A TO-126 package body is nominally 11.0 mm tall, 8.0 mm wide and 3.25 mm thick, with a 3.2 mm mounting hole and 2.28 mm lead pitch. That pitch is the number most footprint libraries get wrong, because it is not 2.54 mm. JEDEC registers the same outline as TO-225AA and STMicroelectronics calls it SOT-32. Below are the toleranced dimensions from three manufacturer drawings, the torque and interface resistance that decide whether the part survives, and the land-pattern math.

Key takeaways

• Three vendor drawings disagree on body width and thickness by enough to matter. Design the keepout for the widest, not the nominal.
• Lead pitch is 2.28 mm, roughly 0.090 in. A 0.1 in grid footprint will fight you.
• The exposed pad is electrically live. On a BJT it is the collector.
• Rated dissipation ranges from 12.5 W to 40 W across parts in the same outline. The package does not set the number.
• Greasing the interface roughly halves case-to-sink resistance. Adding mica roughly triples it back.
• Use a conical compression washer. A split lock washer cannot generate the needed clamping force.

[IMAGE 1: TO-126 transistor bolted to an extruded aluminum heatsink, screw head and conical washer visible | alt: “TO-126 package mounted on an aluminum heatsink showing the exposed metal tab”]

TO-126 dimensions, drawing by drawing

The TO-126 outline is old enough that vendors registered slightly different versions of it. The table reconciles three current or archived drawings. Where they conflict, design to the widest envelope.

Feature	onsemi Case 77-09 Issue AA (mm)	Fairchild pkg code AN, Rev B (mm)	Toshiba TO-126N (mm)
Body height	10.60 to 11.10	11.00 ±0.20	11
Body width	7.40 to 7.80	8.00 ±0.30	8.0
Body thickness	2.40 to 3.00	3.25 ±0.20	3.25
Mounting hole diameter	2.90 to 3.30	3.20 ±0.10	not published
Hole center from top edge	3.80 to 4.20	3.90 ±0.10	not published
Lead pitch	2.04 to 2.54	2.28 typ	not published
Lead width	0.60 to 0.90	0.75 ±0.10	not published
Lead thickness	0.39 to 0.63	0.50 +0.10 / −0.05	not published
Overall length with leads	14.50 to 16.63	16.10 ±0.20	not published

onsemi’s Case 77-09 table caps body width at 7.80 mm and body thickness at 3.00 mm. Fairchild’s drawing gives 8.00 ±0.30 mm and 3.25 ±0.20 mm for the same features. Those thickness ranges do not overlap. A mechanical clearance sized to onsemi’s 3.00 mm maximum will foul a Fairchild-lineage part at 3.45 mm.

Toshiba publishes its TO-126N as 8.0 × 11 × 3.25 mm under internal code 2-8U1A, which sits with the Fairchild envelope. Treat 8.3 mm wide by 3.45 mm thick as the safe worst case unless your BOM is single-sourced.

[IMAGE 2: annotated TO-126 outline drawing with body height, width, thickness, hole diameter and lead pitch called out | alt: “TO-126 package dimension drawing showing 11 mm body height, 8 mm width and 3.2 mm mounting hole”]

The names TO-126 hides behind

The descriptor comes from Transistor Outline Package, Case Style 126, and the revised JEDEC outline system renumbers it TO-225AA. STMicroelectronics refers to the same style as SOT-32. IEC 60191 calls it A56, DIN 41869 calls it 12A3, and GOST 18472-88 calls it KT-27. onsemi’s internal case number is 77, and the company brands the construction Thermopad. Toshiba adds TO-126N. All seven names describe parts that will drop into the same footprint.

Pin order depends on the device, not the package

Central Semiconductor’s package sheet lists three pin styles for the same outline: transistors are emitter, collector, base; SCRs are cathode, anode, gate; triacs are MT1, MT2, gate. In every case the mounting plate is common with pin 2. The die is bonded to a metal sheet whose opposite face is exposed for heatsinking, which is why the tab carries the collector potential on a bipolar transistor. Never assume the pinout from the package.

