Power Semiconductor Basics · Component Guide

A few months back a replacement shipment of stud diodes came back to us with a note attached: half of them “didn’t work.” They weren’t defective. The customer had pulled reverse-polarity diodes out of a welder rectifier and re-ordered standard ones, because nobody on his team could read the part number off the burnt ones. Three hundred pieces, wrong polarity, everyone annoyed.

That story isn’t unusual, and it’s the reason this guide exists. A stud rectifier diode looks like the simplest component in the cabinet — two terminals, screw it in, tighten the nut. It also has more ways to be quietly wrong than almost anything else in the power bay: wrong polarity, wrong torque, a headline current rating quoted at a case temperature you’ll never see, and a heat-flow path that depends entirely on how carefully somebody prepared an aluminium plate.

What a stud rectifier diode actually is

A stud diode is a single PN junction in a hermetic metal case whose base is a threaded stud. Screw it into a heatsink, and the stud becomes both the mechanical fix and one of the two electrical terminals. That dual role is exactly why polarity and torque matter so much.

The other common shape for the same job is a press-fit (disc/capsule) diode, which gets clamped between two heatsinks, and the diode module, which puts several dies in one insulated package with a copper baseplate. Stud diodes sit in between: cheaper than a module, easier to mount than a press-fit stack, and available individually so you can replace one failed device instead of a whole assembly.

Mounting continuity is also why they’re still popular in repair. A plant that built a plating rectifier in 1998 can usually still buy a DO-5 stud that drops into the same hole.

Package families and roughly what each carries

Thread sizes vary between manufacturers, so never assume — check the outline drawing before you order. The ranges below are what we see most often:

Reading a stud diode datasheet without fooling yourself

Here are real numbers from two representative standard-recovery datasheets, to show why headline values can’t be compared across manufacturers:

Look at the first row. One vendor quotes 80 A at a 140 °C case, the other 150 A at 125 °C. Those numbers are not interchangeable, and if you compare only the “80” and the “150” you’ll mis-specify by a wide margin. Always read IF(AV) together with the case temperature it assumes.

The other trap: IF(RMS). In a rectifier running on AC, heating follows the RMS figure, not the average. For a half-sine at 180° conduction the form factor is 1.571, which means a diode carrying 125 A average is really carrying about 196 A RMS.

Conduction angle: the row almost nobody reads

Most datasheets include a small table showing how much ΔRth(j-c) to add when the device operates at a conduction angle other than DC. Typical figures for a DO-5 part:

Add it to the DC value before you do any thermal math. A device running at 30° conduction is not the same thermal animal as one running continuously — the same average current produces noticeably more heat, because RMS current climbs fast as the conduction window narrows.

stud-rectifier-diode-forward-power-loss-vs-conduction-angle

Polarity: the mistake that costs the most time

In a standard polarity stud diode, the stud is the cathode and the top terminal is the anode. In a reverse polarity part, the stud is the anode — these are normally marked with an R in the type number (think 70HFR120 against 70HF120), and sometimes with a different colour seal on the lead.

Put the wrong one in a bridge rectifier and you don’t get a broken machine. You get a short across the transformer secondary, usually accompanied by the smell everybody remembers.

If you can’t read the marking, a multimeter on diode test settles it in five seconds:

Mounting torque and surface preparation

This is where most stud diode failures are born, and it costs nothing to get right.

Too much torque distorts the case and can stress the die. Too little leaves contact resistance high — and under-torquing is by far the more common error of the two.

Stud diode vs diode module vs press-fit disc

Where stud rectifier diodes actually get used

Worked example: rebuilding the rectifier of a 250 A arc welder

A repair shop called us about a dead single-phase bridge. Here’s how the replacement gets sized.

Circuit: single-phase bridge, four stud diodes. Secondary roughly 70 V open circuit, about 28 V across the arc at 250 A.

Step 1 — per-diode current. Each diode conducts half the time, so average forward current per device is 250 / 2 = 125 A. With the 180° form factor of 1.571, that’s about 196 A RMS.

Step 2 — voltage. Secondary peak is roughly 70 × 1.414 = 99 V. On paper a 200 V part satisfies the 2× rule. In practice nobody keeps 200 V studs on the shelf and the price difference is negligible, so the standard replacement is a 400 V class device. It also buys headroom for the inductive spikes that appear when the arc breaks and re-strikes.

Step 3 — pick the device. A 150 A class stud diode checks out: 150 A average at Tc = 125 °C and 235 A RMS, comfortably above the 196 A we need.

Step 4 — losses. Take VF(TO) = 0.80 V and rf = 0.97 mΩ:

Step 5 — thermal. Add the conduction-angle correction: Rth(j-c) = 0.30 + 0.14 = 0.44 K/W. Contact resistance greased and torqued: 0.10 K/W. Junction-to-sink rise is therefore 137 × 0.54 ≈ 74 K.

Allowing Tj up to 150 °C in a 40 °C workshop, that leaves 150 − 40 − 74 = 36 K for the heatsink, or 36 / 137 ≈ 0.26 K/W per diode. Mount all four on one shared plate and the plate itself only needs about 1 K/W — roughly a 300 × 200 × 10 mm aluminium plate in free air, less again with a fan.

Total dissipation across the bridge lands near 550 W. That sounds alarming until you remember the machine is putting out roughly 7 kW at the arc. About 8% loss, which is what these welders have always done.

How stud diodes fail

Frequently asked questions

What does the “R” mean in a stud diode type number?

Reverse polarity — the stud is the anode instead of the cathode. If you’re replacing parts from a bridge, check every position individually; a single plate often carries both polarities side by side.

Can I fit a higher-current diode than the original?

Almost always yes. Higher IF(AV) is electrically safe, and a slightly different VF rarely matters. Watch three things: the stud thread must match, the polarity must match, and if it sits in a parallel string the VF should be close to its neighbours.

Can I parallel stud rectifier diodes?

Yes, but do it deliberately. Use identical part numbers, mount every device on the same heatsink with symmetric spacing, add small current-sharing resistors or accept derating the string by 10–20%, and never assume two diodes split current evenly just because they’re the same type.

There’s no datasheet for the old diode. What now?

Identify the thread with a gauge to find the package family, determine polarity with a multimeter, measure the actual load current and peak reverse voltage, then pick the next standard current class up and a voltage class at least twice the expected peak. We do this regularly for customers restoring older machines — usually it takes a photo and two measurements.

Why does my diode run hot at only half load?

Nine times out of ten it’s mounting, not silicon. Check for paint or anodising under the flange, missing thermal compound, burrs around the hole, and a nut tightened by feel. Verify with a torque screwdriver before hunting electrical causes.

Standard recovery or fast recovery?

For plain line-frequency rectification — welders, chargers, plating supplies — take standard recovery. It’s cheaper and generally gives better surge capability. Fast recovery only earns its price when there’s a switching stage involved, such as an inverter welder or a chopper.

My old part has a metric thread and replacements come in UNF. Now what?

Don’t force it. Metric studs (M10, M12, M16, M20) and UNF types coexist in the market and are not interchangeable. Measure the pitch, then order the matching family — or ask us, we stock both.

Before you order

Nearly every wrong order we’ve seen came down to one of four things: polarity misread, current compared without checking the case temperature, thread guessed instead of measured, or a mounting surface nobody bothered to prepare. All four are free to fix in advance.