question about transistor package types

For the various medium to high power transistor package styles that have some sort of metal heatsink mounting tab (to-220, to-126, to-246, ect), is that tab electrically isolated from the actual transistor?

For instance, I know sometimes people stick to-92 types together with thermal grease to make them track thermally. Can one do a similar thing with the above mentioned types, and similar ones? Not that I specifically want or need to do this. I just want to know if it works that way.

Something I think that I would want to do is mount my heat sinks to the chassis. Can one do this, and then bolt the tabs of these transistors straight to the heat sink? Or do you need some sort of insulator between the tab and the sink.

Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
Usually the transistor's metal case is connected to the collector; often this fact is mentioned on the transistor's datasheet. It's also possible to use an ohmmeter to find out what the case is connected to (if anything).

There are also some all-plastic versions of the TO126 and TO220 packages, which have no metal case. Check the datasheets of the BD139 and BUJ302 for examples.
ok, how is this isolation accomplished? I assume any sort of plastic or other insulator between the tab and the sink would be highly detrimental to proper heat transfer. Besides, in the types with no metal bushing, keeping the mounting bolt isolated seems problematic.

Does this pretty much mean that the heatsink itself needs to be isolated from the rest of the circuit, ie, mounting direct to the chassis is a no-go?

I presume that when I see photos of lines of power transistors bolted to the same sink, they are all in parallel, and the collectors are fine sharing this connection.
There are insulating washers for the transistor body and plastic bushes for the bolts.

Once the transistors are insulated from the heatsink you are free to bolt the heatsink to an earthed chassis.

I built up an amp last week and fell foul of the classic problems.
1/ I forgot to deburr the holes and they broke through the insulating tab when I bolted them up.
2/ I got some swarf behind am insulating tab and that broke through the insulating tab.

A case of more haste less speed.

I always test now with a DMM that there is no short between transistor tab and the heatsink before I power up for the first time.

Here is an example of a few insulating washers.


2003-02-17 7:38 pm
it is possible for hard anodize on a Al heatsink to be the insulating layer - but explicit insulation is safer for hobbyist

for driver Q using PCB mount individual heatsink - you can just let the heatsink float with the Q collector/drain V - although it can be a shock hazard on probing/while debugging
Last edited:
Thermal compound and mica insulators are still very common. There are also sil-pad type insulators that don't require the thermal compound. Generally if there is an exposed metal surface on the transistor body then electrical isolation is required, unless the circuit is designed such that the tab is at ground potential or the heat sink itself is isolated from chassis.

Ideally, the isolation material has a very high dielectric strength but is highly conductive thermally. There aren't too many substances that offer both characteristics, some that do are quite toxic.
so one would grease both sides of this washer? What is the washer made of, and why does it not thermally insolate the transistor from the sink? Or does it indeed insolate it a little, and is just an undesired but necessary side effect.

Grease both sides.

The washer material is deliberately a good heat conductor.
ok I just pulled a couple sinks out of an old guitar amp that I use for parts. There were indeed bushings separating the bolt from the transistor tab. Also, there were thin sheets of plastic about the size of the transistors themselves between the sink and the transistor case. Are these thin sheets typical of the washers you speak of?

As an unrelated side-note, this was an 80 watt solid state guitar amp. It has 2 TIP3055 output transistors in it. Since both transistors are the same, does this imply its some sort of quasi-complimentary output stage? I thought this was uncommon these days (its a Rockman amp from the mid '90s, a rampage R80)
There are different types of insulators. Some do not need grease. I prefer to use mica pads and thermal grease because it works well and is quite cheap. The nylon bushings cost a few pennies each as well. Typically the collector or drain pin is connected to the tab. This is the junction where the heat builds up so it is mounted in contact with the metal tab. Some call it a head slug, typically made of tinned copper. Because metal surfaces are rough microscopically, the thermal grease fills in the gaps for more even heat transfer function. The goal is to minimize the maximum temperature of the transistor die, the silicon crystal that is the transistor. This minimizes the Pd de-rating of the device, although secondary breakdown is usually what limits BJT’s actual SOA (safe operating area). Secondary breakdown does not effect Mosfets so they are strictly thermally limited in terms of SOA. Having mica & grease or any other electrical insulator between the tab and the heatsink will add a resistor, allthough small but greater than metal to metal, in the thermal transfer function. This mean greater Pd de-rating, compared to directly mounting the tab with grease, but you don‘t want to short out the power supply.:whazzat:

I have done, at least with Mosfets, it seems a bit of improvement on typical thermal mounting Here & Here & Here. It is interesting to note that with the type of transistor used here, there are typically two versions with identical datasheets except one is in TO-220 and the other is TO-247. The only difference is Pd. Here I have taken the TO-220 version (they are quite a bit cheaper:D) and mounted them directly to small blocks of (at least 80Cu/20Zn) brass with grease. The blocks are insulated by mica and grease from the heatsink. In an audio power amplifier output transistor, Pd dissipation occurs in spikes. That is, Pd is great but with a short duty cycle. Because the brass is a dense metal, it acts like a filter capacitor in the thermal transfer trading a lower maximum temperature for a higher average temperature. In addition, the surface area of the brass in contact to the heatsink is 3 whole TO-220 mica pads, greatly increasing the surface area thus reducing the overall ‘resistance’ of the mica pads in the thermal transfer. Effectively, to a degree, I have made them into larger transistors. Geek speak translation: I can achieve >70W of output power from just one pair of TO-220 devices.:)
Last edited:


2011-11-17 3:49 pm
As Dan Moos seems completely new to this.

Mica is a clear material and is generally supplied in pre-cut shapes just slightly larger that the transistor (or IC or module) body. It is an electrical insulator but does conduct heat. Under a microscope you would see that the heatsink and the transistor aren't very flat, the silicon grease is used to fill the mounds and pits in the two surfaces to help transfer the heat more effectively.

Sil Pads are a sort of rubbery sheet, made of a silicon material that is an electrical insulator but also conducts heat. As it is soft, as opposed to the hard mica, it doesn't require the silicon grease to fill in the imperfections between the two metal surfaces.

There are other materials too but the two above are the mosty widely encountered.
basically you need to join the transistor to the heatsink without any air in between.

The thermal compound's only job is to exclude the air.

The mica or silpad's job is to electrically isolate the transistor from all other parts of the circuit.

Neither mica, nor kapton, nor silpad, nor kerafol are good thermal conductors.
They all achieve adequate conductivity by reducing the thermal resistance, by being very thin.

The exceptions are diamond and silver and copper and aluminium and aluminium oxide and few other exotics that are very good thermal conductors. Of those mentioned only aluminium oxide is a good electrical insulator and a good thermal conductor. That why it got a mention (anodise) in some earlier posts.