1. Time and Frequency measurement principles

Block diagram on Figure 1 demonstrates input signal transformation to series of time and frequency measurement data points.

First, input signal gets into input amplifier. The generic role of input amplifier circuits is matching the input to signal source, and optionally: attenuate or amplify it, remove DC offset and/or filter out high-frequency noise.

Please note: There are several kinds of input amplifier circuits (configurable input amplifiers, prescalers, fixed input amplifiers), please refer to 4.5 Input signal conditioning/Input Amplifiers chapter for more details about them.

After the input amplifiers the signal is digitized using one (inputs C, EA, ER) or 2 comparators (inputs A, B, D, E). It works the following way: when the signal crosses the set trigger level, comparator generates a positive (in case of transition from below to above the trigger level) or negative (in case of transition from above to below the trigger level) slope. Please refer to Figure 2 for an example.

Please note: For most measurement modes on inputs with 2 comparators, output of only one comparator is used for measuring the signal from this input. However, there are measurement modes using both comparators implicitly:

  • Rise/Fall Time and Slew Rate measurements use 2 comparators for getting Time Interval between lowest (10%) and highest (90%) levels of the signal (see details in 5.4.3 Rise Time, Fall Time, Rise-Fall Time);
  • Pulse width and Duty Cycle measurements use 2 comparators to produce pulses on positive and negative slope of a signal and measure Time Interval between these;
  • Frequency and Period Average use 2 comparators to implement implicit wide hysteresis targeted on improving noisy signals measurements (see details in 5.1.1 Frequency/Period Average measurements chapter).

Second comparator can also be explicitly selected for measurement. E.g., Time Interval A, A2 will measure time interval between signal level set by trigger level A and signal level set by trigger level A2 (can be useful for measuring characteristics of multi-level signals, e.g. TDR measurements – see 5.2.4 Measuring Time Interval between different trigger points of the same signal). It can also be explicitly used as a source of start or stop arming.

Digitized signals from physical inputs are just pulse trains which can be multiplexed to 4 internal measurement channels. Each measurement channel starts with Wide Hysteresis block. Wide Hysteresis block just passes through the signal for most measurement functions except Frequency, Period Average, their Smart versions, TIE, and Frequency Ratio for which implicit wide hysteresis is used for better noise tolerance (see 5.1.1 Frequency/Period Average measurements).

The resulting pulse train is then counted independently in each measurement channel. Measurement logic specific for each measurement mode makes snapshots of channel’s pulse counter, timestamps it and adds channel number, forming series of so-called raw results.

Raw results are further post-processed in calculation block to form series of final value-timestamps pairs.

See Figure 2 for a visualized example of signal processing happening inside CNT-104S.

Figure 1. Input signal to Time & Frequency measurement results transformation
Figure 2. Example of input signal to result transformation for Frequency measurement mode with Wide Hysteresis

2. Voltage measurement principles

Measuring voltage of the input signal is available only on inputs A, B, D, E. On each input there are 2 dedicated comparators used to search for signal’s lower and upper levels. 4 inputs can be measured in parallel.

Adaptive search algorithm is used which depends on Voltage Mode setting. Each voltage mode implies minimum input signal frequency which the voltage measurement can handle. E.g. in Normal Voltage mode (default) CNT-104S is able to measure voltage of signals with frequencies starting with 100 Hz and above. Allowing lower frequencies makes the voltage measurement slower so it is not recommended to set Voltage mode below Normal unless really needed (one might want to consider using Autoset instead).

Table below summarizes Voltage Modes available in CNT-104S:

Voltage ModeMinimum FrequencyAverage Time to measure 1 voltage sample
Very Slow1 Hz15 s
Slow10 Hz1.5 s
Normal100 Hz450 ms
Fast1 kHz65 ms
Very Fast10 kHz30 ms
Table 1. Voltage Modes
Figure 3. Input signal to Voltage measurement results transformation.

3. Sample Interval

On each measurement channel CNT-104S can produce gap-free samples back-to-back as fast as 1 sample per 50 ns which corresponds to setting Sample Interval to 0 s. However, for most cases one doesn’t need samples to be generated with such a high frequency, then using non-zero Sample Interval can be considered. In this case samples in each measurement channel will be generated not faster than once per set Sample Interval.

Figure 4. Example of how sample interval works with multi-channel Frequency measurement.

