The DO-4 package is a threaded-stud diode outline: a 10-32 UNF-2A stud on one face, a 7/16 in hermetic hex body, and a flexible pigtail lead on the other face. Its correct JEDEC name is DO-203AA. Several reference pages call it DO-213AA, which is wrong. Below are the registered dimensions, the mounting torque limits, measured interface thermal resistance, and the parts you can still buy.
Key takeaways
- The JEDEC designation is DO-203AA. DO-213AA is a MiniMELF surface-mount body and has nothing to do with stud diodes.
- Stud thread is 10-32 UNF-2A per FED-STD-H28. Hex is 7/16 in, 10.77 mm to 11.1 mm across flats. Overall height is 15.24 mm to 20.32 mm.
- Torque the nut to 1.36 N·m to 1.69 N·m (12 lbf·in to 15 lbf·in) dry. Never thread the stud into an untapped hole and never exceed 15 lbf·in.
- A greased bare metal joint measures about 0.2 °C/W. A dry 3 mil mica washer measures 1.6 °C/W, an eight-fold penalty that usually sets your heat sink size.
- Polarity is family-specific. The stud is the cathode on the 1N1199A rectifier series and the anode on the 1N2970B zener series.
- Lead times from Microchip on non-stocked DO-4 parts run 40 to 45 weeks. Treat DO-4 as a repair and legacy package, not a new-design default.
[IMAGE 1: Photograph of a DO-4 stud rectifier next to a DO-5 device on a machined aluminum heat sink, nut and conical washer visible | alt: “DO-4 package stud-mount diode with 10-32 stud and flexible lead, shown beside a larger DO-5 device”]
What the DO-4 package is
DO-4 is a JEDEC diode outline for stud-mounted power rectifiers and Zeners, registered as DO-203AA. The body is a hermetic 7/16 in hex with a 10-32 UNF-2A threaded stud on one face and a flexible lead on the other. The stud carries both current and heat into the heat sink.
JEDEC issued outline numbers sequentially from 1957. In late 1968 it switched to a category system with suffix letters and re-registered the older outlines as time allowed, which is how DO-4 became DO-203AA and DO-5 became DO-203AB. Motorola’s mounting note observes that the original short names never really lost currency among engineers, and distributors still list both: DigiKey’s filter string is “DO-203AA, DO-4, Stud”.
One naming error is worth killing. DO-213AA is MiniMELF, also called SOD-80, a glass surface-mount cylinder roughly 1.6 mm in diameter and 3.5 mm long. The confusion appears traceable to a manufacturer document: Microsemi’s 10 W Zener datasheet T4-LDS-0075 Rev. 3 labels its cover illustration “DO-213AA (DO-4)” while the mechanical section one page later correctly states DO-4 (DO-203AA). If a parametric search on DO-4 returns nothing, search DO-203AA.
DO-4 dimensions
Letter codes below follow the registered outline drawing as published in Microsemi T4-LDS-0075 Rev. 3. Inches are the controlling units; millimeters are conversions.
| Dim | Inches (min–max) | Millimeters (min–max) | Notes |
| C | — to 0.250 | — to 6.35 | Terminal 1 flat surface |
| C1 | 0.012 to 0.065 | 0.30 to 1.65 | Terminal 1 flat surface |
| CD | 0.255 to 0.424 | 6.48 to 10.77 | Cannot exceed HF |
| CH | 0.300 to 0.405 | 7.62 to 10.29 | |
| HF | 0.424 to 0.437 | 10.77 to 11.1 | Hex across flats, nominal 7/16 in |
| HT1 | 0.075 to 0.175 | 1.91 to 4.45 | Chamfer or undercut optional |
| HT2 | 0.060 to 0.175 | 1.52 to 4.45 | Chamfer or undercut optional |
| OAH | 0.600 to 0.800 | 15.24 to 20.32 | Overall height |
| SD | 0.1697 max | 4.31 max | Pitch diameter of plated threads |
| SL | 0.422 to 0.453 | 10.72 to 11.51 | Stud length |
| SU | — to 0.078 | — to 1.98 | |
| UD | 0.163 to 0.189 | 4.14 to 4.80 | Incomplete or undercut thread length |
| ΦT | 0.060 to 0.095 | 1.52 to 2.41 | Terminal diameter |
Three of these bite in practice. SD caps the pitch diameter of the plated thread at 0.1697 in (4.31 mm); heavy tin or solder plating can push a stud past the gauge limit so it binds in a correctly tapped 10-32 UNF-2B hole. Blame the plating before you re-tap the sink. The seating plane has a minimum base diameter of 0.403 in (10.24 mm) because the chamfer at the hex end is optional, so the contact face is not always the full hex. And OAH spans 15.24 mm to 20.32 mm, a 5 mm range, which matters if the device sits under a cover.
