DaqExpert documentation

Updates are now verified before anything is installed: every update package is digitally signed and each file is checked, so a damaged or altered download is refused and your installation stays untouched. The application is no longer closed until the update has been downloaded and verified. The application selection screen is calmer: the day countdown no longer sits on every card. A licensed application shows a green Licensed badge; the countdown appears, in red, only in the last 5 days. One line at the bottom shows who the licence belongs to and until when.

New

• Signed updates: each update package carries a signed list of its files with their fingerprints. A package that is not signed by DaqExpert, or whose content differs from what was signed, is refused before anything is installed.
• The update checks that this computer meets its requirements (Windows, .NET Framework, WebView2 runtime, disk space) before replacing any file, and says plainly when a full reinstall is needed instead.

Improved

• Application selection: no more permanent day counter on each card. A licensed application shows a green Licensed badge; Expires in N days appears in red only in the last 5 days. An expired licence shows License expired — Renew, an application not included shows Request a trial. The licence holder and expiry date are shown once, at the bottom of the screen. Available in English, French, German and Spanish.

Fixed

• A failed or refused update download no longer leaves the application closed: it is only closed once the new version is downloaded and verified.
• Closing the application selection screen, or the login screen, no longer leaves DaqExpert running in the background without any window. Each application that is closed leaves memory; the selection screen stays available while an application is open and closes with the last one.
• Acoustic Mapping 3D: the DaqExpert Link tab no longer shows the technical description of the link between the page and DaqExpert, and the Send a test message button is gone. That button produced simulated levels that looked like a real measurement and could be captured on a point.

FluxProbe becomes a pure acquisition tool fed by the task running in Data Acquisition: as soon as Measure starts, its channels are available in the tool, the recording streams live (flux curve and per-pole deviation refreshed every second), the TDMS is saved automatically and the analysis runs by itself when the countdown ends or on Stop and analyse. Machine parameters (stator voltage, current, active and reactive power, field, temperature) can be read from DAQ channels or entered manually and are shown live and in the report. Calculated data recording: the RMS and mean of a periodic signal no longer drift for three seconds after a discontinuity in 10 ms averaging mode, the Harmonics group only holds measured values, the single-phase power quantities no longer carry meaningless harmonic columns, and the Statistics group gains a Valid flag. Acoustic Mapping 3D replaces the Acoustic mapping tool: acoustic campaigns on a 3D model of the unit, with levels computed by DaqExpert from the microphone signal.

New

• FluxProbe (air-gap flux probes): the tool reads the task running in Data Acquisition — no hardware to choose, no file to load. Open the tool, start Measure, and the channels are listed with their sample rate; start the recording and the live view shows the synchronous flux curve and the deviation of each pole against the 3 % threshold while it records. The recording is saved as TDMS in C:\DataAcquisition\FluxProbe and analysed automatically on the whole number of complete revolutions, both at the end of the countdown and on Stop and analyse. A new Machine parameters tab (02) reads stator voltage and current, active and reactive power, field current and voltage and stator temperature from DAQ channels (gain, offset, DC or AC) or manual values, shown in the top bar every half second and as test conditions in the results. Send to Report sends the printable report composed by the tool (conditions, synthesis, pole map, flux per pole, per-pole deviations, method), rendered page by page. A flux probe sampled under 10 kS/s is flagged before recording; a recording under-sampled for the measured speed and number of poles is flagged after analysis.
• Acoustic Mapping 3D replaces the Acoustic mapping tool. The campaign is built on a parametric 3D model of the unit (vertical, horizontal Francis, Pelton, bulb) or on a free floor plan with an imported image; measurement points can be generated (ISO 3744 / 3746 surface, survey, diagnostic) or placed by hand and dragged in the 3D view, and every measurement is stored per operating condition (background noise, no-load, synchronised, load steps, user-defined conditions). Levels can be entered by hand from a sound level meter or come from DaqExpert: choose the microphone among the signals of the running task, and DaqExpert computes every second LAeq, LCeq, LCpeak, LZeq and the 1/3-octave spectrum over a 30-second sliding window (A and C weightings per IEC 61672-1, bands per IEC 61260-1). The sound power LwA, the maps per condition, the campaign report (PDF, HTML, CSV) and Send to Report are in the tool; the campaign is saved automatically and restored when the tool is reopened, and exported files are stored in C:\DataAcquisition\Acoustic. The microphone channel should be declared in pascals in the acquisition task; any other unit is used as if one unit were one pascal, and the tool says so. The tool follows the application language (English, French, German, Spanish), including its campaign report and PDF.

