Cool the tip with a closed glycol loop and leave the compressor out of it. A vortex tube sized for a small camera housing drinks about 8 SCFM of dry, oil-free air at 100 PSIG [6.9 bar] to deliver 275 BTU/h (81 W). The 1-2 bar, low-flow scheme usually tried first moves almost nothing: 0.5 SCFM of cool air is worth about 3 W. Feeding that tube continuously means roughly 1.5 kW of compressor input, plus a dryer and filters. The same 24 W duty is a 1 L/min glycol loop, a magnetic-drive pump and one 120 mm radiator at the crane base, where you have room, drawing about 25 W. Here is the arithmetic for both routes, and the housing that holds the camera at 52 °C in a 105 °C ambient.
The duty you have to hit
Start with three numbers: the camera's heat, its temperature limit and the two ambients. A machine-vision camera dissipates 8-15 W; this example uses 15 W. Most industrial camera datasheets allow 60 °C operation, so hold the case at 52 °C and keep 8 K of margin. Call the tip 105 °C and the crane base 40 °C, where the machinery lives.
Size the body at 200 × 150 × 120 mm [8 × 6 × 5 in] outside, which is 0.14 m² of surface. A bare stainless wall, at the 3.7 W/m²K reference value this site uses for stainless, lets in 3.7 × 0.14 × 55 ≈ 28 W, nearly twice the camera's own load. Wrap the walls in 15 mm of silica aerogel blanket, k ≈ 0.02 W/m·K, good for 200 °C service where polymer foams are already at their limit. The wall assembly then falls to about 1 W/m²K, and the leak drops to 1 × 0.14 × 55 ≈ 8 W. A dual-pane window with a 6 mm dry gap, 60 × 60 mm clear, adds about 1 W.
Add it up: 15 + 8 + 1 = 24 W, which is 82 BTU/h. Put that number on your spec, because everything downstream sizes from it.
What a vortex tube costs in air
Vortex tubes are sold in sizes that consume 2-100 SCFM at 100 PSIG. The smallest enclosure coolers deliver about 275 BTU/h on 8 SCFM, and the largest reach 5,600 BTU/h on up to 100 SCFM with cold-end air down to -40 °F [-40 °C]. Against an 82 BTU/h duty, the smallest unit is still three times oversized, so it cycles on a thermostat instead of running flat out.
The air is the real bill. Eight SCFM is 480 scf per hour, and 6,000 running hours — 250 days at 24 h — burn 6,000 × 480 ≈ 2.9 million scf. A compressor delivers roughly 4 SCFM per horsepower at 100 psi, so you would be feeding a 24 W load with about 2 hp, call it 1.5 kW of input. Over 6,000 running hours that is 9,000 kWh, against 150 kWh for the pump-and-fan loop. Add a dryer sized for the full flow; heatless desiccant spends 15-20% of its air on purge. In dusty crane service, coalescing filters come due around every three months.
The air has to be dry as well. Air saturated at 100 psi sheds water in the hose and again at the cold end, the worst possible place, right beside optics. Keep the hot end open to atmosphere, because back pressure trims the output.
Why a 1-2 bar line cools almost nothing
One SCFM of air carries about 0.6 W per kelvin. Half an SCFM arriving 10 K below the box temperature moves 3 W, and carrying the full 24 W at that same 10 K would take about 4 SCFM of genuinely cool air. After 6 m of hose through a 105 °C stretch, unconditioned air arrives near hose-wall temperature, and the difference you paid for is gone before the box. Low-pressure air earns its keep one way: slight positive pressure that keeps grit out of a breather path, and a sealed window removes even that job.
Seal the lens behind the dual pane. The dust then lands on the outer pane, where you wipe it on a schedule instead of chasing it inside the box.
The compressor, if you go air anyway
Air wins in one case: a dry 6-7 bar header already on the crane. Take a 10 SCFM branch through a point-of-use filter, regulator and dryer. Run the smallest enclosure cooler on its maker's relief valve and thermostat, and the housing stays sealed between cycles.
Buying machinery for 8-10 SCFM continuous puts you in rotary-screw class. Oil-free airends are priced 30-65% above oil-flooded equivalents and last 20,000-40,000 h against 40,000-80,000 h. You can also buy oil-freeness with filtration. ISO 8573-1 grades compressed air by oil content, and its Class 1 grade allows 0.01 mg/m³; coalescing plus carbon stages after an oil-flooded screw get you there. Either machine is 1.5 kW of hardware moving 24 W out of one small box.
Small piston compressors sit at medium duty, 50-75% duty cycle, so continuous service pushes a single piston past its class. A duplex pair alternating covers the flow, with receivers and dryers to maintain as well. The numbers point one way: no new compressor for a 24 W load.
The glycol loop, with the arithmetic
Fill the loop with 30% propylene glycol: specific heat 3.8 kJ/kg·K, freezing point about -15 °C, inhibitors inside, dye added so a weep shows colour. One litre per minute is 0.0172 kg/s and carries 0.0172 × 3,800 ≈ 65 W/K, so the 24 W tip duty warms it only 0.4 K on the way through.
Bolt the camera to an aluminium cold plate on the rear wall, through a bracket under 0.5 K/W, and the case sits 15 × 0.5 ≈ 7 K above the plate. Run two lines 6 m each way: PTFE-core or silicone hose for the hot metres, both fine at 105 °C where polyurethane is not. At 8 mm bore with 10 mm of aerogel wrap, each line loses about 0.1 W per metre per kelvin. Keep the runs on the cooler boom with only the last 3 m of each in the 105 °C stretch, where they pick up 2 × 3 × 0.1 × 60 ≈ 36 W.
