Brazing brass NPT hubs onto a submersible junction box

Braze brass NPT couplings into the box wall, then thread a submersible-rated brass cable gland into each one: 1/2 in and 3/4 in NPT both cut 14 threads per inch under ASME B1.20.1, so either size gives you the standard entry you are after. The converted joint holds water at two seals — the braze fillet where the coupling meets the wall, and the gland's compression seal on the cable jacket — and the third path, wicking through the strands, is stopped by the cable, not the box. Freshwater loads every one of those seals at 0.433 psi per ft of depth, so a box at 20 ft lives with about 8.7 psi and you prove the work at 13 psi. Get the box maker's brazing process in writing before heat touches the shell.

Can a standard conduit hub replace the factory's brazed connector?

For the thread, yes; for the seal, no. A conduit hub puts a 1/2 or 3/4 in NPT thread on the wall but carries no elastomer against the cable jacket, so water walks straight in along the cable at 0.433 psi per ft of depth.

The factory connector is two parts in one: a brazed feedthrough and a cable seal. Your conversion splits that job — the coupling carries the thread, a submersible gland does the sealing. Fittings made for flexible liquid-tight conduit clamp a convoluted conduit, and their geometry never matches a round jacketed cable, so leave them off this build.

On a hose-rated box the question plays out differently, and conduit hub vs sealing locknut for outdoor top entries works through that case. For genuine immersion, treat sealing penetrations in a submersible enclosure as the companion read: every opening you add is a seal you have to prove.

How much pressure does each entry have to hold?

Freshwater pushes 0.433 psi per ft of depth (water at 62.4 lb/ft³) and seawater about 0.445 psi per ft, so your maximum depth times that factor is the load on every seal at once. Flip it around: 1 psi of head per 2.31 ft of freshwater. Every extra 10 ft of depth adds 4.3 psi, and depth is the only variable that moves the number.

That load lands on three places at once: the braze fillet, the NPT thread, and the gland seal on the jacket. A fourth path runs inside the cable, where the same differential pressure pushes water along the conductor strands and the core even with a perfect jacket seal.

For scale, IEC 60529's second digit 7 means 1 m [3.3 ft] for 30 min, digit 8 means continuous immersion on conditions agreed between maker and user, and IP66 is powerful jets — an IP66 box is an above-water rating. NEMA 4X is the same story: hose washdown. Both are manufacturer ratings; our stainless boxes are rated NEMA 4X / IP66 by the manufacturer and are not UL listed, and neither code says anything about sitting your box at depth.

Worked example: your brass box hangs 20 ft down with four cables entering the bottom.

• Depth pressure: 20 ft × 0.433 psi/ft = 8.7 psi freshwater; 20 × 0.445 = 8.9 psi seawater.
• Proof pressure: 1.5 × 8.7 psi ≈ 13 psi, the factor pressure-vessel practice uses for hydrostatic proof, held 10 min or more.
• Cable side: four jackets at 0.50 in OD, each seated mid-range of a gland sealing 0.35–0.55 in.

What brass should the couplings and glands be?

Dezincification-resistant (DZR) brass for everything you braze or thread on, because corrosion references put dezincification risk above roughly 15 % zinc, and a dezincified fitting leaks as porous copper. Selective zinc loss leaves a sponge of copper that holds no pressure, and it starts where you cannot see it — inside the fillet.

Keep the entry one metal. A brass coupling with a brass gland and a brass plug for unused entries has no galvanic driver; put a zinc-plated steel or aluminium gland on that assembly in seawater and the gland becomes the sacrificial part. Order the couplings with an alloy certificate, and leave surplus fittings with no traceable paper in the drawer.

One more gate before you buy parts: if the box was qualified under a military or navy specification, cutting new entries into it can void that qualification. Get the box builder's filler, flux and heat limits in writing, or buy their brazed connectors and keep the qualification intact.

What do you need before you light the torch?

The job takes a short list, and the test rig belongs on it before the first flame: four certified brass NPT couplings, silver filler that flows at 620–750 °C [1150–1400 °F], flux rated for brass, glands matched to your jackets, and a water bath for the proof test.

How do you braze an NPT coupling into the box wall?