Sample Interval hints:

  • Sample Interval clock is not synchronized to signal, meaning actual Sample Interval between 2 consecutive samples can be less or more than Sample Interval set by the user.
  • When doing parallel measurement, sample interval is applied to each measured series independently. E.g., when measuring Frequency A, B, D, E and using 1 ms Sample Interval, one will get 1000 Frequency Samples per second from each input.
  • Sample Interval for averaging measurement functions (Frequency, Period Average and their Smart alternatives) acts as an averaging gate. The larger the gate (Sample Interval) – the greater the resolution.
  • The CNT-104S contains two cascaded memories for result saving. The first is the cache buffer that can holdup to 20000 raw samples with a maximum writing speed of 20 million Samples/s. The second is the main memory, that can hold up to 32 million results, with a maximum writing speed of 12.5 million samples per second, or 80 ns between samples. The data is written to the cache buffer in parallel from up to 4 inputs, meaning the capture speed is independent of the number of inputs used. The data transfer from the fast cache buffer to the slower main memory is done in serial, so the maximum capture speed varies with the number of channels  Setting max number of samples of samples to 20k for a single channel measurement, or 5k for a 4 channel measurement, will guarantee a sampling rate of 20 million values/s, at a sample interval of 50 ns. When setting a larger number of samples , the Sample Interval (given that signal period is less than or equal to the set Sample Interval) must be increased to minimum 80 ns for 1-channel and minimum 320 ns for 4-channel measurements, to avoid cache buffer overflow. If not, the measurement is aborted to avoid data loss.

Example. Measurement function is Frequency A,B,D,E, Sample Interval is 50 ns, Sample Count is set to 10000 and period of all input signals is 50 MHz. 4 measurement channels are used in parallel, each supposed to deliver samples at a rate of 20 million samples per second, which is greater than the speed cache buffer can be fetched with (12.5 million samples). Total number of samples to be generated is 4 x (10000 + 1) = 40004 (N + 1 raw samples are needed to calculate N frequencies), which is twice greater than measurement logic buffer capacity. So, the buffer will overflow and measurement will be aborted. Solution would be to either increase Sample Interval to 320 ns (4 channels x 80 ns, giving 12.5 million samples per second total) or to decrease Sample Count to 4999 (giving 4 x (4999 + 1) = 20000 samples total).

To avoid this situation please watch out for warning text when setting Sample Count and/or Sample Interval or warning icon when choosing measurement function and inputs.

Figure 5. Warning icon on Function/Inputs selection dialog (see clickable red button with exclamation sign to the left of Cancel).
Figure 6. Warning text when editing Sample Count
  • When using Sample Interval close to 50 ns, actual sample interval might vary between 50 ns and 100 ns. In case of Time Interval measurement – it can result in generating less samples than was ordered by Sample Count setting. To avoid this please consider setting Sample Interval to 0 if you need minimal sample interval between samples. Setting Time Interval to 0 disables Sample Interval clock, meaning taking samples as fast as possible (close to 50 ns if signal period is 50 ns or greater).
  • Sample Interval setting doesn’t affect Voltage measurement where interval between samples directly depends on Voltage Mode (please see Voltage measurement principles for details).
  • Sample Interval setting doesn’t affect manual Totalize measurement where interval between samples is always 100 ms.

4. Input signal conditioning/Input Amplifiers

4.1. Overview

The input amplifiers are used for adapting the widely varying signals in the ambient world to the measuring logic of the timer/Analyzer.

These amplifiers have many controls, and it is essential to understand how these controls work together and affect the signal.

4.2. Configurable amplifiers

Input amplifiers for inputs A, B, D & E are configurable (Inputs section of the Settings dialog) and allow to select a bunch of parameters described in the following sections.

Figure 7. Input amplifiers configuration menu

4.2.1. Impedance

Impedance setting allows to match the impedance of the input to signal source. One can choose between 1 MΩ or 50 Ω impedance.

CAUTION: Switching the impedance to 50 Ω when the input voltage is above 12 Vrms may cause permanent damage to the input circuitry.

4.2.2. Attenuation

This setting allows attenuating the signal by factor of 10 if its dynamic range exceeds ±5 V. 1x, 10x and AUTO attenuation options are available. If AUTO is selected then first sample of any voltage measurement (including the ones performed during Autoset and Auto-trigger) will be analyzed to determine the dynamic range of the signal. If the range exceeds ±5 V then 10x attenuation will be switched on automatically. 1x will be used otherwise.

