How a data centre PDU bracket is manufactured

A rack PDU bracket is the cheapest part in the cabinet and the one most likely to be rejected on site. It is a 2 mm steel cantilever, usually two of them per PDU, whose only job is to hold 4–12 kg of sockets and cable at a fixed position in a 482.6 mm rack while somebody plugs a C13 lead into it at 02:00. When it fails, it sags, the PDU drops out of the U space, or the earth bond reads open.

Data centre PDU bracket manufacturing is therefore a study in three numbers: the hole-to-bend distance that stops the ear tearing, the deflection that keeps the PDU level under load, and the film thickness that decides whether the bracket is earthed or insulated.

This walk-through covers a 1U horizontal PDU bracket and its vertical 0U variant, 2.0 mm CRCA, powder coated, from the laser blank to the continuity check, with the geometry, load and coating limits a drawing needs.

What the bracket has to line up with

A rack PDU does not define its own mounting; the rack does. IEC 60297-3-100 fixes the 482.6 mm (19 in) series: the panel width across the mounting flanges, the 465.1 mm hole centres, the 44.45 mm unit pitch and the mounting hole pattern. A bracket that is 1 mm out on the hole centres will bolt to one rack and not the next, and every rack brand in a modern data hall draws from the same series.

The practical consequences for the drawing:

• 465.1 ±0.3 mm between the two mounting hole centres, measured across the ear centres.
• 44.45 mm U pitch, so a 1U bracket occupies one U and the holes sit 22.2 mm from the U boundary on a half-pitch pattern.
• 9.5 mm square for M6 cage nuts, or Ø6.5 mm clearance for a direct screw. The square is the more demanding feature because a cage nut needs all four corners.
• Ear depth of 40–60 mm from the rack post to the PDU body; deeper ears put the PDU into the cable bend radius.

Vertical 0U PDUs are different. They hang on two L-brackets, one at the top and one at the bottom of a 1,800–2,000 mm rear post, and the brackets carry the weight plus the pull of 20–30 cables. The mounting position is set by the PDU length, not by the U pitch, and the brackets are usually handed — a left and a right that must not be swapped in the kitting tray.

PDU bracket manufacturing: blanking and forming the ear

At low and medium volume the blank comes off the 12 kW fibre laser on its 40 ft × 16 ft bed, nested with the cage-nut squares, the ear holes and the cable-tie slots in one pass. Above roughly 5,000 brackets a year, a progressive die on a 200–300 T press is cheaper and, more importantly, holds the hole pattern in the tool rather than in the nesting file. A 250 T punching press with a standard 9.5 mm square station can also pierce the cage-nut pattern in a second operation without a dedicated tool.

Forming is two 90° bends and, on the 0U variant, a third offset bend that clears the rear door. Three details decide the outcome.

Hole-to-bend distance. The rule is 2.5t + R from the bend line to the nearest edge of the hole. On 2 mm CRCA with a 2 mm inside radius that is 7 mm. Put a 6.5 mm hole 4 mm from the bend line and the hole stretches into an oval, the material thins at the outer fibre, and a crack starts. The hole-to-bend distance article works through the case where the customer cannot move the hole and the bend has to be relocated instead.

Minimum flange for the die. On the 200 T brake with a 4,450 mm bed, a 1.5 mm flange on 2 mm sheet needs a 6 mm V-die; the standard 16 mm V-die will not grip it. Small brackets are formed on smaller tooling, which is why a 40 mm ear is comfortable and a 12 mm ear is not.

Springback. 2 mm CRCA at 90° springs back 1–2°, and the 0U offset bend returns more. The brake compensates through the Delem control, but the offset dimension is what drifts, not the angle, and a 0.5 mm drift on the offset puts the bracket 0.5 mm off the post.

The bend allowance you need in the blank

A 90° bend on 2 mm CRCA with a 2 mm inside radius gives back roughly 4.2 mm of bend allowance, so the developed length is shorter than the sum of the outside legs by about 4.2 mm per bend. Two bends, and a blank cut from the outside dimensions comes out 8.4 mm long, which throws the hole centres out by more than the tolerance. Run the numbers through the bend allowance calculator before the DXF is released; correcting a blank after the first article costs a tool revision or a re-cut batch.

The 0U offset and the door swing

Vertical PDUs sit behind the rear door and behind 20–30 power cables. The bracket's offset bend is what buys that space, and it is the dimension that fails first on a new rack. Allow 40–50 mm from the rear post to the PDU body for a 10 mm cable bundle bending through 90°: a standard IEC C13 lead is 7–8 mm in diameter and needs a bend radius of at least four times that, 30 mm, or the conductors work-harden over time.

The offset itself is a Z-bend with two opposed 90° bends, each springing back 1–2° and each carrying its own bend allowance. An offset that closes by 0.5 mm per bracket puts the PDU back panel 2 mm closer to the door frame than the drawing allows, and the site team finds it when the door will not latch. State the offset as a ±0.5 mm dimension from the post face, not from the bend line.

