Signal Waveforms and Sampling

The amplitude of the microwave power delivered by a magnetron is determined by its high-voltage DC power supply, as it essentially follows the waveform of the magnetron anode current. This waveform can range from a low-ripple type (CW) to power-line rectified (high-ripple) to pulse-modulated. Modulated signals of various kinds can also be produced by other generator types.

In addition, all measured quantities may vary periodically for different reasons (e.g., a mode stirrer installed in the microwave cavity, objects moving on a belt in the applicator, or a combination of both).

The Homer collects voltage samples from its four detectors, and can be commanded to compute results either

• based on each individual sample, or

• by calculating special averages over a sampling burst, which is a collection of samples taken consecutively during a certain length of time with a defined sampling rate.

It is essential, especially with autotuning, to configure the timing of the signal sampling such as to arrive at meaningful, steady results; otherwise the supplied data may widely fluctuate, resulting in apparent noise and improper, unstable and seemingly random tuning.

Moreover, with signal power waveforms which drop to zero in parts of their period, an appropriate triggering must also be employed to activate the frequency counter only when the signal is nonzero; otherwise the measured frequency will be meaningless.

The user should be aware of the signal waveform type involved and understand how the sampling works to set up the sampling, triggering and frequency counting appropriately.

A summary of parameters controlling the sampling is shown in Fig. 25 and described below. The measured signal (blue and red trace) is sampled at points indicated by the circles.

Fig. 25.  Signal waveform and its sampling.

Triggering

When a requirement to make a measurement occurs in the system, the sampling, depending on the status of the Free Run flag, may start either immediately or after a triggering event has occurred.

• If Free Run = TRUE, sampling starts without delay. Repeated sampling bursts generally start at random instants of the signal waveform. This setting of Free Run is intended for CW or low-ripple signal waveforms.

• If Free Run = FALSE, sampling starts at the instant when the microwave power level, depending on the chosen trigger slope, exceeds or drops below a set triggering level L_trg, which is defined as a percentage of the (V_4 max – V_4 min) difference, where V₄ is the reference detector output voltage. (The Homer determines V_4 min and V_4 max before each sampling burst.)

A signal sampling is normally followed by a frequency measurement; the same free run/triggering logic is used.

Waiting for a triggering event continues for a limited timeout interval T_max. If no triggering event occurs during this interval, measurement will start automatically on T_max expiration. This may result in slow measurement cadence, fluctuating results and incorrect frequency measurement.

Sampling Parameters

The sampling procedure collects N_s samples (1 to 4095), spaced in time by the sampling step Dt_s = 1/f_s, where f_s (10 Hz to 200 kHz) is the sampling rate. The whole sampling process takes a time denoted sampling time or sampling duration T_s, where

or

Any two of the three parameters involved in the equations can be chosen (subject to their bounds); the third one must be computed using the above formulas.

In case of CW or low-ripple signals, the sampling parameters can, in principle, be arbitrary (of course, subject to the above constraints).

A case of periodic waveforms (resulting, e.g., from using a high-ripple magnetron power supply) is illustrated in Fig. 25. The sampling time T_s in such cases should ideally cover exactly one signal period T_p (thicker red portion of the waveform) or its integral multiples, otherwise the mean power will not be computed accurately. To aid in proper settings, the sampling parameters in Rectified mode are defined indirectly in terms of the ripple repetition frequency, the number of periods to be covered, and the total sample count.

A particularly undesirable situation may occur when the user does not recognize that the signal is of high-ripple type and employs CW sampling with a short sampling time relative to period T_p. An example of such a situation is shown as case A in Fig. 25 (densely packed green circles). The sampling here covers only a small fraction of the ripple period, yielding an average value V_{4 avg} which strongly depends on the instant the sampling has started. Since in CW mode the sampling starts at random, the result can fluctuate practically between V_4 min and V_4 max. To make the situation worse, if the frequency counting interval (T_c in Fig. 25) includes periods with near-zero power level, the frequency measurement may be incorrect.