Vishay’s outline for the same package agrees: 10-32 UNF-2A stud with an M5 × 0.8 metric option, 20.30 mm maximum height, terminal diameter 1.80 mm ± 0.20 mm. Package mass is roughly 7.0 g for the 12 A rectifier and 7.5 g for the 10 W Zener.
DO-4 versus DO-5
These two outlines get cross-shopped constantly and the datasheets rarely sit side by side. Values below come from the Vishay VS-1N1…A series datasheet (document 93493), the Vishay 35 A to 60 A stud datasheet (document 93492), and Motorola application note AN1040.
| Parameter | DO-4 (DO-203AA) | DO-5 (DO-203AB) |
| Stud thread | 10-32 UNF-2A | 1/4-28 |
| Hex across flats | 7/16 in (10.77–11.1 mm) | 11/16 in |
| Max nut torque, dry threads | 1.69 N·m (15 lbf·in) | 3.4 N·m (30 lbf·in) |
| Max nut torque, lubricated | 1.30 N·m (11.55 lbf·in) | 2.3 N·m (20 lbf·in) |
| Torque on hex, lubricated | 1.43 N·m (12.65 lbf·in) | 3.2 N·m (28 lbf·in) |
| RθJC, representative rectifier | 2.0 °C/W at 12 A | 0.65–1.1 °C/W at 35–60 A |
| RθCS, greased flat surface | 0.5 °C/W | 0.25 °C/W |
| Measured joint, greased metal | 0.2 °C/W | 0.1 °C/W |
| Measured joint, mica + grease | 0.8 °C/W (3 mil mica) | 0.6 °C/W (5 mil mica) |
| Typical stud Zener rating | 10 W | 50 W |
| Approximate mass | 7 g | 17 g |
The short version: DO-5 buys you roughly three times the die-to-case thermal path, twice the torque budget, and five times the Zener power, at two and a half times the mass and a bigger hole in the sink. [INTERNAL LINK: DO-5 stud package -> DO-5 (DO-203AB) package reference]
Polarity: the stud is not always the cathode
This is the most common DO-4 design error, and the package name gives you no help. The R suffix means “reverse of this family’s standard”, not “anode” and not “cathode”.
- Vishay VS-1N1199A through VS-1N3673A rectifiers: base part number is cathode to case. Add R for anode to case, for example 1N1199RA.
- Microsemi 1N2970B through 1N3015B, 10 W Zeners: standard polarity is anode to stud. Suffix RB gives cathode to stud.
- Microsemi 1N3993A through 1N3998A, same package and power: standard polarity is cathode to stud. Suffix RA gives anode to stud.
Two Zener families in the same outline, opposite conventions. In a bridge or any design sharing one grounded sink, picking the wrong variant either shorts a rail through the heat sink or forces you into mica and bushings you had not budgeted thermally. Read the polarity line on the datasheet every time.
Screening prefixes ride in front: JAN, JANTX or JANTXV per MIL-PRF-19500/124 on the Zeners, and MIL-S-19500/260 for the Vishay 12 A rectifiers. An e3 suffix marks the RoHS-compliant commercial build. A full string looks like JANTXV1N2970RBe3.
[IMAGE 2: Labeled cross-section of a stud diode joint showing heat sink, thermal compound, hex seating face, mica washer and insulating bushing, conical washer, and nut | alt: “DO-4 stud diode mounting stack-up with mica insulator, conical washer and nut”]
Thermal design: run the chain before you pick a heat sink
Junction temperature follows the usual series chain, TJ = TA + PD × (RθJC + RθCS + RθSA). Two things surprise people about DO-4.
First, RθJC belongs to the die, not the package. The 12 A rectifier in DO-4 is 2.0 °C/W. The 10 W Zener in the identical outline is 12 °C/W. Same hex, same stud, six times the thermal resistance, because the Zener die is far smaller.