The limits to put on the drawing

The deflection column deserves attention because it is the one number a design engineer can change for free. Doubling the ear depth from 40 mm to 60 mm raises the second moment of area by 50% and cuts deflection by a third; adding 0.5 mm of thickness roughly halves it. Neither costs anything at the press. Adding a fourth screw to the rack post does.

Coating, masking and the earth path

PDU brackets are almost always powder coated, usually in the same RAL as the rack. Two things go wrong and both are electrical.

Film on the mounting face. A 70 µm coat on the ear face means the bracket no longer sits flush on the post; it stands 0.07 mm proud and, because the screws compress the film unevenly, the bracket can rock. The PDU then sits out of plane by 1–2 mm over 482.6 mm, which reads as a crooked PDU. Mask the mounting face to 20 µm maximum, and specify it as a masked zone.

Film on the bond path. Polyester powder has a dielectric strength of roughly 30–50 kV/mm, so a 70 µm film holds off 2–4 kV. A bond tester pushing a low current sees an open circuit, and the installer has to remove coating on site, which is where the scratch and the rework start. Mask a 25 mm pad around the earthing stud, or fit a stud that stands proud of the film and use an external-tooth washer under the nut.

There is a third, more subtle failure. If the bracket is the only metal path between the PDU chassis and the rack frame, the PDU's protective earth runs through two powder-coated joints. Even with the pads masked, the path has to be verified, not assumed: a continuity check at 0.1 Ω across the assembled joint is a two-minute test on the first article and a five-minute argument avoided at commissioning.

Zinc whiskers and the wrong finish

Pre-galvanised sheet is cheaper than coated CRCA and needs no paint. Electroplated zinc, however, can grow whiskers — single crystals a few micrometres across and up to tens of micrometres long — under compressive stress, and a data hall is the one place where a conductive whisker across two terminals is a known failure mode. Powder-coated CRCA, or a zinc-nickel plated bracket where conductivity is needed, avoids the question. If the drawing says galvanised, ask whether the finish is hot-dip (no whisker risk, but 60–100 µm of zinc that changes the hole sizes) or electroplated (whisker risk).

Inspection before the bracket leaves

The acceptance checks that catch field failures are short.

• Hole pattern on a checking fixture or with a calliper across the two outer holes. 465.1 ±0.3 mm, and the 9.5 mm squares checked with a go/no-go pin after coating, not before.
• Ear angle and offset against a square, with the offset dimension measured from the post face rather than from the bend.
• Load check on the first article: hang 2× the rated PDU mass for 24 h and measure deflection at the centre. A bracket that has taken a set inside 24 h is under-designed, not under-manufactured.
• Continuity across the masked joint at 0.1 Ω.
• Film thickness on the face and the bond pad, with the pad reading reported separately because it is the one place a high reading is a failure.

What Ganesh actually runs

Brackets are lasered on the 12 kW machine or punched on the 20–300 T press line, formed on the 200 T 5-axis brake with a 4,450 mm bed, and powder coated through seven-tank pretreatment and the 18 ft × 14 ft booth. Checking fixtures and weld fixtures are built in the in-house tool room, which is also where a progressive die for a high-volume bracket would be made, and coated parts are checked for film thickness and cross-hatch adhesion before dispatch. The plants work to ISO 9001:2015 and ISO 14001:2015, and first article inspection and PPAP documentation are available on request.

The data centre racks product page covers the frames, rails and blanking plates on the same route, and bending for rack frames and cabinets explains how the bend sequence changes when brackets and rails are formed from the same coil.

What the rack drawing has to settle

1. Rack standard and the hole pattern, with the 465.1 mm centres toleranced.
2. PDU mass, mounting position and any cable pull the bracket must resist.
3. Material, thickness and finish, with pre-galvanised accepted or rejected.
4. Bend radius, ear depth and the hole-to-bend distance as a minimum note.
5. Masked zones: mounting face, bond pad, anything that must stay conductive.
6. Cage-nut squares, and whether they are pierced before or after coating.
7. Load and deflection acceptance, if the bracket carries a vertical PDU.
8. Continuity limit and the test method.

Most PDU bracket rejections come from items 1, 5 and 6 being left open. Send the drawing with the PDU data sheet to request a quote, and the rack load calculator will give the load figures to check the ear against. The data centre hub indexes the related guides on containment panels, blanking plates and rails.

When a bracket is the wrong answer

Three cases where the part should not be made at all.

The PDU is too heavy for a rear-post bracket. Above about 15 kg, a vertical PDU needs a full-length mounting rail or a shelf, not two L-brackets. Bolting a heavier unit to the same bracket moves the failure from deflection to fatigue at the bend.

The rack has no rear posts. Some open-frame racks only have front mounting angles. A bracket designed for a rear post will not fit, and the adaptor that makes it fit usually puts the PDU in the door swing.

The bracket has to be aluminium and anodised. A 2 mm aluminium bracket saves about 60% of the mass, which matters on a swing-out frame, but anodising is not a process we run. If the design depends on a hard anodic finish for wear at the ear, that part comes from an anodiser, not from us.

For everything else the part is simple, and that is exactly why it gets under-specified. The three numbers — 7 mm hole-to-bend, 2 mm deflection and 20 µm on the bond pad — cover most of what goes wrong.