I work as a field service technician on hydrogen systems for pilot plants, test labs, and a few manufacturing sites that run small gas skids around the clock. Most of my week is spent around tubing, regulators, stack enclosures, and cabinets where a leak can stay invisible until something else tells you the day is about to go wrong. That is why I never treat a hydrogen leak detector as a box to check off on a purchase order. I treat it like one of the few tools in the room that can catch a bad situation before my nose, ears, or instincts have any chance to.
What I look for before I trust a detector in the field
The first thing I care about is how the detector behaves in a real room, not how polished the brochure sounds. Hydrogen moves fast, rises fast, and sneaks through tiny faults that look harmless during assembly. I have seen a fitting pass a pressure hold test in the morning, then start whispering gas by the afternoon once the system warmed up and settled. A detector that reacts quickly and recovers cleanly tells me much more than one with a long feature list.
I usually start with the sensor type, the alarm logic, and the housing, because those three details decide whether the unit will survive beyond the first month. In an electrolyzer room, I want to know if the sensor drifts after repeated exposure and how easy it is to bump test it without shutting half the system down. Some sites want fixed detection tied into ventilation and shutdown relays, while others need a handheld unit for commissioning and maintenance rounds. Those are different jobs, and I do not pretend one device covers both well.
Placement matters more than many buyers admit. I like to walk the piping run, mark high points, look at cabinet ceilings, and pay attention to fans or louvers that can push gas away from where a detector was mounted by habit instead of thought. Six inches can matter. A customer last spring had a sensor mounted where the airflow from a cooling fan kept diluting the leak plume, and the detector looked lazy until we moved it to a quieter pocket near the top of the enclosure.
Why the source matters as much as the sensor
I have bought detectors through big industrial catalogs, local distributors, and a few specialized suppliers, and I can tell pretty quickly who actually understands hydrogen service. A general safety shop may sell the unit, but that does not mean the staff can answer basic questions about cross-sensitivity, calibration intervals, or whether the detector is suited for a humid cabinet that cycles temperature all day. I would rather work with a seller who can talk through the application than one who only repeats the data sheet back to me.
One resource I have pointed people toward for product comparisons is Wasserstoffleckdetektor, especially when they want to compare portable options without guessing from generic listings. I still read the specifications myself and match them against the site conditions. A clean product page does not replace engineering judgment, but it saves time when I am narrowing down which units are even worth bringing into a hazardous area review.
I also pay attention to what comes after the sale. Spare filters, replacement sensors, calibration gas availability, and clear manuals matter long after the cardboard box is gone. I learned that the hard way on a university test stand that used a detector nobody on site knew how to maintain, and after about 18 months it became a wall ornament with a blinking fault code. Cheap units can get expensive fast.
The mistakes I keep seeing during installation and setup
The most common mistake is treating detector placement like smoke alarm placement. Hydrogen is not smoke, and a room with active ventilation can turn a neat ceiling mount into dead space if the gas is being swept sideways before it collects. I usually sketch the room on paper, note the likely leak points, then check where the warm equipment is because convection can change the path more than people expect. That extra 10 minutes has saved me from bad installs more than once.
I also see people skip bump tests because the detector powers on without a fault. That tells me the electronics are alive, not that the sensing path is ready to catch a live leak. Two minutes is enough. During startup on a fuel cell bench, I once watched a brand-new unit sit silent during a controlled check because the alarm thresholds had been left in the wrong mode after factory setup. The screen looked fine until we challenged it.
Another problem is burying the detector in a maintenance plan that nobody owns. If a site cannot tell me who checks calibration every month, every quarter, or at whatever interval the manufacturer and process risk call for, I assume the detector will be neglected. Paperwork sounds dull, but that paperwork decides whether the sensor is trusted on day 400 or ignored because people learned the alarms were drifting. False confidence is worse than open doubt.
How I balance handheld detectors and fixed systems
I carry a handheld detector because I do not trust fixed systems to answer every question during service work. A fixed unit protects the room. A handheld helps me chase a suspect fitting, a valve stem, or a compression joint hidden behind cable trays and sample lines. They work together, and I am much calmer when I have both available during commissioning.
Handheld units are great for tracing a problem after a maintenance event, especially if someone opened a panel, changed a regulator, or remade three or four tube connections in a cramped cabinet. I move slowly, give the sensor time to respond, and watch for repeated peaks in the same spot instead of waving the instrument around like a wand. Slow beats flashy. I have found tiny leaks that way which would never have been obvious from pressure decay alone during a short test window.
Fixed systems earn their keep once the site is running routine operations and people stop paying close attention to every hiss and every fitting. I like fixed detectors tied into local beacons, ventilation, and a clear alarm sequence that operators can explain without opening a binder. If the alarm goes off at 2 a.m., the response needs to be obvious. A detector is only useful if the people on shift know what it is telling them and what they should do next.
After enough years around hydrogen skids, I have stopped looking for a perfect detector and started looking for one that fits the room, the people, and the maintenance habits it will live with. The best unit I have ever used was not the fanciest one, but it responded fast, stayed stable, and made sense to the crew that depended on it every day. That is usually the real test. If I can walk into a site six months later and see that the detector is still trusted, still checked, and still placed where it can do its job, I know the choice was probably the right one.