Second, the interface term is large enough to dominate. Motorola measured DO-203AA joints at a 15 in-lb test torque and published this:
| Interface condition | RθCS (°C/W) |
| Bare metal to metal, dry | 0.3 |
| Bare metal to metal, greased | 0.2 |
| 3 mil mica, dry | 1.6 |
| 3 mil mica, greased | 0.8 |
| Vishay datasheet value, smooth flat greased surface | 0.5 |
Worked example. Take a VS-1N1206A, 600 V and 12 A, dissipating 10 W with ambient at 55 °C. The part is rated to 200 °C junction, but design to 150 °C for life. Allowed rise is 95 °C, so total thermal resistance must be at or under 9.5 °C/W. With RθJC of 2.0 and a greased bare joint at 0.5, the heat sink gets 7.0 °C/W. Insert dry mica and RθCS becomes 1.6, leaving only 5.9 °C/W for the sink. Grease that same mica and you recover to 6.7 °C/W. The insulator, not the diode, decided the heat sink size.
Surface preparation is part of the spec. Vishay calls for heat sink flatness under 0.03 mm and surface roughness under 0.02 mm to DIN/ISO 1302, with no particle thicker than 0.05 mm in the joint. AN1040’s looser general guidance for extruded aluminum is 4 mils per inch total indicator reading and a 50 to 60 microinch finish. [INTERNAL LINK: heat sink sizing -> junction temperature and thermal resistance calculation]
Mounting a DO-4 without cracking the die
Vishay and Motorola agree on the failure modes, and they are both mechanical. Excessive torque warps the hex base. So does driving the stud into an unthreaded or partly threaded hole and letting it cut its own path. Either can crack the die, and the resulting part often passes an initial test and dies weeks later.
The only recommended fastening method is a nut and a washer. Working sequence:
- Deburr and clean the sink hole, then verify the tap with a 10-32 UNF-2B gauge. Remove anodizing under the seating face if the stud must conduct.
- Apply a thin, even film of silicone or petroleum-based thermal compound to the hex face and spread it with a spatula. Vishay names Penetrox for metal-to-metal joints.
- Start the stud straight by hand. If it does not turn freely, stop and check the plating and the tap.
- Torque the nut to 1.36 N·m to 1.69 N·m (12 lbf·in to 15 lbf·in) on dry threads, or 1.07 N·m to 1.30 N·m (9.45 lbf·in to 11.55 lbf·in) if the threads are lubricated. Torquing against the device case with lubricated threads is allowed at 1.17 N·m to 1.43 N·m.
- Attach the flexible lead last, with slack and strain relief.
Use a conical (Belleville) washer, not a split lock washer. A number 6 split washer flattens at roughly 50 lb of force where the joint wants 150 lb to 300 lb, and a conical washer holds near-constant pressure across 20 to 80 percent of its deflection, which absorbs thermal cycling. Tighten to about half the washer’s free height.
The flexible lead is a glass-to-metal seal and takes neither bending nor axial load. Route it with slack, use braided or stranded wire, keep the run short and the gauge large, and never let the eyelet support the device. Solder temperature is the one place the two sources disagree: Microsemi caps solder-pad temperature at 260 °C for 10 s, while Vishay’s mounting note gives 260 °C to 275 °C for a lead-free joint to the eyelet. Use the tighter number.
If the stud node cannot share the sink potential, isolate the whole heat sink rather than the device. When that is impossible, DO-4 mica kits exist, including NTE’s DO-4MICA washer and DO-4 mounting kits from surplus specialists. Budget the 0.8 °C/W to 1.6 °C/W penalty from the table above before you commit.
What you can actually buy in DO-4
| Part | Type | Key ratings | Sourcing note |
| VS-1N1199A to VS-1N3673A (Vishay) | Standard recovery rectifier | 50 V to 1000 V, 12 A, IFSM 240 A at 60 Hz, I²t 260 A²s at 10 ms | Widest voltage ladder in the package |
| 1N1343 (Microchip) | Rectifier | 150 V, 16 A | Stocked at DigiKey |
| S21160 (Microchip) | Standard recovery | 1600 V, 22 A, VF 1.2 V at 30 A | $28.98 each at 100, 45-week factory lead time |
| UES705 (Microchip) | Fast recovery, trr 50 ns | 300 V, 20 A | $41.88 each at 100, 40-week lead time |
| UFR3020 (Microchip) | Fast recovery, trr 35 ns | 200 V, 30 A, VF 975 mV at 30 A | $59.25 each at 100, 40-week lead time |
| 1N2970B to 1N3015B (Microsemi) | 10 W Zener | 6.8 V to 200 V, RθJC 12 °C/W, derate 0.083 W/°C above 55 °C | JAN, JANTX, JANTXV to MIL-PRF-19500/124 |
| MBR3545 to MBR35100R | Schottky | 45 V to 100 V, 35 A, IFSM 600 A, VF 0.68 V at 35 A | Highest continuous current commonly offered in DO-4 |
Prices and lead times above are DigiKey listings for Microchip parts. The lead times are the headline: 40 to 45 weeks on anything not already in a bin. Design accordingly, and if you are sustaining equipment, buy the spares now rather than at failure.