Drill to the coupling's spigot with 0.002–0.005 in [0.05–0.13 mm] of clearance, flux both faces, heat the wall evenly, and feed filler until a fillet closes the full 360 degrees around the joint. Brazing handbooks call that clearance the capillary window where silver filler flows.

1. Strip the shell: gaskets, terminal blocks, potting and any lining come out. You braze a bare box, because the heat destroys everything else.
2. Lay out your four entries on the bottom or a side wall, at least one coupling diameter apart and clear of the lid flange, so no flame licks the gasket face.
3. Machine each hole to the coupling spigot at 0.002–0.005 in [0.05–0.13 mm] diametral clearance.
4. Degrease both parts and brush flux over the whole joint area, inside the hole and on the spigot.
5. Heat the wall with a soft neutral flame — the wall, not the coupling — until the flux turns glass-clear.
6. Touch 1/32 in filler rod to the joint and let capillary action pull it around; keep the joint under about 800 °C [1470 °F], since filler flows from 620 °C and zinc leaves the alloy as you push past 800 °C.
7. Cool in air, then scrub the joint in hot water until no flux film remains; trapped flux corrodes brass under immersion.
8. Chase the NPT threads with a chaser or the gland body, and sight around every fillet for a complete, shiny ring.

When entries sit close together, braze the higher-temperature joint first: reheating a neighbour can drop its fillet while you are watching the new one.

How do you seal four cables into the new threads?

One gland per cable, made up 1.5 to 3 turns past hand tight with sealant rated for continuous immersion, and each jacket OD sitting mid-range of its gland's seal range. NPT tapers 0.75 in per ft of diameter (1 in 16), so the thread wedges and the sealant fills what the wedge misses.

Sizing is a two-minute datasheet job: G3/4 cable gland sizing and thread compatibility shows how the thread and the seal range pair up, and choosing a sealed G-thread to NPT adapter covers mixing thread families. A 0.50 in jacket mid-range of 0.35–0.55 in works; the same jacket at the top of a 0.40–0.50 in range leaves the seal nowhere to follow the cable.

Then close the fourth path. Water-blocked cable or potted conductor ends stop capillary travel through the strands, which is the leak that survives a perfect gland — stopping water wicking through a cable covers the cable-side fixes.

How do you prove the box still holds at depth?

Test the finished box, not the parts: all four glands fitted with their cables, 1.5 times the depth pressure held for 10 minutes, and the whole box under water so you can watch for bubbles. For the 20 ft example that is 13 psi of regulated air; the regulator, not the valve, sets the pressure, and a bubble test at these levels is safe shop practice.

Hydrostatic filling works too when you have the pump: pressurize outside-in to the same 13 psi. Either way, open the box afterwards and read the moisture — the location names the failed seal.

Rework the one entry that failed and repeat the test at full pressure. A joint that holds 5 psi and weeps at 13 psi has fillet porosity, and it will only get worse with time.

What mistakes flood a brazed conversion?

Four failures account for nearly every flooded conversion: flux left on the joint, a fillet that misses part of the ring, a jacket at the edge of its seal range, and open conductor ends. Each has a cheap fix before the box goes under, and an expensive one after.

• Flux residue: hot-water scrub right after cooling, while the residue is soft.
• A partial fillet: reheat and re-flow; a 300-degree ring holds nothing at depth, so the fillet closes the full 360.
• The seal-range edge: move up a gland size so the jacket sits mid-range, where the elastomer can follow it.
• Open strands: pot the ends or order water-blocked cable for the run.
• Overheating: stay under 800 °C and braze the higher-temperature joints first so a neighbour's fillet never re-melts.

Decision: with the builder's process data in hand, braze your four couplings, gland the cables, and prove the box at 1.5 times depth pressure. If the builder withholds the process, or the box's qualification hangs on their connectors, buy their parts instead. And if the duty turns out to be washdown rather than immersion, skip the brass work: lay your glands out in the cutout designer on a 316L stainless body — custom stainless enclosures with factory cutouts are delivered in 10 business days, counted from your approval of the drawing — and the cutouts, glands and windows guide carries the layout rules. Those boxes are rated NEMA 4X / IP66 by the manufacturer and are not UL listed; for real depth you stay with a qualified submersible shell like the brass one you already own.

Enclosures for this job

Catalog sizes that suit the job this article answers, picked from its subject; open one for its full specifications.

Questions