What a TO-126 can actually dissipate

Treating TO-126 as a wattage class is the most common design error here. Rated dissipation across parts sharing this outline spans better than three to one.

Part	Type	V(BR)CEO	I C	P TOT at T C = 25 °C	Rθ(j-c) max	Rθ(j-amb) max
BD139 (ST)	NPN BJT	80 V	1.5 A	12.5 W	10 °C/W	100 °C/W
BD140 (ST)	PNP BJT	−80 V	−1.5 A	12.5 W	10 °C/W	100 °C/W
MJE340 (onsemi)	NPN BJT	300 V	0.5 A	20 W	6.25 °C/W	not specified
BD679 (ST)	NPN Darlington	80 V	4 A	40 W	3.12 °C/W	100 °C/W
BD681 (ST)	NPN Darlington	100 V	4 A	40 W	3.12 °C/W	100 °C/W

ST rates the BD139 at 12.5 W with the case at 25 °C but only 1.25 W with ambient at 25 °C. onsemi derates the MJE340 at 0.16 W/°C above 25 °C, the reciprocal of its 6.25 °C/W junction-to-case figure.

Note that junction-to-ambient stays near 100 °C/W regardless. In free air every one of these parts is a one-watt device. The differences only appear once a heatsink exists.

That creates a sourcing trap. DigiKey’s parametric field for BD13916STU reads 1.25 W, while the same field for 2N4922G in the same package reads 30 W. One is referenced to ambient, one to case. Filtering a catalog on power mixes the two silently.

How to mount a TO-126

onsemi’s AN1040 classifies the TO-126 as a plastic body mount part, meaning the clamping force lands on plastic rather than on a metal flange. The recommended fasteners are machine screws, eyelets, or clips, always with a conical compression washer. The package hole accepts an M3 or a #4 screw.

The stack, from the sink upward: heatsink, thermal compound or pad, insulator if required, package, conical washer, flat washer, screw. Screw heads must not bear directly on the plastic, because they are not flat enough to distribute force and the case can crack.

Two surface rules govern the sink. Flatness better than 4 mils per inch is adequate in most cases, and a 50 to 60 microinch finish is sufficient; finer costs money without lowering contact resistance. The heatsink hole should clear the fastener and no more, so for plastic packages choose a smaller screw rather than a bigger hole.

Now the numbers that decide the design. onsemi measured interface resistance for both plastic packages at their respective test torques.

Package	Test torque	Metal to metal, dry	Metal to metal, greased	With 2 mil mica, dry	With 2 mil mica, greased
TO-126 (contact pad 1/4 in × 3/8 in)	6 in-lb (0.68 N·m)	2.0 °C/W	1.3 °C/W	4.3 °C/W	3.3 °C/W
TO-220AB	8 in-lb (0.90 N·m)	1.2 °C/W	1.0 °C/W	3.4 °C/W	1.6 °C/W

These values come from onsemi’s applications laboratory, with the caveat that dry-interface figures vary widely because they depend heavily on surface condition. The TO-126 penalty against TO-220 is structural: its contact pad is only about 1/4 in by 3/8 in, roughly 60 mm², against a body face of about 88 mm².

Clips beat screws here. A clip presses on the plastic directly over the die, while a single screw at one end lifts the package under the die.

[IMAGE 3: exploded view of the mounting stack, heatsink through screw | alt: “Exploded view of TO-126 mounting hardware stack with conical washer, insulator and M3 screw”]

Isolating the tab without losing the heat path

If the sink is grounded and the collector is not, an insulator goes in the path. Mica plus grease costs about 2 °C/W over a greased bare joint, per the table above.

Silicone elastomer pads trade some of that back for assembly speed. Fischer Elektronik’s WK 126, cut for the TO126 outline at 11 × 9 mm, is published at 0.45 K/W with 6.5 kV isolation. Vendor pad figures and system figures are measured differently, so do not subtract one from the other. AN1040 states plainly that interface data from different manufacturers is not in good agreement.