4.2.3. Coupling

Use the AC coupling feature to eliminate unwanted DC signal components or keep DC offset by using DC coupling.

Hint: Always use AC coupling when the AC signal is superimposed on a DC voltage that is higher than the trigger level setting range. However, we recommend AC coupling in many other measurement situations as well. When you measure symmetrical signals, such as sine and square/triangle waves, AC coupling filters out all DC components. This means that a 0 V trigger level is always centered around the middle of the signal where triggering is most stable.

Figure 8. AC coupling a symmetrical signal.

Hint: Signals with changing duty cycle or with a very low or high duty cycle do require DC coupling. Figure 8 shows how pulses can be missed, while Figure 9 shows that triggering does not occur at all because the signal amplitude and the hysteresis band (please see for explanation of what is Hysteresis Band) are not centered.

Figure 9. Missing trigger events due to AC coupling of signal with varying duty cycle.
Figure 10. No triggering due to AC coupling of signal with low duty cycle.

Hint: always use DC coupling for signals below 10 Hz.

4.2.4. Filter

This setting allows you to apply analog low-pass filter for signals with high-frequency noise or interference. 10 kHz and 100 kHz low-pass filters are available. All filters have a signal rejection slope of approx. 20 dB / decade.

Hint: keep filters off unless you cannot obtain stable readings otherwise.

Hint: it is not recommended to use filters for pulse signals as filters affect pulse signal shape.

4.2.5. Preamp

Pre-amplifier allows to amplify the signal to improve sensitivity for signals with amplitudes below 100 mVpp.

Hint: avoid using pre-amplification if signal amplitude is above 100 mVpp. Prefer using Autoset instead – it will turn on pre-amplification only if necessary

4.2.6. Trigger Level

Set trigger level. Setting proper trigger level is essential for getting accurate and stable results. So it is advised to keep Trigger Mode Auto letting the instrument select adequate trigger levels. Please see measurement functions description for details.

In Auto and Relative Trigger Modes, the instrument performs voltage measurement (using current Voltage Mode setting) – so-called auto-trigger – before each Time or Frequency measurement which delays the measurement start. In cases when it is undesirable, please use Manual Trigger Mode.

When measuring non-continuous signals or single cycles, auto-trigger might fail to measure signal voltage range correctly which won’t allow setting trigger levels properly and might result in wrong measurement results. It is advised to use Manual Trigger Mode in this case.

In Manual Trigger Mode in most cases one can get the best results if Trigger Level is set to the center of signal voltage range. It will help avoid capturing signal edge artifacts and in most cases the middle of the signal voltage range will be the point with maximum slew rate, which minimizes timing trigger error.

Setting trigger level close to signal minimum or maximum level can result in intermittent readings and/or unreliable result. For example, measured Frequency value twice greater or twice lower than actual can be a typical consequence of poor trigger level choice when measuring pulse signals with significant artifacts on edges.

Please note: Actual triggering does not occur when the input signal crosses the trigger level at 50 percent of the amplitude, but when the input signal has crossed the entire hysteresis band (Figure 11). Which causes measurement timing errors.

Figure 11. Trigger hysteresis

The hysteresis band is about 20 mV with attenuation 1x, and 200 mV with attenuation 10x. The hysteresis compensation reduces hysteresis trigger error to <2 mV

To keep the hysteresis trigger error low, the attenuator setting should be 1x when possible. Use the 10x position only when input signals have excessively large amplitudes, or when you need to set trigger levels exceeding the -5 V to +5 V window.

4.3. Input Amplifiers with fixed configuration

The following inputs have fixed amplifier parameters:

  • EA (External Arming) is fixed to 1 kΩ impedance and approx. 1.5 V trigger level;
  • ER (External Reference) is fixed to 50 Ω impedance and accepts signals with 0.1 to 5 Vrms amplitude;
  • C (Prescaler input) is fixed to 50 Ω impedance.

5. Arming

Arming, in general, gives the opportunity to start and stop a measurement when an external qualifier event occurs.

Arming can initialize either a single sample acquisition (Arm on Sample) or a single measurement session defined by sample count (Arm on Block). Measurement can be started and stopped by rising or falling edge of signals on instrument’s inputs and delayed from 0 ns (delay off) to 2 s with 10 ns resolution step. In case of Totalize measurement mode, timer (set by a combination of Sample Interval and Stop Source set to Timer) can be used as a source of stop arming.

