An updated API client is available for EMScope software version 4.1 and later. It extends the v4 interface with several protocol and usability improvements:

• Binary amplitude frames: measurement data is now transmitted as binary WebSocket frames ([flags : uint8][padding : 3 bytes][float32 values …]) instead of JSON, reducing overhead on high-frequency sweeps. The client decodes them transparently.
• Cached frequency axis: the server sends the frequency array only when the RBW or display range changes — not on every sweep. self.freq is updated automatically and shared across consecutive binary frames.
• has_sweep_data(msg): a new helper method that returns True for both binary amplitude frames (EMI mode) and JSON values messages (impedance mode), replacing the manual “values” in msg check used in v4.
• repetition flag: self.repetition is set to True when a quasi-peak repeat sweep was triggered (200 Hz / 9 kHz bands only).
• Convenience setters: dedicated methods (set_detector_type(), set_measure_channel(), set_display_range(), set_amp_units(), etc.) alongside the generic set_value().
• Parallel multi-detector measurement: an additional example demonstrates simultaneous PK + QP + AV acquisition using three concurrent WebSocket connections and asyncio.gather.
• Reports with a verdict: request_report() sends the frequency range and report fields the server needs (before, the report never arrived), and the reply sets self.verdict (“PASS”, “FAIL” or “NOT_EVALUATED”, which is never a pass), self.detectors, self.not_evaluated, self.overload_traces and self.source (the traces evaluated by a tabs report). Frequencies in self.report are in Hz. set_value(“trace_id”, n) names the trace in the first column of the report rows; without it the column is None.
• Instrument in use: if another session holds the instrument (another session_uuid), connect() raises DeviceInUseError instead of returning with empty device data.
• More sweep flags: self.dwell_reduced (the spectrogram fast mode shortened the dwell: quasi-peak and average uncalibrated) and self.overload_held (the trace contains data measured with overload).
• User-defined modal channels: set_modal_coefficients() and set_neutral() define the modal channels the same way as the advanced modal options of the web interface (see Modal channels below).
• Number of points: the spectrum has three interpolated points between bins (one in software 4.3 and earlier), about 23 475 points for the whole 9 kHz band. Take the length from self.freq; do not assume a fixed value.
• Reference level: the server starts at 115 dBuV (100 in software 4.3 and earlier). With the attenuator in “auto” it sets the attenuation, and with it the noise floor: call set_reference_level() at the start of every session.

The api_v5.py and api_example_v5.py files are provided. The api_example_v5_multi_detector.py file demonstrates simultaneous PK + QP + AV measurement using parallel connections. The api_example_v5_modal.py file measures user-defined modal channels on a three-phase instrument.

Required libraries: websockets, matplotlib. Python 3.8 or later is required.

For a detailed description of the underlying WebSocket protocol, refer to the WebSocket API documentation.

In modal mode (set_mode(“modal”)) every modal channel is a weighted sum of the instrument inputs, computed before the FFT: c1·In1 + c2·In2 + …. The coefficients are defined with set_modal_coefficients() and require the advanced_modal license (request_licenses() lists it). Without the license the server ignores them and does not reply.

• The argument is a dictionary {configuration: {channel: {“c1”: value, …}}}. Each value is a real number or a [real, imaginary] pair, with both parts within -1 and +1 (+1 is applied as 0.99997). Only the configurations given are changed, but every channel given needs all the coefficients of its configuration; anything else raises ValueError before sending.
• The channel names are fixed; their content is free. For example, the common mode of L1 against the neutral, (L1 + N) / 2, is cm with c1 = 0.5, c2 = 0, c3 = 0, c4 = 0.5.
• The instrument measures two modal channels per sweep: cm with dm1, and dm2 with dm3. With more than two traces the two pairs are measured in consecutive sweeps, so place the channels that must be measured at the same instant in the same pair.
• Every measuring connection must call set_visible(True): on RX4 models the instrument measures only the inputs of the visible traces.
• With coefficients 1 and -1 a differential channel can reach twice the full scale of one input: it saturates and the trace shows OVERLOAD, as when an input overranges (before software 5.0: without an overload flag). The default channels use 0.5 and -0.5: add 6.02 dB to obtain the full difference.
• The coefficients apply to the RBWs up to 30 MHz. With RBW “120” (30 - 110 MHz) the modal channels are fixed.
• Send them at the start of every session: a server restart returns them to the defaults. Software 4.3 and earlier also lose them on a band change that loads another firmware (for example to “120” and back); there, send them again after set_rbw().

The api_example_v5_modal.py file measures the common and the differential mode of each phase against the neutral (six channels) with two coefficient configurations, one after the other.

The api.py and api_example.py files are provided for use with EMScope software versions 3.13 and earlier.

Required libraries: websockets, json. The matplotlib library is used in the example script.

A fully redesigned API client is available for EMScope software version 4.0. It introduces a cleaner interface, automatic RBW and display range management, robust session handling, and transparent keepalive ping/pong support.

The api_v4.py and api_example_v4.py files are provided.

Required libraries: websockets, matplotlib. Python 3.8 or later is required.