Improved

• The Statistics group of the Calculated file now carries a Valid channel: 1 when every value of the row was computed on that window, 0 when a value was held from a previous window (power meter without a complete cycle, loss of signal) or when a column reserved in advance is not available yet (velocity methods with a settling delay). It can be used as a filter in DIAdem or Excel to hide points that are not measurements.

Fixed

• Accel to Vel: in the Time window, the calculated velocity no longer disappears or turns into straight lines when its method delays it (frequency-domain integration, zero-phase filtering): it is kept however long the delay, even beyond the 60 seconds the chart retains, and the delay notice now says how far to scroll back to see it next to the acceleration.
• Time chart scrollbar: its track now always covers the last 60 seconds (or 1.5 times the visible window if longer), like the Time History chart. Dragging back into that span always shows data, and a window parked in the past keeps its data until it leaves those 60 seconds - before, the track length depended on what had been done earlier and a parked window could empty itself from the left within seconds.
• In 10 ms (and 20 ms) averaging mode, the RMS and mean of a periodic signal were wrong for exactly three seconds after a discontinuity of the signal (a loop junction of a replayed file, a breaker operation, a fault on a current channel): up to 7 % on the RMS and 1 V on the mean, while the power quantities on the same window stayed right. The whole-cycle safeguard introduced in 0.7.0.1 rejected the signal's period as soon as one interval between two crossings was out of line, and the three seconds of history kept in front of each window carried that one interval for exactly that long. A single irregular interval among many no longer disqualifies a period; a handful of irregular intervals still does, which is what keeps a slowly moved sensor from being taken for a periodic signal.
• Single-phase power meter: the recorded Harmonics group still contained sixteen columns P_Amp1X, P_Phase1X, Q_Amp1X and so on, all at zero. The amplitude or phase of a power that is already an average over the window has no physical meaning, so these columns are no longer written. Measured channels keep their harmonics.
• The Harmonics group was written from the first window with zeros and a Valid flag at 0 whenever a keyphasor was configured but never produced a valid pulse, for instance when the channel declared as keyphasor is a plain sine. A recording in which no 1X was ever measured now has no Harmonics group at all. When the first measurement comes late, the group is created at that moment and the earlier windows are filled in as zeros flagged Valid 0, so that the Statistics, Harmonics and Virtual Channels groups keep the same number of rows and stay aligned in the CSV and Excel exports.

Two display defects where the recorded file was right and the screen was not. A trend of a slowly moving sensor (a displacement probe moved by hand, for instance) was drawn as flat steps instead of following the movement, on the Trend chart, the Trend Table and the Trend Dashboard alike. And with the order/omega velocity method, the velocity shown on screen read a few percent low and the Time window drew straight lines across the waveform while the recorded velocity was correct. Remote monitoring trends also receive every point the operator sees, compressed, instead of one point per transmission.

Improved

• The channel values panel of the main window now shows the same statistics as the Trend Table and the Trend Dashboard, computed once per refresh instead of being recomputed separately for that panel. Its values follow the trend calculation window and now include the 1X/2X items.

Fixed

• Trend, Trend Table and Trend Dashboard drew a slowly moving sensor as flat steps. The statistics window has a safeguard for periodic signals: it extends the calculation window so that it always spans whole cycles, which is what keeps the RMS of a voltage or a vibration flat. That safeguard also fired on signals that are not periodic at all — a sensor moved up and down twice within ten seconds was taken for a signal with a period of several seconds, the one-second window grew to ten seconds, and the displayed mean stayed frozen until the next movement entered the window, with jumps to intermediate levels in between. The recorded Calculated file was not affected because it works on shorter stretches of signal. The safeguard now only engages on signals that genuinely repeat: regular intervals, and a period under two seconds. Voltages, currents and vibrations keep exactly the same statistics as before.
• Order/omega velocity: the value on screen could read several percent low with extremes beyond the physical amplitude, and the Time window drew straight lines across the velocity waveform, while the recorded velocity was correct. The channels of a display snapshot were read one after the other while an acquisition block could be written in between, so the keyphasor and the acceleration no longer covered the same instants; the edge detector then saw one block twice and never its neighbour, and a whole block of revolutions came out wrong. Blocks are now written, and snapshots read, as a unit; the acceleration and its keyphasor are also paired by time stamp. In the Time window, each new slice of the velocity waveform now reaches the chart — previously every second slice between two redraws was dropped, and the chart bridged the gaps with straight lines.
• The advisories shown above the Time chart for the velocity methods (method delay, keyphasor silent, settings refused) always appeared in French. They now follow the application language.
• Remote monitoring trends were sparse and late on a slow internet connection. The remote view only received one trend point per transmission, and each transmission carried about 450 KB of raw samples; on a site with a weak upload link a transmission took several seconds, so the remote trend showed one point every 2 to 7 seconds where the application draws ten per second, and it lagged by up to 8 seconds. Transmissions are now compressed (about three times smaller) and each one carries every trend point the application plotted since the previous successful transmission, with its own time stamp, so the remote trend has exactly the application's points even when a transmission is delayed. A gap in the remote trend now means data was really lost, and it is drawn as a gap rather than bridged by a straight line.