Accept Level

Some measured quantities, e.g., the reflection coefficient, are meaningless during zero-power signal portions. These meaningless values would introduce error into the average reflection coefficient and reflected power computations, and should therefore be discarded. Sample acceptance/rejection is controlled by a parameter named Accept Level (L_acc). It is defined as a percentage of the peak reference detector voltage V_4 max, below which the samples are considered invalid and are ignored in reflection coefficient and reflected power computations. The invalid region is the grayed area in Fig. 25; the invalid samples are indicated by the yellow-filled circles.

Processing of the assembly of the collected samples continues depending on the user-selected mode of sampling (also referred to as Waveform). Three types of signal waveforms (and equally named modes of sampling) are distinguished: CW, Rectified, and Pulsed. All will be described in detail below. (Simulation program HoSim.exe can “generate” all three types of the waveforms.)

In addition, a special situation may occur, which can be designated as a Mode Stirrer case, when the incident power may theoretically be constant, providing no clue for triggering, but the reflection coefficient and the reflected power oscillate periodically. This situation will be treated separately.

Modes of Signal Sampling

There are three signal sampling modes available, named CW, Rectified, and Pulsed. Each of them is suitable for a different microwave signal modulation envelope (waveform).

CW Sampling Mode

The CW sampling mode is suitable for signals essentially constant over time or with a relatively low ripple in it. Such waveforms result for instance when feeding magnetrons by constant or low-ripple anode voltage. This type of signal is referred to as CW (continuous wave) or Low-Ripple. See more details in CW Signal Sampling.

In the CW mode, the detector voltages are sampled over a defined sampling time T_s. After being collected, the mean values of the voltages are calculated. The final results of interest (reflection coefficient, incident power) are computed from these averaged voltages. This is in contrast with the Rectified and Pulsed modes, where the results are computed for each individual sample, and only then averaged.

Rectified Sampling Mode

Many magnetron generators use high voltage which is simply a result of unfiltered rectification of up-transformed one-phase or three-phase power line voltage. Microwave power envelope then pulses with repetition rate f_r equal to power line frequency f_p or its integral multiple (typically 1, 2, 3, 6 or 12). Such type of signal is referred to as Rectified. See more details in Rectified Signal Sampling. Rectified-type sampling is also useful to handle the Mode Stirrer situation.

Pulsed Sampling Mode

In some applications, microwave generators are modulated by fast rectangular pulses with durations in some cases as short as 20 microseconds. This type of signal is referred to as Pulsed. The Homer is capable of processing pulsed signals with pulse durations down to 50 microseconds, including autotuning. See details in Pulsed Signal Sampling.

What Is the Difference

In the CW mode, the detector voltages are sampled over a defined sampling time T_s. After being collected, the mean values of these voltages are calculated. The final results of interest (reflection coefficient, incident power) are computed from these averaged voltages.

In the Rectified and Pulsed sampling modes, the detector voltages are also sampled over a specified sampling time T_s, possibly using triggering. After being collected, the samples are, unlike in the CW mode, processed individually to yield particular results (reflection coefficient, incident power, reflected power) for each sample. These particular results, not voltages, are then averaged in a special manner to obtain final results. The individual results can also be viewed.

Communication Problems in Rectified/Pulsed Modes

Communication problems may sometimes be experienced in Rectified and Pulsed modes of sampling. This is because the sampling, in particular the sending of individual samples, may take quite a long time (e.g., more than 10 seconds to transfer 4095 samples).

During this period, the system does not respond to commands from the PC; the commands are either ordered in a queue or may be lost.

As a result, the Homer may exhibit seemingly erratic behavior when it takes the commands from the queue and executes them.

A progress bar on the Signal toolbar shows progress in receiving of the individual samples.

Tips to Avoid Difficulties

• Do not click indiscriminately when encountering no response from the Homer: wait a few seconds first, then try to stop the Homer internal measurement by clicking the toolbar button or the equivalent Homer|Stop Homer menu item.

• Do not Send Samples when you want a faster measurement (uncheck Send Samples checkbox in the Homer Sampling Setup window or deactivate the toolbar button).