The TO-220F package is a TO-220 with the mounting tab moulded inside the plastic, so it bolts to a grounded heatsink with no washer and no bushing. Three vendors publish isolation ratings of 1500 V, 2500 V and 4500 V for that construction — under three different tests. Below: what those numbers actually mean, the dimensions that differ from a standard TO-220, and the torque ceiling that is less than half.
Key takeaways
- Isolation ratings are not comparable across vendors. onsemi rates 4500 VRMS for 0.3 s; Vishay rates 1500 VAC for 1 minute.
- Infineon caps mounting torque at 0.5 N·m for TO-220 Full-PAK against roughly 1.1 N·m for a standard TO-220.
- The mounting hole is smaller: 3.00–3.40 mm on onsemi’s CASE 221AH against 3.54–3.73 mm on Vishay’s TO-220AB.
- Encapsulation costs 1.5 to 2.7 °C/W junction-to-case across three vendor pairs — roughly what a thermal pad costs anyway.
- The package rating covers the tab, not the air path. onsemi states that strike and creepage distance must still be provided.
What the TO-220F package is
TO-220F is a three-lead TO-220 in which the mounting flange is fully encapsulated in the moulding compound, leaving no exposed metal and no electrical connection between tab and die. NXP catalogs it as SOT186 with the industry codes TO-220F and SC-67. Lead pitch stays at 2.54 mm.
The name fragments badly across vendors. ST and most distributors write TO-220FP. onsemi calls the outline TO-220 FULLPAK and registers FULLPAK as a trademark; its case number is 221AH. Infineon writes TO-220 Full-PAK, Vishay writes TO-220 FullPAK, and Vishay’s separate ITO-220AB is a different isolated construction again. The F and FP suffixes both abbreviate “full pack”.
[IMAGE 1: Front and rear views of a TO-220F alongside a standard TO-220AB, showing the encapsulated flange against exposed metal | alt: “TO-220F package with fully encapsulated tab compared to a standard TO-220 with exposed metal flange”]
The isolation rating means nothing without the test condition
| Part and document | Package | Rated isolation | Test condition |
| onsemi MBRF2060CTG, MBRF2060CT/D Rev 9, Jul 2016 | TO-220 FULLPAK | 4500 VRMS | t = 0.3 s, RH ≤ 30%, TA = 25 °C, all leads shorted to heatsink |
| ST STP16NF06FP, CD00002501 Rev 8 | TO-220FP | 2500 V | DC insulation withstand voltage |
| Vishay MBRF20100CT, doc 89033, 11 Jun 2018 | ITO-220AB | 1500 VAC | Terminal to heatsink, t = 1 min |
Three numbers, three tests, no common basis. A 4500 V rating held for 0.3 s is a different claim from 1500 V held for 60 s, and a DC withstand figure is a different claim again from an AC one. Convert nothing. If your safety standard calls for a one-minute AC dielectric test, only the Vishay-style spec answers it directly.
One more limit sits underneath all three. onsemi’s datasheet note states that proper strike and creepage distance must be provided. The rating covers the encapsulation between the leads and the heatsink surface — not the external air path around the package body, and not the clearance to whatever the leads run to next.
TO-220F dimensions: two envelopes under one name
| Dimension | onsemi CASE 221AH, Issue F | NXP SOT186, 10 Feb 2016 | TO-220AB reference, Vishay 95222 |
| Package length, D | 14.70–15.30 mm | 16.4–17.0 mm | 14.85–15.25 mm |
| Package width, E | 9.70–10.30 mm | 9.6–10.2 mm | 10.11–10.51 mm |
| Body thickness / seated height, A | 4.30–4.70 mm | 4.0–4.4 mm | 4.25–4.65 mm |
| Lead pitch, e | 2.54 mm BSC | 2.54 mm nominal | 2.41–2.67 mm |
| Mounting hole | P 3.00–3.40 mm | — | ØP 3.54–3.73 mm |
| Terminations | 3 | 3 | 3 |
Two results matter. The mounting hole shrinks. onsemi’s full-pack hole runs 3.00–3.40 mm against 3.54–3.73 mm on Vishay’s TO-220AB. An M3 screw fits both; anything larger fits only the standard part. A hardware kit carried over from a TO-220AB design will not thread through.
And “TO-220F” is not one envelope. onsemi’s CASE 221AH is dimensionally close to a standard TO-220 at 14.70–15.30 mm. NXP’s SOT186 gives package length as 16.4–17.0 mm, roughly 1.5 mm longer. Both use the symbol D, and NXP labels its parameter “package length” explicitly. Check the specific drawing before you assume a common vertical clearance.
Lead pitch is unchanged at 2.54 mm, so the PCB footprint carries over.
What the encapsulated tab costs thermally
| Part and package | Junction-to-case | Penalty against its own standard version |
| Vishay MBR20100CT, TO-220AB | 2.0 °C/W | — |
| Vishay MBRF20100CT, ITO-220AB | 3.5 °C/W | +1.5 °C/W |
| ST STP16NF06, TO-220 | 3.33 °C/W | — |
| ST STP16NF06FP, TO-220FP | 6.0 °C/W | +2.67 °C/W |
| onsemi MBRF2060CTG, TO-220 FULLPAK | 4.0 °C/W | No standard-package counterpart in the same datasheet |
Across two true same-die pairs the encapsulation costs between 1.5 and 2.67 °C/W. That is the number worth holding onto, because a thermal pad costs roughly the same. The isolation you delete from the assembly is the isolation you were going to add anyway.
The comparison flips only when you can afford grease and a bare tab — an ungrounded or floating heatsink. There the standard package keeps its full advantage.