The application selection screen now has a single button, Cancel, and it takes you back to wherever you came from — the application you were working in, even if you had minimized it, or the login screen. It replaces the two buttons that were there before. Exit was the risky one: used from this screen it could close the whole suite, including an acquisition left running, and on Data Acquisition that meant losing the configuration, which is only saved on a clean exit.

Improved

• One button instead of two on the application selection screen. Cancel returns you to the application that opened it, bringing it back to the front even when it was minimized, or to the login screen when you came from there. Back to login and Exit have been removed: a single action now covers both directions.
• Data Reporting: the report list and outline now sit on a grey side panel, and the working area — the page you are editing — is white. It was the other way round. The thin white strip above the title bar is gone: the window now uses the same frameless layout as Data Analysis.
• The company logo is fetched again during the session when it could not be obtained at start-up. Until now the logo shared by all the machines of a company was requested once, when the main window opened: if the update server was unreachable at that moment (no network on site), the machine stayed without its logo until the next launch. The application now retries at growing intervals (1, 5, 15, then every 30 minutes) until the logo is displayed or the server answers that there is none for this machine.
• Test tools: the twelve tool windows now share the same header controls, in the same place: Send to Report, then minimize, maximize and close. Sudden Short-Circuit gains the minimize and maximize buttons it lacked, Generator Efficiency gains maximize, FluxProbe gains Send to Report, and the button reads Send to Report everywhere instead of an arrow. In TIF & THD, the Acquisition, Excel and PNG actions leave the header: Acquisition sits at the top of the Acquisition tab and the two exports in the Results toolbar, next to Print and JSON.
• Test tools: for a given theme, the twelve tool windows now share the exact background colour of the main window (the hosted pages TIF & THD, Sudden Short-Circuit, Generator Efficiency and FluxProbe used a lighter panel colour), and their minimize, maximize and close buttons look the same everywhere: same size and shape, same light hover, no red close button.

Fixed

• Leaving the application selection screen could close everything. The Exit button shut the suite down whenever no other window was on screen — and opening this screen from an application hides that application, so the condition was met even with work in progress. There is no longer any way to close a running application from this screen.
• The orbit chart's filter produced a spurious loop at the start of the figure. The filter kept its internal state from one refresh to the next while being fed the whole last second of signal again at every refresh, thirty times a second; the state inherited from the end of the previous second was applied to the beginning of the current one. With a narrow band-pass used to isolate the 1X component, the first revolutions of the orbit were drawn with a wrong radius and phase, flickering from one refresh to the next. The filter now processes each sample exactly once, so the orbit starts clean. The two axes are also paired by their time stamps instead of by position, which removes a small, intermittent skew between X and Y.
• Remote monitoring now shows the operator's orbit. The remote orbit was drawn from the raw signal and from whichever two channels came first, while the operator was looking at a filtered orbit of the pair they had chosen. The application now tells the remote view which filter it applies, to which axis, which pairs it displays and whether it recentres the figure, and the remote view applies exactly the same processing; the remote orbit chart, which had stopped drawing altogether in the previous live-monitor release, draws again. Remote curves also recover cleanly after a missed transmission instead of stitching two disjoint stretches together.
• Filtered trend values were averaged over ten seconds instead of the calculation window. On a Trend chart or Trend Table, a filtered channel was computed from the last ten seconds of signal, while the unfiltered channel uses the calculation window (one second by default). The filtered series therefore reacted ten times more slowly: one second after a step change the unfiltered value was correct while the filtered value had moved only about one seventh of the way and still read roughly 40% low, and it took the full ten seconds to catch up. Both now use the same window, so a filtered series responds exactly like its unfiltered counterpart. Existing filtered trends will look sharper than before, which is what they should have shown all along.
• FluxProbe, 3D machine tab: the controls column on the right now scrolls, so the analysis anomalies and the legend are reachable again. They were cut off at the bottom while the 3D view itself stays fixed.