Counterfeit exposure is real here. These are 1960s JEDEC numbers, hermetic metal-and-glass, heavily brokered, and often specified as JAN grade for traction, welding, battery-charger and aerospace repairs. Buy franchised where you can. When you cannot, curve-trace forward voltage and reverse leakage against the datasheet before installation, and treat suspiciously crisp marking on an old part number as a reason to test, not a reason to relax. [INTERNAL LINK: counterfeit screening -> detecting counterfeit and remarked semiconductors]
Selecting a DO-4 device
- Check current first. Above roughly 35 A continuous, DO-4 is out of headroom. Move to DO-5 or a modern package.
- Ask whether a stud is genuinely required. For a new design, an isolated TO-247 or a screw-terminal module deletes the insulator stack, the pigtail and the 40-week lead time. DO-4 earns its place when you are repairing installed equipment, need hermetic metal-glass construction, need JAN screening, or need the surge and I²t rating a hermetic stud delivers.
- Pick polarity to match the node your heat sink sits at, then confirm it on the datasheet rather than the package name.
- Compute RθSA with the insulator you will actually use, not the bare-joint number.
- Search distributors on DO-203AA as well as DO-4. Listings are inconsistent and you will miss stock otherwise.
[INTERNAL LINK: TO-247 thermal performance -> TO-247 package mounting and thermal guide]
Frequently asked questions
Is DO-4 the same as DO-203AA?
Yes. DO-203AA is the current JEDEC registration for the outline originally registered as DO-4, part of the renumbering JEDEC began in late 1968. Both names appear on datasheets and distributor filters, frequently as “DO-203AA (DO-4)”. DO-213AA is a different package entirely and is not a DO-4 alias.
What size is the DO-4 stud thread?
10-32 UNF-2A per FED-STD-H28, on a 7/16 in hex body. Stud length runs 10.72 mm to 11.51 mm. Vishay also offers a metric variant with an M5 × 0.8 stud. Maximum pitch diameter after plating is 0.1697 in (4.31 mm), which is the figure to check when a stud will not thread.
How much torque should a DO-4 diode get?
1.36 N·m to 1.69 N·m (12 lbf·in to 15 lbf·in) applied to the nut with dry threads, per Vishay document 93493. Microsemi independently caps the 10-32 nut at 15 in-lb. Lubricated threads drop the window to 1.07 N·m to 1.30 N·m. Never tighten by turning the heat sink or by tapping the stud in.
Is the stud the anode or the cathode?
It depends on the part family, not the package. On the Vishay 1N1199A rectifier series the stud is the cathode and an R suffix reverses it. On the Microsemi 1N2970B Zener series the stud is the anode and an RB suffix reverses it. The suffix always means “opposite of this family’s default”.
Can I mount a DO-4 diode on a PCB?
Only as a chassis or panel-mount part whose stud bolts to a metal surface and whose pigtail is soldered or crimped to the board. The stud is the thermal path; a bare board cannot carry 10 W to 20 W out of a 7/16 in hex. Bolt the device to a sink or chassis, then wire to the PCB with slack.
What replaces DO-4 in new designs?
For most new equipment, an isolated TO-247 or TO-264 rectifier, or a screw-terminal diode module. Those give you built-in isolation, an ordinary footprint and normal availability. Stay with DO-4 when hermeticity, JAN screening, surge rating or drop-in compatibility with installed hardware makes the decision for you. [INTERNAL LINK: DO-41 package -> DO-41 axial rectifier package reference]
What to do next
If you are replacing a failed part, match the JEDEC number and the polarity suffix exactly, order from franchised stock, and torque to 15 lbf·in on a greased, deburred, flat seat. If you are choosing a stud diode for a build that must ship this year, price the DO-4 option against an isolated TO-247 with the 40 to 45 week lead time included, and take the modern package unless hermeticity or military screening rules it out. And whatever you choose, size the heat sink from the insulated interface number, not the bare-joint one. That single substitution is what turns a marginal design into one that survives its thermal cycles.