Conformal pads also relax. In one 1000-hour study of a CHO-THERM 1688 pad, interface impedance fell from 0.90 to 0.70 °C/W while the mounting torque decayed from 6 in-lb to 3 in-lb. With a non-conformal material that torque loss would have raised resistance instead. Conical washers exist to absorb exactly this.

For screw isolation, add an insulating bushing inside the package hole so the threads cannot reach the metal base.

Worked example: a BD139 at 3 W

Assume a BD139 dissipating 3 W in a 55 °C enclosure, with the junction held to 125 °C for margin against the 150 °C limit.

Total allowable resistance is (125 − 55) / 3 = 23.3 °C/W. Junction-to-case takes 10 °C/W. A greased, non-isolated joint takes 1.3 °C/W. The heatsink must therefore be 12.0 °C/W or better. Insert mica and grease instead and the interface takes 3.3 °C/W, leaving 10.0 °C/W for the sink.

For scale: a Fischer SK 129 63,5 STS extrusion, rated 4.5 °C/W and listed for SOT-32 among other packages, clears this with room to spare. Aavid’s 5775 slip-on clip for TO-126, part numbers 577500B00000G and 577500U00000G, is 0.63 mm sheet aluminum plotted on a 0 to 20 °C/W scale and does not.

Without any heatsink the same part sees 100 °C/W, so 55 °C ambient and a 125 °C junction limit allows only 0.70 W. The 1.25 W free-air figure on the datasheet assumes 25 °C ambient and a junction at the 150 °C limit.

PCB footprint: hole and pad sizing

Start from the largest lead cross-section. Case 77-09 permits lead width up to 0.90 mm and thickness up to 0.63 mm, giving a diagonal of 1.10 mm.

IPC-2222 sets minimum hole size at maximum lead diameter plus 0.25 mm at Level A, 0.20 mm at Level B and 0.15 mm at Level C. Pad diameter is that hole size plus twice the 0.05 mm annular ring plus the fabrication allowance, 0.6 mm at Level A and 0.5 mm at Level B.

Producibility level	Minimum finished hole	Pad diameter	Copper gap at 2.28 mm pitch
Level A	1.35 mm	2.05 mm	0.23 mm
Level B	1.30 mm	1.90 mm	0.38 mm

Level A round pads leave a 0.23 mm gap, marginal for many fabricators’ solder-mask registration. Either accept Level B geometry or use oblong pads with the long axis perpendicular to the lead row. This is the practical consequence of the 2.28 mm pitch.

Six mistakes that put TO-126 parts in the returns bin

1. Heatsink hole larger than the package hole. Tightening then pulls the package into the hole, and the resulting distortion can crack the die.
2. Punched sink holes. A punched hole depresses the surrounding metal; the device either deforms to the crater or bridges it, losing contact area.
3. Split lock washers. A typical #6 split washer flattens at about 50 pounds, while 150 to 300 pounds is needed for good heat transfer.
4. Bending leads at the body. AN1040 advises a bend radius greater than 1/16 in for TO-225AA, no twisting at the case, and no axial motion of the lead relative to the case. Axial force above 4 pounds may cause permanent damage.
5. Soldering too hot or too close. The limit is 260 °C for no more than 5 seconds, applied further than 1/8 in from the plastic case.
6. Aggressive cleaning solvents. Alcohol and unchlorinated solvents are generally fine; gasoline and chlorinated freon can swell the encapsulant and damage the die.

Sourcing, lifecycle, and compliance

Moisture sensitivity does not apply. J-STD-020 covers nonhermetic solid-state surface mount devices, and defines SMD as plastic encapsulated surface mount packages and other moisture-permeable packages. A through-hole TO-126 falls outside that scope, so expect no MSL rating. The datasheet’s lead soldering limit governs instead.

Automotive. Among onsemi’s TO-225 bipolar parts, part numbers carrying an S or NSV prefix are qualified to AEC-Q101. The base part number is not.

Lifecycle is the real risk. DigiKey lists KSA1381ESTU, a 300 V PNP in TO-126-3, as obsolete. 2N4923G carries a 27-week lead time at $1.31 in ones and $0.38748 at 1,000, while 2N4922G shows 16 weeks. Same package, same vendor, eleven weeks apart.

