SEL Event Report Format: EVE, CEV, and COMTRADE Explained

An SEL relay does not write one thing when it records a fault. It writes several, they have different resolutions, and the one that arrives in your inbox is usually the lowest resolution of the set. Engineers routinely conclude that a record is unusable for waveform work when in fact a much better version of the same event was still sitting on the relay.

This page describes what each form contains, which command produces it, and the single distinction that causes the most wasted analysis: filtered against unfiltered data.

The two families of SEL event report

Schweitzer Engineering Laboratories relays expose event data in two fundamentally different ways.

• The ASCII event report. A text report retrieved over the serial or Telnet interface with the EVE command. It is a table: one column per analog quantity, one character column per digital element, one row per sample, followed by a summary and the settings that were in service. It is readable without any software, which is exactly why it circulates by email.
• The binary COMTRADE report. A standard file set conforming to IEEE C37.111, which the SEL-421 manual cites as C37.111-1999. It is not readable in a terminal and it is not retrieved with a command that prints to screen. As that manual puts it, file transfer is the only mechanism for retrieving high resolution COMTRADE data from the relay.

Both are generated from the same captured event. They are not interchangeable, because the ASCII report is decimated and often filtered while the COMTRADE capture is not.

The commands, and what each one returns

Command syntax varies across the SEL range, so treat the table below as the shape of the scheme rather than as a universal reference, and confirm against the instruction manual for your specific relay. The resolutions quoted are those documented for the SEL-351.

The nnnnn in the COMTRADE filenames is the unique event identification number shown by the event history command, which is how you match a COMTRADE capture to the row in the history that describes it. If you have a trip target and a timestamp but no waveform, that number is the thing to ask the substation for.

Filtered against unfiltered: the distinction that matters

This is the part that quietly ruins analyses. A filtered event report has been through the relay’s digital filter before it was written. An unfiltered report, requested by appending R to the command, holds the sampled values as the relay took them.

They answer different questions, and only one of them is usually the question you have.

• Filtered data tells you what the relay saw. If you are asking whether an element should have picked up, whether a distance zone was correctly restrained, or why a timer started when it did, filtered data is the correct source, because it is the input the protection logic actually acted on.
• Unfiltered data tells you what was on the wiring. CT saturation, decaying DC offset, transformer inrush and harmonic content are all features the digital filter exists to suppress. Look for them in filtered data and you will not find them, and you will wrongly conclude they were absent.

The trap is that a filtered record looks perfectly healthy. It plots as clean sinusoids because that is what filtering produces. Nothing on screen announces that the raw current was a saturated waveform collapsing to near zero for part of every cycle. If you are investigating a suspected instrument transformer problem and your record came from a plain EVE command, the evidence you want was removed before the file was written.

The SEL-351 manual makes the same point from the other direction for voltage: with the PT connection set to delta, unfiltered event reports are the only means by which the signals applied to the individual voltage terminals can be directly observed. Some measurements simply do not exist in the filtered form.

Four samples per cycle is not a waveform

The default EVE output on an SEL-351 is four samples per cycle. That is enough to read the order in which elements asserted and to see roughly when current changed magnitude. It is not enough for waveform analysis, and the reason is arithmetic rather than opinion.

Four samples per cycle can represent the fundamental and nothing above the second harmonic. The second harmonic ratio is the standard discriminator between transformer inrush and an internal fault, and the odd harmonic signature is a principal indicator of CT saturation. Neither survives at that rate. A record at four samples per cycle can support a statement about sequence of operation and cannot support a statement about waveform shape, and it is worth being explicit about which of the two you are making.

For context, the SEL-351 offers 16 samples per cycle in both filtered and unfiltered form, 32 samples per cycle unfiltered from EVE R, and 128 samples per cycle unfiltered in the high resolution COMTRADE capture. The SEL-421 expresses the same idea in absolute terms, with selectable effective sampling rates for its high resolution raw COMTRADE storage. If someone sends you a four sample per cycle report, a far better record of that same event very probably still exists on the relay.

Digital channels, debounce, and why two sources disagree

A detail worth knowing before you time anything to the millisecond. The high resolution COMTRADE capture records the raw control input status, while the standard event report and the Sequential Events Recorder show input state after the debounce timer has been applied.

The SEL-421 manual states this directly: control input state changes appear to occur faster in COMTRADE oscillography files than in event reports or SER reports, because of the control input debounce time delays, and contact bounce may be visible when the raw data are viewed. On the SEL-351 the pre debounce status of the input contacts can be viewed by using the EVE R or CEV R commands, which the manual recommends for analysing contact bounce problems with connected equipment.

So when a breaker auxiliary contact appears to change state a few milliseconds earlier in one record than another, that is not a clock error and not a fault. It is the debounce timer, and the raw capture is the one that reflects the physical wiring.

Getting an SEL event into a waveform tool

A .cev file is a compressed ASCII event report. It is not COMTRADE, and renaming it will not make it COMTRADE. There are two routes.

1. Convert the event report. The SEL acSELerator Analytic Assistant, distributed as SEL-5601, reads event reports and exports COMTRADE. SYNCHROWAVE Event is the SEL program the manuals point to for viewing COMTRADE data directly.
2. Retrieve the native COMTRADE capture instead. Where the relay stores one, pulling the .CFG, .DAT and .HDR set by file transfer is the better path. It is unfiltered, it is higher resolution, and it avoids a conversion step that can only lose information.

Once the event is in COMTRADE it is portable, and the vendor question stops mattering. The same file opens in a COMTRADE viewer alongside records from any other manufacturer, which is what makes multi ended and multi vendor events analysable at all. The SEL-421 manual is candid about this too, noting that many third party software suppliers can provide programs to display and manipulate COMTRADE files.

Two header details that surprise people

Both come from the SEL-421 revision history and both change what a downstream tool sees.

• Digital channel count is a setting. The event report digital setting, ERDIG, can be set to some or all, controlling how many Relay Word bits are written into COMTRADE event reports. A record that seems to be missing the element you wanted is often a record made with the narrower setting, not evidence that the element never operated.
• Station identifiers are sanitised. The station ID label in the COMTRADE configuration file is stripped of non alphanumeric characters to comply with C37.111-1999. If the station name in the file does not match your naming convention exactly, that is usually why.

What to ask for

If you are requesting a record from site and you intend to do waveform analysis rather than sequence of operation analysis, the request that avoids a second trip is this: the native COMTRADE capture for the event, all three files including the header, plus the event history entry so the event number can be matched. If only an event report is available, ask for the unfiltered form.

The general method for working through the record once you have it is set out in how to read a fault record, and what is a COMTRADE file covers what the CFG and DAT files contain once the format question is settled.

Frequently asked questions

What is an SEL event report?

The record an SEL protective relay writes when a trigger condition occurs, in two families: a human readable ASCII report retrieved with the EVE command, and a binary COMTRADE file set retrieved by file transfer. Same event, different resolution.

What is a CEV file?

A compressed ASCII event report from an SEL relay. It holds the same content as the equivalent EVE output in a more compact encoding, and it needs converting before a general waveform tool can read it.

Is an SEL event report the same as COMTRADE?

No. COMTRADE is the vendor neutral IEEE C37.111 interchange format. The SEL ASCII event report is a proprietary text layout. Many SEL relays can produce both from the same event, and the COMTRADE version is normally the higher resolution one.

Which report should I ask for?

For sequence of operation and element behaviour, the filtered event report, because it shows what the relay acted on. For anything involving waveform shape, saturation, harmonics or DC offset, the unfiltered high resolution COMTRADE capture, because the filtered form has already removed what you are looking for.