Table 1 shows possible arming modes and their specifics. Modes absent in this table are not supported.

Start Arming SourceStop Arming SourceArm onMeasurement FunctionComment
OffOffN/AAnyArming is not used, measurement is started and stopped normally
InputOffBlockAny, except TotalizeStart arming input initialize measurement session (block). Measurement session ends when all samples (number is set by Sample Count) have been collected.

Figure 12. Frequency measurement with measurement session repeated after start arming pulses

Arming start pulses that occur during measurement session are neglected (see ).

Figure 13. Frequency measurement with arming start pulses occurring during the measurement session
InputInputBlockAny, except TotalizeStart arming starts measurement session (block). Measurement session ends when either all samples (number is set by Sample Count) have been collected or by signal front on stop arming input (whatever comes first).

Figure 14. Frequency measurement with measurement session controlled by start and stop arming signals
OffTimerSampleTotalizeTotalize measurement is started by pressing Restart button, Sample Interval defines Totalize Gate length. The Analyzer counts number of events during the gate and produces single sample (1 sample per series).
InputTimerSampleTotalizeStart arming pulses start Totalize Gates (gate length is defined by Sample Interval). Each gate produces 1 sample (per series). Measurement session ends once required number of samples (set by Sample Count) was collected.

Figure 15. Totalize measurement in timed mode (stop arming signal set to timer)
InputOffSampleAny, except TotalizeStart arming is used as a pacing clock – single sample measurement is executed just after the arming event.

Figure 16. Pulse width measurement with arming signal used as a pacing clock In case of Frequency/Period Average measurement, samples are measured with dead-time, no back-to-back. Maximum total number of samples is reduced to up to 16 million.

Figure 17. Frequency measurement initialized by arming pulses
InputInputSampleTotalize, Frequency, Period Average, Smart Frequency, Smart Period AverageStart arming events start Gate (gate length is defined by Sample Interval). Stop arming events end Gate. Each gate produces 1 sample (per series). Measurement session ends once required number of samples (set by Sample Count) was collected.
In case start and stop arming settings are not the same, neasurements are performed with dead-time, no back-to-back. Maximum total number of samples is reduced to up to 16 million.

Figure 18. Frequency measurement in a gate created by arming signals

When arming start and stop conditions are the same, measurements are performed continuously (back-to-back) with no dead-time.

Figure 19. Frequency measurement when arming stop is set the same as arming start (back-to-back measurements) In case Smart versions of Frequency/Period Average are used, each sub-gate (Sample Interval divided by 1000) is being armed, i.e. one sample is calculated using data from 1000 arming gates.
InputInputSampleAny, except Totalize, Frequency, Period Average, Smart Frequency, Smart Period AverageStart arming is used as a pacing clock – a single sample measurement is executed just after active edge of arming pulse. Stop arming delays registration of timestamps other than the first one.   Depending on particular measurement mode, two, three or four timestamps are registered that give a single result.

Figure 20. Pulse width measurement with arming on sample and start and stop arming active, case 1

Figure 21. Pulse width measurement with arming on sample, start and stop arming active, case 2
Table 2. Arming modes

Start arming is useful for measurement of frequency in signals, such as the following:

  • Pulse modulated RF signals (bursts) Single-shot events or non-cyclic signals.
  • Pulsed signals where pulse width or pulse positions can vary. Signals with frequency variations versus time (“profiling”).
  • A selected part of a complex waveform signal.

Signal sources that generate complex wave forms like pulsed RF, or sweep signals, usually also produce a sync signal that coincides with the start of a sweep, or start of a radar burst. These sync signals can be used to arm the Analyzer. See Figure 21.

Figure 22. Sync signal used as start arming starts the measurement.

You normally use stop arming together with start arming. That means that the external gating signal controls both the start and the stop of the measurement. Such a gating signal can be used to measure the frequency of an RF burst signal. Here the position of the external gate must be inside a burst. See Figure 22.

Figure 23. Start and stop arming together is used for burst signal gating.

Note that burst measurements with access to an external sync signal are performed in the normal Frequency mode. In time interval measurements, you can use the stop arming signal as a sort of “external trigger Hold Off signal”, blocking stop trigger during the external period. See Figure 24.

Figure 24. Using start/stop arming as an external Hold-Off

Please note: start arming setup time is up to 5 ns. Which means that the measurement is actually started within 5 ns since start arming event.