying a low-level input can produce an adequate trigger signal, but excessive gain also adds noise. And high-level inputs can saturate or overload an amplifier, thus introducing delay errors. Designers can overcome these problems by adding an automatic gain-control (AGC) loop that prevents first-stage saturation yet maintains adequate gain for proper operation.

Instrument-loading effects can also cause problems by interfering with the circuit under test. A typical input stage presents an equivalent impedance of a few tens of picofarads in shunt with a resistance—typically, 1 to 10 MV or 50 V. You want to be sure the instrument’s input impedance doesn’t alter the circuit under test. Added capacitance and resistance may alter phase shift or frequency. Any loading depends on your test setup, the probes you’re using, and the characteristics of your circuit.

Monitor the DUT’s performance for malfunctions when you connect any instrument. To combat unexpected loading effects, always use an oscilloscope to examine any signal you plan to measure. Then, look for changes in wave shape, transition time, or voltage levels when you connect the counter. Adjust the counter’s trigger levels for stable readings, and then disconnect the oscilloscope, noting any display changes that might indicate marginal trigger settings or ground-loop effects introduced by either instrument.

Optimal measurements require that an instrument’s resistive component must either match or exceed circuit impedance by at least a factor of 10. But input capacitance can load the DUT, widening a pulse width or increasing a waveform’s transition times. To measure a load-sensitive waveform, use high-impedance low-capacitance active probes. In digital circuits, you can use a spare section of a hex inverter or buffer to gain isolation at the expense of introducing extra (but fixed) delays. In analog circuits, a broadband unity-gain buffer connected between the test point and counter isolates the instrument, but a fixed resistor may do the job at lower cost.

Unsynchronized clock and input signals inevitably introduce a quantization error of one or more counts in counter-based time measurement. Depending on the time of arrival of a clock pulse, the leading edge of an input signal may cause the counter to receive no pulse, a partial pulse, or a full pulse. A similar condition applies to the trailing edge of the input pulse. Energy in partial pulses may or may not trigger the counter, causing an erroneous reading.

One technique designers have traditionally used to improve time-measurement resolution and minimize quantization errors borrows from dual-slope analog-measurement techniques. In Figure 3, the positive edge of an incoming pulse of width T triggers an integrator, which accumulates charge until the first internal clock pulse arrives.

Then, a discharge circuit drains the integrated charge at a fixed, slower rate (for example 1/1000th of the charging rate) while a secondary counter accumulates N clock pulses. The interval T1 = N/1000 adds to the basic measurement T.

The falling edge of interval T triggers a second integration (not shown), which ends when the next internal clock pulse arrives and drains the integrator at a 1/1000th rate, accumulating M clock pulses. The interval: T2 = M/1000 subtracts from interval T, making the corrected pulse interval:

Ttrue = T + T1 – T2

Applications Can Cause Errors
The sources of errors can arise outside of an instrument, too. You can introduce time-measurement errors in several ways, even with a freshly calibrated instrument that’s in perfect operating condition. For example, if you’re using an external trigger source to initiate a measurement and you inadvertently substitute a longer input cable, the longer cable will delay arrival of an input pulse, thus making it arrive late with respect to the trigger.

The cable delay is fixed and measurable, though, and a “smart’’ instrument can compensate for the cable delay. You can preset a delay control or specify an appropriate time offset in data-acquisition software.

Same Time Tomorrow?

Dedicated time-measurement instruments now offer more than the basic measurements noted earlier. Today, you can obtain statistical-measurement features—arguably the most exciting development in recent years. Where once a test operator had to make repeated individual measurements and guess the values of rapidly changing least-significant digits, modern counters store and display cumulative results of many measurements. These instruments can display pulse widths, transition times, and jitter in bar graph or numerical formats.

In addition, considerable functional overlap exists among time-measurement instruments, modulation-domain analyzers, and oscilloscopes. New oscilloscopes recently introduced by Tektronix and LeCroy offer time statistics. And modulation-domain analyzers combine many timing-analysis features of scopes and counters. Given the breadth of tools available, it’s embarrassing if you don’t know the time. T&MW

Bradley J. Thompson has been writing for Test & Measurement World since 1986. Currently, he serves as a Contributing Technical Editor and works as an independent electronics consultant and writer.

The Basic Measurement Series periodically reviews fundamental techniques and technologies for test professionals.