What Is Zero Pad Measurement Gaps?

Zero pad measurement gaps is an optional setting on Measurements and Pan-42 Measurements export jobs. When enabled, PowerRadar inserts zero-value records for intervals where a sensor or meter did not report, rather than leaving those intervals absent from the file.

The result is a continuous timeline at the chosen resolution - every interval in the export window has a row, whether or not the device produced real data for it.

Why use it?

• Keep row counts consistent across export runs so downstream BI tools and data pipelines don't misalign charts or produce false drops.
• Avoid gaps that break time-series joins in a data warehouse.
• Make silent intervals explicit - a zero row signals clearly that a device reported nothing for that interval, rather than leaving the question open.

Where to Find It

Zero pad measurement gaps is a checkbox in the New export job configuration form, located below the Data resolution setting and above the Data transfer method section.

1. In the sidebar, open the Account Dashboard and click Auto-Export.
2. Click Create New Export Job in the top right.
3. From the Data to export dropdown, select Measurements or Pan-42 Measurements. The Zero pad measurement gaps checkbox appears once either of these types is selected.

This option is not available for Catalog or Meters Measurements job types.

How It Works

A padded row is structurally identical to a real measurement row. It includes the device ID, device name, site ID, site name, timestamp, and resolution - but every numeric field is set to 0:

• For Measurements jobs: power(W), energy(Wh), current(A), voltage(V), power_factor.
• For Pan-42 Measurements jobs: all of the above plus reactive_power(VAR), consumed_active_energy(Wh), and frequency(Hz) when present.

PowerRadar fills three types of gaps for each active device in the export window:

• Header gap - intervals between the start of the export window and the device's first real measurement.
• Mid-stream gap - intervals between two consecutive real measurements that are more than one resolution step apart.
• Tail gap - intervals between the device's last real measurement and the end of the export window.

There is no gap-length ceiling. Every missing interval within the active window is padded, regardless of how long the device was silent. A device that went offline for 24 hours mid-window at 1-minute resolution will generate up to 1,440 padded rows for that gap alone.

The Data sample preview in the export configuration form does not reflect your zero-padding choice - it always shows the same fixed example. You will see padded rows in your actual export files after the first scheduled run.

What Gets Padded - and What Doesn't

Zero padding only activates for devices that sent at least one real measurement during the export window. One measurement is enough - there is no minimum count or percentage threshold.

If a device sent no data at all during the export window, it is omitted from the file entirely. PowerRadar has no record to anchor the padding to, so no rows are generated for that device - zero or otherwise.

In practice:

• A sensor with consistent reporting and a few missing intervals → all gaps filled with zero rows. ✓
• A sensor that went completely offline for the entire export window → not in the file at all.
• A sensor that came back online mid-window after a long outage → padded from the start of the window up to its first real measurement (header gap), and filled for any mid-stream or tail gaps too.

Late Data and Zero Rows

When zero padding is enabled, there is one scenario where the same timestamp can appear in two consecutive export files with different values: late-arriving measurements.

Here is how it happens:

1. The export job runs for a given window. A device has not yet reported for interval T, so the exporter writes a zero row for that timestamp.
2. The device's real measurement for interval T arrives late - after the export already ran.
3. The next scheduled export run picks up the real measurement and delivers it, producing a row for the same device_id and measurement_time(UTC) - this time with the actual values.

What this means for your downstream system: if you are ingesting export files into a data store, use measurement_time(UTC) combined with device_id (and phase for PAN-42) as the key and apply an upsert (insert or update) rather than a plain insert. A later file can overwrite a zero row from an earlier one with the correct value, and your system needs to handle that gracefully.

If your pipeline only ever appends rows and cannot overwrite, you may end up with duplicate entries for the same timestamp - one with zeros, one with real data. In that case, either disable zero padding or add a deduplication step that keeps the non-zero row when a conflict exists.

Zero Padding in PAN-42 Exports

PAN-42 meters report three phase channels (R, S, T) independently, and zero padding works at the phase level- not the meter level. Each phase channel is padded separately based on its own reporting history.

When zero padding is enabled, the exporter checks each phase channel for every timestamp in the export window. If a phase channel has no data for a given timestamp, a zero row is written for that phase at that timestamp- even if the other two phase channels on the same meter do have real data for that same moment.

What this means in practice: for a given device_id and measurement_time(UTC), you may see rows for all three phases - some with real values, some with zeros - depending on which phases reported data for that interval. This is expected behavior when zero padding is on.

How to read your data correctly: always use the combination of device_id, phase, and measurement_time(UTC) as the unique identifier per row. When you see a row with all numeric fields at 0 alongside rows with real values for the same device and timestamp, the zero row means that specific phase did not report for that interval.

Note - PAN-42 row counts grow quickly. A three-phase PAN-42 meter that is silent for an hour at 1-minute resolution generates up to 180 padded rows (3 phases × 60 intervals) for that hour alone. Factor this into your batch size planning.

Effect on File Size and Batch Splits

Padded rows count toward the batch size limit exactly the same as real measurement rows.

If your export contains 8,000 real measurements and 3,000 padded rows, the total is 11,000 rows. This exceeds the 10,000-row file limit for SFTP and FTPS - the export is automatically split into two files. Your downstream system needs to handle and reassemble split deliveries.

When to Turn It On - and When to Leave It Off

Turn it on when:

• Your BI tool, dashboard, or analytics pipeline expects a row for every time interval and breaks or shows incorrect results when rows are missing.
• You are performing a time-series join across multiple sensors and need all sensors to share the same set of timestamps.
• You want a clear signal in your data that a device was silent for a specific interval - a zero row is explicit; a missing row is ambiguous.

Leave it off when:

• Your downstream system handles sparse data on its own and doesn't need evenly-spaced records.
• You need to distinguish "the device genuinely reported zero watts" from "the device reported nothing." With zero padding on, both states produce a row with power(W) = 0, making them indistinguishable.
• You are running a PAN-42 export and your system requires a strict one-to-one match between real phase readings and rows in the file. In this case, leave zero padding off to ensure every row in your export corresponds to an actual phase measurement.

Related Articles

• Export Jobs: Measurements Interface
• Export Jobs: Pan-42 Measurements Interface
• Export Jobs: Setup
• Export Jobs: Pulse Meter Measurements Interface