[IMAGE 2: Thermal path diagram comparing a TO-220F on bare greased heatsink against a standard TO-220 with a thermal pad | alt: “TO-220F package thermal path compared to a standard TO-220 with an insulating thermal pad”]
Mounting: half the torque of a standard TO-220
Infineon’s application note AN2006-12 (rev 2.1, 12 September 2022) sets the constraint plainly: recommended mounting torque is 0.5 N·m and must not be exceeded. Vishay specifies 10 in·lb maximum, about 1.13 N·m, for TO-220AB and ITO-220AB in doc 89033. The full-pack ceiling is under half.
The reason is structural. On a standard TO-220 the screw bears on metal. On a full-pack it bears on the moulding compound, and that compound is the isolation. Crush it and the dielectric goes with it.
Infineon’s other requirements are equally specific:
- A rectangular washer between screw head and tab, sized so it does not damage the plastic body.
- No self-tapping screws.
- Clip mounting preferred, because the force lands above the die. Saddle clips for heatsinks under 5 mm thick, U-clips above that, both delivering 15 N to 50 N. Groove-anchored proprietary clips reach 25 N to 50 N.
- Heatsink surface roughness RZ ≤ 10 µm and flatness ≤ 10 µm over a 15 mm reference length.
- Thermal grease still helps, reducing interface resistance by 1.2 to 1.5 K/W.
onsemi shows clip mounting as the typical technique for its FULLPAK parts, which is consistent with the torque limit.
Worked example: sizing a heatsink for a full-pack rectifier
Take onsemi’s MBRF2060CTG, a dual 20 A / 60 V Schottky in TO-220 FULLPAK. Per leg, RθJC is 4.0 °C/W and forward voltage is 0.75 V at 10 A with TC = 125 °C.
Dissipation per leg 10 A × 0.75 V = 7.5 W (both legs: 15 W into one case)
Junction-to-case rise 7.5 W × 4.0 °C/W = 30 °C
Junction derated 150 °C → case must stay ≤ 120 °C
Ambient 45 °C (120 − 45) / 15 = 5.0 °C/W case-to-ambient allowed
Greased contact 0.5 heatsink must be ≤ 4.5 °C/W
No washer, no bushing, no isolator in that stack. Run the same budget on a standard TO-220 with a thermal pad and you land in the same region, which is the point of the previous section.
[IMAGE 3: Clip-mounted TO-220F on an extruded heatsink with a saddle clip, annotated with contact force | alt: “TO-220F package clip mounted to an extruded heatsink showing saddle clip contact”]
Five mistakes that cause isolation failures
- Torquing a full-pack to the standard TO-220 spec. The ceiling is 0.5 N·m, not 1.1 N·m, and the compound you crush is the dielectric.
- Comparing isolation ratings across datasheets. 4500 V for 0.3 s and 1500 V for 60 s are not on the same scale.
- Assuming the package rating covers creepage. onsemi requires strike and creepage distance to be provided externally.
- Reusing a TO-220AB hardware kit. The full-pack hole is 3.00–3.40 mm.
- Using a self-tapping screw into a tapped sink. Infineon rules it out; the swarf and the uncontrolled torque both land on the plastic.
Frequently asked questions
What does the F in TO-220F mean?
Full pack. The mounting flange is encapsulated in the moulding compound rather than exposed, so the tab carries no electrical connection to the die. ST and most distributors write the same package as TO-220FP; onsemi calls its version TO-220 FULLPAK; NXP catalogs it as SOT186 with the industry code SC-67.
Is TO-220F the same as TO-220FP?
In practice yes — both abbreviate full pack and describe a three-lead TO-220 with an encapsulated tab. The dimensions are not automatically identical across vendors, though. onsemi’s CASE 221AH gives package length as 14.70–15.30 mm while NXP’s SOT186 gives 16.4–17.0 mm. Check the drawing tied to your order code.
What is the isolation voltage of a TO-220F package?
There is no single figure. onsemi specifies 4500 VRMS for 0.3 s at 25 °C and 30% relative humidity; ST specifies 2500 V DC withstand; Vishay specifies 1500 VAC for one minute on its ITO-220AB. Match the rating to the test your safety standard demands rather than picking the largest number.
Can I mount a TO-220F directly to a grounded heatsink?
Yes — that is the reason the package exists. No mica washer, no shoulder bushing, no isolating pad. You still need thermal grease for contact, a torque limit of 0.5 N·m per Infineon, and external creepage and strike distance that the package rating does not cover.
What torque should I use on a TO-220F?
Infineon’s AN2006-12 recommends 0.5 N·m and states it must not be exceeded. Use a rectangular washer under the screw head and no self-tapping screws. Clip mounting is preferable because the load lands over the die; saddle and U-clips deliver 15 N to 50 N of contact force.
What to do next
If your heatsink is grounded and the thermal budget has 2 °C/W of slack: specify the full-pack. It deletes the washer, the bushing and the assembly step that most often shorts a tab to chassis, at a thermal cost close to the pad it replaces.
If your heatsink floats: keep the standard TO-220 and grease it. Encapsulation buys nothing you need and costs 1.5 to 2.7 °C/W.
Before you accept an isolation rating: read the test condition beside it. Duration, waveform and humidity all move the number, and none of the three published figures converts into the others.
Before releasing the mounting drawing: set the torque at 0.5 N·m, specify a rectangular washer, and check the hole against 3.00 mm worst case rather than a TO-220AB kit.
If you can use a clip: use one. It puts the force over the die and removes the torque question entirely.