The radiator therefore sees 24 + 36 = 60 W. A 120 mm fan-forced liquid-to-air exchanger runs about 15 W/K, or 0.05-0.1 °C/W on a typical datasheet. The loop then holds 60 ÷ 15 = 4 K above the 40 °C base air: coolant at 45 °C, plate at 45 °C, camera case at 52 °C, every surface above the dew point so nothing wets. A 15 W magnetic-drive pump and an 8 W fan power it, about 25 W in total.
Leaks are the usual fear, so design them visible. Run the loop at about 1 bar, fit dry-break couplings at the housing, and put a pressure switch on the loop to alarm on loss. A weep of dyed glycol shows colour long before it drips, and you can drop the box at the tip without draining the system. The same reasoning behind closed-loop enclosure cooling where fumes are in the air applies here; for larger heat loads, see cooling a fanless vision computer at hot sites.
The housing you put on the drawing
The body is a one-piece welded stainless box with no breather and no drain. Choose 316L where the site sees salt air or chlorides from washdown and pickling; 304 serves an inland mill. It is rated NEMA 4X / IP66 by the manufacturer and not UL listed. In a US spec, NEC 110.28 keeps an IP code from standing in for the enclosure type, so the type number goes on the drawing.
The IP66 jet test in IEC 60529 drives a 12.5 mm nozzle at 100 L/min from 2.5-3 m for at least 3 min. Your two bulkhead fittings, the Ethernet gland and the window gasket each have to hold that level, because the assembly takes the rating of its weakest part. The cutouts, glands and window guide sizes the holes, and window seal spacing on long panes covers wide glass. A lens facing spray takes its gasket thinking from the camera enclosure built for CNC coolant. Keep the window gasket in silicone, good to 200 °C, where EPDM tops out at 150 °C.
Before you open the box at height, isolate the camera supply and prove it dead, and treat every surface above 60 °C as a burn hazard. A sachet of indicating desiccant finishes the interior, since a sealed box has no new moisture source.
What to order
| Line | What to specify | The number that sizes it |
|---|---|---|
| Body | One-piece 316L (or 304 inland), welded seam, your cutouts | 200 × 150 × 120 mm [8 × 6 × 5 in] outside |
| Rating | Sealed, no breather, no drain | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| Window | Dual pane, 6 mm dry gap, silicone gasket | 60 × 60 mm clear aperture |
| Insulation | Silica aerogel blanket on walls and lid | 15 mm |
| Cooling interface | Bolted aluminium cold plate, two 1/4 in bulkhead fittings | Camera bracket < 0.5 K/W |
| Cable entry | One Ethernet gland, chosen with the box | Holds the body's rating |
| Interior | Indicating desiccant sachet, camera on the plate | Case at 52 °C at 105 °C ambient |
| Base skid | Magnetic-drive pump, 120 mm radiator, pressure alarm | 60 W at 4 K approach |
Order the sealed body through custom enclosures. You set the material, the three dimensions and every cutout; it is built from one piece and delivered in 10 business days anywhere in the world, counted from your approval of the drawing. Buy the wet side — plate, couplings, pump and radiator — from a liquid-cooling supplier against the 60 W duty. If the inside layout is still open, the enclosure selector checks the body size around the camera and the plate.
Catalog dimensions to compare
Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.
| Enclosure | Material | Outside H x W x D / display | Rating |
|---|---|---|---|
| 16 × 12 × 8 in | 304 stainless | 16 x 12 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 24 × 20 × 8 in | 304 stainless | 24 x 20 x 8 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 30 × 24 × 10 in | 304 stainless | 30 x 24 x 10 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
| 10 × 8 × 6 in | 304 stainless | 10 x 8 x 6 in | Rated NEMA 4X / IP66 by the manufacturer; not UL listed |
Questions
Will a vortex tube run on a 1-2 bar supply?
No. Makers publish their flow and cooling figures at a 100 PSIG [6.9 bar] inlet, and the small units draw about 8 SCFM there. At 2 bar you get a mild draft with no useful cold fraction. If the site offers only 2 bar, the glycol loop does the cooling and the air line stays for tooling.
How long would a bottled CO2 supply cool the camera?
About two days of continuous duty per 13.6 kg [30 lb] bottle. Liquid CO2 gives up roughly 300 kJ of usable cooling per kilogram as it expands, so 13.6 kg holds about 4.1 MJ; at a 24 W duty that is 4.1 MJ ÷ 24 W ≈ 47 hours. It suits short campaigns and intermittent vision tasks, not a crane running all shift.
Water or glycol in the loop?
30% propylene glycol. Specific heat 3.8 kJ/kg·K, freezing point about -15 °C, corrosion inhibitors included, and dye so a weep is visible. Plain water carries 4.2 kJ/kg·K but freezes and grows algae, while straight glycol carries less heat and pumps harder. Plan a coolant change every two to three years.
Does a housing with coolant fittings still carry a NEMA type?
Yes, when the fittings, gland and window gasket hold the same level as the welded body, because the assembly takes its rating from the weakest part. The body here is rated NEMA 4X / IP66 by the manufacturer and not UL listed, and the IP66 jet test means a 12.5 mm nozzle at 100 L/min for at least 3 min — powerful jets, never immersion.
Does positive air pressure keep the lens clean?
Only a little, and only while the air flows. Half an SCFM of filtered air is worth about 3 W of cooling at 10 K and holds slight positive pressure against grit at a breather path; in a sealed box there is no such path and nothing to purge. Seal the lens behind the dual-pane window and wipe the outer pane on schedule.
What if the crane base is as hot as the tip?
Add a chiller at the base and keep the tip hardware identical. A small vapor-compression chiller runs a coefficient of performance of 2-3, a Peltier liquid chiller about 0.4-0.7, both sized for the 60 W loop duty. That is the whole point of putting the machinery at the base, where size and service access stop mattering.
