Phoenix Journal · LEV Testing
On-torch, on-arm and ambient extraction are not interchangeable - they sit on a scale from capture-at-source to general dilution. Here is how to choose, and how to prove your choice keeps welders out of the plume.
Welding fume control
Choosing between on-torch, on-arm and ambient extraction is not really a kit decision - it is a question of how close you can get the capture to the arc, and how you prove it works.
Since the HSE changed its enforcement expectations in 2019, all indoor welding needs effective engineering control, and mild steel fume is treated as a carcinogen alongside stainless and other alloys. That reclassification took the debate away from "do we need extraction at all" and moved it firmly onto "which method actually keeps the welder out of the plume, and can you demonstrate it". The three broad options - on-torch, on-arm and ambient - are not interchangeable. They sit on a scale from capture-at-source to general dilution, and the further you drift from the arc, the harder the numbers become.
The reason the numbers matter is manganese. It is present in almost all mild steel fume, and the workplace exposure limit is now demanding: 0.05 mg/m³ for the respirable fraction as an 8-hour time-weighted average, with 0.2 mg/m³ inhalable. The respirable figure is roughly a tenth of the old limit, and the HSE is candid that it will be exceeded during many welding tasks unless controls are genuinely effective and used properly. That single line is what should drive your choice of extraction - not the sales brochure.
On-torch extraction puts the capture point millimetres from the arc, built into the MIG torch itself. From a pure capture-efficiency standpoint it is the strongest option, because nothing beats being at the source. Done well it pulls fume before it ever reaches the breathing zone, and the welder keeps full freedom of movement without stopping to reposition a hood. The catch is set-up. Capture depends on maintaining enough induced airflow at the nozzle - research points to keeping induced velocity at or above roughly 0.25 m/s and torch inclination within about 30° - and if the vacuum drops, the shielding gas is disturbed, or the welder angles the torch too far, capture falls away quickly. It suits repetitive MIG work far better than intricate positional welding where the torch is constantly turning.
On-arm extraction - the articulated hood on a flexible arm - is the workhorse of most fabrication shops. It is flexible, it copes with changing workpieces, and a single unit can serve a range of tasks. But it is only as good as its placement. The rule of thumb is to keep the hood within about one duct diameter of the weld and directly over the rising plume, so an eight-inch arm wants to sit around eight inches from the work. Move the job, and the hood has to move with it. In practice the biggest failure mode is a hood left parked too far back while the welder chases the seam - capture collapses with distance, and the welder never notices because the arm looks like it is doing something.
Ambient or general extraction cleans the whole room rather than the source. It has a place - controlling background haze, protecting others in the workshop, and handling situations where source capture genuinely cannot reach - but it is supplementary, not a primary control. Because it relies on diluting and moving large air volumes, it does little to stop fume passing through the welder's breathing zone on its way up. The hierarchy is settled: capture at source first, room ventilation second. Leaning on ambient alone is where COSHH assessments tend to come apart.
The numbers that decide it
Whichever method you pick, the same physics applies: extraction velocity drops away sharply with distance from the source, so a control that is effective at 100 mm can be near useless at 300 mm. The figures below are the ones an inspector - and your LEV test engineer - will care about.
Those three numbers frame every decision. The exposure limit tells you how hard the control has to work, the 14-month interval tells you the control has to keep working and be proven to, and the reclassification tells you it applies to every welder, every material, indoors as standard.
For most shops the honest answer is a combination. On-torch or a well-placed on-arm system does the heavy lifting at the source; ambient extraction mops up the residual background; and where even good source capture cannot get exposures below the limit, it is supplemented with suitable respiratory protective equipment. RPE is also the default for outdoor welding, where you cannot rely on any fixed extraction at all. The point is that RPE supports engineering control - it does not replace it, and a face mask on its own is not a compliant answer to a carcinogen.
Method choice is only half the job. Buying good extraction and then bolting it to the wall is one of the most common - and most expensive - mistakes we see. Fume control is a system: the right capture method, positioned correctly, used consistently, maintained, and backed by training and supervision so the welder actually keeps the hood where it belongs. If you want to go deeper on that wider duty, our guide on how to manage welding fume beyond just buying extraction walks through the parts a purchase order alone will never cover. And where the work moves into tanks, vessels or other tight enclosures, the calculus shifts again - see welding in confined spaces and managing the risks, where dilution ventilation and rescue planning matter as much as the extractor itself.
Whatever you install, the law expects you to demonstrate it works. Under COSHH Regulation 9, any LEV system - on-torch, on-arm or ambient - must have a thorough examination and test at least every 14 months, carried out by a competent person who checks capture performance against how the system is actually used, not just that the fan spins. That test is your evidence that the control still pulls fume where it needs to, and it is the first thing an inspector will ask to see. A system that passed on paper two years ago, with a hood now parked a foot too far from the arc, will not survive scrutiny and will not be protecting your welders.
Questions
For capture efficiency, on-torch extraction is the strongest option because it sits millimetres from the arc, so fume is pulled before it reaches the welder's breathing zone. It suits repetitive MIG work and lets the welder move freely. On-arm hoods are more flexible across changing workpieces but only work when kept close to the plume - roughly one duct diameter away - so the best choice depends on the type of welding you do.
No. Ambient extraction dilutes and moves large volumes of air across the whole room, but it does little to stop fume passing through the welder's breathing zone as it rises. The control hierarchy is clear: capture at source first, room ventilation as a supplement. Ambient systems are useful for background haze and protecting others nearby, but on their own they will not meet a COSHH assessment for a carcinogen.
The UK limit is 0.05 mg/m3 for the respirable fraction as an 8-hour time-weighted average, with 0.2 mg/m3 for the inhalable fraction. The respirable figure is around a tenth of the previous limit. The HSE is clear that this limit will be exceeded during many welding tasks unless controls are genuinely effective and used correctly, which is why capture-at-source matters so much.
Under COSHH Regulation 9, any local exhaust ventilation system - on-torch, on-arm or ambient - must have a thorough examination and test at least every 14 months by a competent person. The test checks capture performance against how the system is actually used, not just that the fan runs. It is your evidence the control still works and the first thing an HSE inspector will ask to see.
Phoenix Duct Clean · by the numbers
Phoenix examines and tests local exhaust ventilation to HSG258 and COSHH - measured, reported and certificated, UK-wide.