Notes from the loop · leak detection

Choosing leak detection for a secondary loop

A conductive sensing rope reports distance by measuring resistance between two wires, and that sentence decides what it can detect.

3 decisions, in order · position resolution 0.1 m · locating repeatability ±0.6 m plus 0.25 % of channel

Buyers usually start with the third one. Getting the sensing principle wrong makes the siting irrelevant; getting the siting wrong makes a correct principle alarm twice a day.

We supply the loop between the coolant distribution unit and the cold plate; neither part is in the catalogue.

Ask for the specification sheet

Coiled conductive leak detection sensing cable with its moulded connector on a workbench

Conductive sensing rope · the two wires the distance is read between

01 / Sensing principles

Each principle, and what it detects

What each principle detects, and how precisely it reports.
Principle	Detects	Locating
Conductive, distributed	Any conductive fluid: water, glycol blends, acids and bases	Distance-read
Capacitive	Water and non-conductive liquids	Zone or point
Fibre-optic, distributed	Water and hydrocarbons; some variants intrinsically safe	Metre scale
Pressure or mass balance	A loss of fluid anywhere in the loop	Loop level only
Acoustic or ultrasonic	Pressurised spray and jet leaks	Approximate

Each one detects

Conductive rope dominates data-centre white space because it is the cheapest way to get a distance reading. Pressure and mass balance catch a micro-leak that never reaches a floor, and they cannot say which rack.

02 / The fluid decides

The most common mis-selection in this category

The glycol floor

A nominal 25 % propylene glycol mix is the design basis for most single-phase direct-to-chip deployments. Sensing cable in this class detects blends at 15 % propylene glycol by weight or higher, covering PG15, PG25 and PG35.

Where the signal stops

One family is tested across 10 % to 55 % propylene glycol mixtures: the loop's concentration decides the envelope. Near-deionised water and dielectric fluids give weak or no signal. A DI loop needs the principle chosen for that fluid, and no accessory fixes it.

03 / Siting

Installation rules cause more trouble than the electronics

Sensing cable secured with clips along an underfloor run, receding into the distance

The clearance that removes most nuisance alarms

Keep sensing cable 6 ft, about 1.8 m, clear of the front of a downflow air unit's discharge. Cable inside that plume alarms on contamination and cannot be located: the published complaint is an alarm twice a day, with no water.

Clearance 1.8 m (6 ft) · clips 0.9–1.2 m · tags 3 m · 500 m channel: 417–556 clips, 167 tags

Hall in section, discharge plume and cable run

04 / Why the panel alarms

An alarm with no water

The threshold arithmetic

The two setpoints

A leak setpoint of 150 µA for 20 seconds, and a contamination setpoint of 50 µA for 120 seconds. Water bridging the wires drives 300 µA.

The contamination band

Material that conducts poorly, solder flux being the classic, attracts moisture and produces 1 to 300 µA. That band crosses the leak setpoint with no water present. The contamination setpoint is adjustable, typically 25 to 300 µA.

Remedy: isopropyl wipe, or a warm soapy wash and 6–48 h drying

05 / What 0.1 m means

Position resolution is not locating repeatability

How to read the table

Position resolution is the granularity of the reported distance, and it stays at 0.1 m on every channel length. A 500 m channel and a 50 m one report the same resolution and different accuracy.

Three figures quotations merge. The length in both formulas is the whole channel.
Channel length	Position resolution	Locating accuracy	Locating repeatability
50 m	0.1 m	±0.85 m	±0.73 m
100 m	0.1 m	±1.10 m	±0.85 m
250 m	0.1 m	±1.85 m	±1.23 m
500 m	0.1 m	±3.10 m	±1.85 m

Accuracy = ±0.6 m + 0.5 % of channel · Repeatability = ±0.6 m + 0.25 % of channel

06 / The accessory bill

What else has to be bought

Leader cable and terminator

A 15 ft (4.57 m) leader cable and an end-of-line terminator, both required for a supervised loop.

One of each per channel

Non-sensing jumper

10 ft (3.05 m), wherever sensing cable crosses over itself. A distance-read channel averages two wet points on one cable: a leak touching it twice is reported once, in the wrong place.

One per crossover

Clips and tags

J-clips every 0.9–1.2 m, tags every 3 m.

500 m channel: 417–556 clips

Power and enclosure

Power supply, an enclosure or rack bracket, extension cable.

One enclosure per panel

07 / Maintenance and settings

What to test, and what to set

Test annually

Test cable break and leak detection at least once a year, monthly preferred. Monitor cable current monthly where the panel logs it.

500 to 1,024 events

Controllers in this class store the last 500 to 1,024 events and trend cable current daily for 288 to 365 days.

5 to 990 seconds

Adjustable in 5-second steps, ±2 seconds, on distance-read controllers; fixed at 10 seconds or 2 minutes on single-zone units.

Set the response time for the space.

Scope · no coolant distribution units, no cold plates

08 / The codes behind this post

The selection, from the codes up

EMS-LD-R500 Locating sensing rope, 500 m Resolution 0.1 m; accuracy ±3.1 m, repeatability ±1.85 m. PDF
EMS-LD-P2 Surface point probe with local alarm Point report, no distance along the run. PDF
EMS-LD-C8 8-channel distance-read panel Response 5–995 s in 5 s steps. PDF
EMS-LD-C2 2-channel zone panel Response 10 s or 2 min. PDF
EMS-LD-X15 Non-sensing extension, 15 m Covers no located run. PDF
EMS-LD-T1 End-of-line terminator One per loop and one per controller. Without it a cut cable reads as a quiet loop. PDF
EMS-TL-K300 Loop telemetry kit, row 100 to 300 kW Matched pair, flow, dP, leak channels. PDF

08 / Specification sheet

Send the loop fluid and the channel length

Ask for the specification sheet

The reply returns the specification sheet, the drawing and the datasheet for the code.

Ask for the specification sheet Leak detection