Second sourcing is easier than it looks. onsemi states that BD677, BD677A, BD679 and BD679A are equivalent to MJE800, MJE801, MJE802 and MJE803, and the BD139 has a direct PNP complement in the BD140. Qualify on junction-to-case resistance, not on the package name.

Frequently asked questions

What is a TO-126 package?

TO-126 is a three-lead, plastic-molded through-hole package for medium-power discretes such as transistors, thyristors and triacs. The die sits on a metal sheet whose exposed face forms a heatsinking pad, and the body carries a hole for a screw or clip. Body size is roughly 11 mm by 8 mm by 3.25 mm.

Is TO-126 the same as TO-225AA?

Yes. TO-126 is the original JEDEC descriptor, and the revised outline system renumbers it TO-225AA. Datasheets use the two interchangeably, sometimes on the same page. onsemi additionally labels it Case 77 and ST labels it SOT-32. All refer to one footprint.

What is the lead spacing on a TO-126?

2.28 mm typical, about 0.090 in. Fairchild’s drawing gives 2.28 mm typ and onsemi permits 2.04 to 2.54 mm. It is not the 2.54 mm pitch used by TO-220, and footprint libraries that assume 0.1 in spacing will misalign the outer two leads by roughly a quarter millimeter each.

What size screw fits a TO-126 mounting hole?

The hole is 2.90 to 3.30 mm across, so M3 is the natural fit and #4 also clears. Add a conical compression washer under the head, and keep the heatsink hole no larger than the package hole.

How much power can a TO-126 dissipate without a heatsink?

About 1.25 W at 25 °C ambient, set by a junction-to-ambient resistance near 100 °C/W. At 55 °C ambient with a 125 °C junction target the figure drops to roughly 0.7 W. Case-referenced ratings of 12.5 W to 40 W require a real heatsink and a real interface.

Can I substitute a TO-220 part for a TO-126?

Mechanically no. The pitch differs, the body is larger, and the mounting hole sits differently. Thermally a TO-220 is usually the better choice: its greased metal-to-metal interface measures 1.0 °C/W against 1.3 °C/W for TO-126, and its contact area is larger.

What to do next

Under about 1 W in still air, mount the part by its leads and skip the hardware. Between 1 W and roughly 4 W, use a clip-on sink and a greased bare joint, verified with the arithmetic above. Above that, or wherever the sink is grounded, budget the isolator at 3.3 °C/W and check the heatsink number falls out positive before you commit the layout.

Draw the footprint at 2.28 mm pitch with a 1.30 mm finished hole and 1.90 mm pads. Then pull the drawing for the exact part number in your BOM and confirm the body envelope, because the vendors do not agree.

Internal links proposed

• [INTERNAL LINK: TO-220 package dimensions and mounting → TO-220 reference page]
• [INTERNAL LINK: calculating junction temperature from thermal resistance → thermal design fundamentals]
• [INTERNAL LINK: through-hole land pattern rules in IPC-2221 → PCB footprint design guide]
• [INTERNAL LINK: choosing thermal interface materials → TIM selection guide]
• [INTERNAL LINK: reading a transistor datasheet’s absolute maximum ratings → datasheet interpretation]

External sources

• onsemi AN1040/D Rev 5, Mounting Considerations For Power Semiconductors: https://www.onsemi.cn/download/application-notes/pdf/an1040-d.pdf
• onsemi MJE340/D Rev 12, Case 77-09 Issue AA: https://www.farnell.com/datasheets/1747980.pdf
• STMicroelectronics BD135/BD136/BD139/BD140 Rev 5: https://www.st.com/resource/en/datasheet/bd139.pdf
• STMicroelectronics BD677 family Rev 5: https://www.st.com/resource/en/datasheet/bd677.pdf
• Toshiba TO-126N package data: https://toshiba.semicon-storage.com/ap-en/semiconductor/design-development/package/detail.TO-126N.html
• Fairchild TO-126 package drawing Rev B: http://www.icminer.com/pakage/www.fairchildsemi.com/to126.pdf