Test & Calibration Report

Firmware v1.4, build 212d560. Testing carried out 25–28 August 2026.

Every figure in this document was measured. Nothing is estimated, extrapolated or taken from a datasheet. Where a limit was not reached, it says so.

Method

Three independent sources of evidence were used, so that no single instrument's opinion could go unchallenged.

The file analysis is the part that matters most. A recorder that reports its own success is not evidence. Each capture was pulled from the card and independently verified — frame count recomputed from file size, refresh rate recomputed from frame count and duration, and every frame compared against the one before it.

1. Recording integrity

Continuous unattended recording of a full 512-slot universe, on mains and on battery, with a dropped-frame counter running throughout.

Total: 5,290,145 consecutive frames across 36.7 unit-hours and 2.7 GB written. Zero frames dropped.

Sustained write rate was 20.0 KB/s against a card rated at 10 MB/s — a margin of roughly 465×. Throughput is not the constraint; the capture buffer holds 364 ms, so only a stall longer than that can lose a frame, and none occurred.

Every file passed independent verification: header consistent, all frame data present, no trailing data, refresh rate recomputed within 0.01 Hz of the stored value, and the universe demonstrably changing throughout.

What this test could have failed

It did, twice, during development — and both faults were invisible from the unit's own screen. They were found because the files were checked independently rather than trusted. Both are fixed in this build, and the results above are from firmware carrying those fixes.

2. Timing accuracy — transmit

VERSE transmitting, measured simultaneously by the Swisson XMT-120A and by a second VERSE inline on the same cable.

Every break agrees within 1 µs. Every mark agrees within 1 µs. Every refresh rate agrees within 0.1 Hz. Slot count read 512 on both instruments in all three modes.

Against the DMX512-A (ANSI E1.11) minimums of 88 µs break and 8 µs mark-after-break, VERSE transmits at approximately twice and two and a half times the minimum.

3. Timing accuracy — receive

The same comparison in the other direction, both instruments watching one source.

The source here was a third-party controller with an unusually long mark. Both instruments reported it, and agreed. That is the useful result: VERSE does not flatter a signal, and does not invent a fault that is not there.

4. Unit-to-unit agreement

Two VERSE units reading the same line at the same time, driven by the Swisson as a known-good source.

Two independent units agree to within 1 µs.

This also establishes something about the instrument itself. Reading a rock-steady source, VERSE reports 1 µs of spread on break and 2 µs on mark — essentially its 0.5 µs resolution floor. The same unit reading a poorly-behaved source reports a spread of 109–6970 µs. The instrument is not adding jitter of its own; it is reporting what is there.

5. Interrupted recording

A recorder used unattended must survive losing power. Tested on two units, by two methods — a firmware reset and a hardware power cut — with recordings running.

Verified independently: all four interrupted files were confirmed intact off the card, with frame counts recovered correctly.

6. Fault detection

A budget DMX console was connected deliberately. VERSE logged, unprompted:

58634  MAB_SHORT   1 us (minimum 8)   x67
62635  MAB_SHORT   0 us (minimum 8)   x131
70635  MAB_SHORT   0 us (minimum 8)   x256
86635  MAB_SHORT   4 us (minimum 8)   x512

That console emits a mark-after-break of 0–4 µs against a required minimum of 8 — frequently none at all. VERSE identified it within seconds and described it in plain words, without being told what to look for.

Summary

What is not claimed

Stated plainly, because a report that only makes claims is worth less than one that marks its own boundaries.

• This is not a calibration certificate. It is a cross-check against one professional reference instrument, on one bench, on one day.
• No traceable calibration to a national standard is claimed or implied.
• VERSE is not certified test equipment and carries no accuracy specification. The figures above describe measured agreement, not a guarantee.
• Recording limits were not reached. Ten hours is the longest run performed, not a maximum. A 32 GB card holds roughly 395 hours of a full universe, a 128 GB card over 1,500. On a FAT32 card a single recording is capped at 4 GB — about 56 hours — by the filesystem; exFAT cards, which is how 64 GB and larger cards ship, have no practical single-file limit.
• Card performance varies. Testing used A1-rated cards. A slower or worn card may stall longer than the 364 ms capture buffer. VERSE counts any frame it drops and reports it on screen and in the file, so a lossy recording can always be told from a clean one.