TO220 FET with insufficient clearance between drain and source pads on PCB?

[treez]

Hello,
We have seen competitors' Switch Mode Power Supplies which have TO220 FETs in them which are assembled without having the legs bent so that there is sufficient clearance between drain and source. How are they getting away with this?

As you can see, for a TO220 FET, the through_hole of the pad needs to be 1.2mm in diameter, the “annulus ring” of copper of the pad needs to be at least 0.3mm thick. This means that there is a maximum of 0.74mm of clearance between Drain and Source PCB copper, and the flyback has a peak Vds of 600V.

0.74mm is nowhere near enough clearance for 600V, so how are they getting away with it?


PCB design tutorial (clearance for voltages tabulated on page 9 [top])
https://www.alternatezone.com/electronics/files/PCBDesignTutorialRevA.pdf

TO220 FET datasheet
https://www.st.com/web/en/resource/technical/document/datasheet/DM00049184.pdf

 

[FlapJack]

Was the pcb routed out between the pins. A picture would help.

 

[Warpspeed]

Yup.
There certainly seems to be a lot of infectious optimism with TO220 packages.

If the voltages are pretty high, the only practical way to do it is to bend the drain leg outward (like a tripod).
That allows greater clearances for not only high voltages, but also wider pads and tracks for very high currents.

I am in awe of some of the TO220 mosfets now rated to carry 150+ amps continuous.
I look at those skinny legs on a TO220 package, and just wonder.......

For anything serious, the TO247 package looks like a much better proposition.

 

[treez]

There certainly seems to be a lot of infectious optimism with TO220 packages.
If the voltages are pretty high, the only practical way to do it is to bend the drain leg outward (like a tripod).
That allows greater clearances for not only high voltages, but also wider pads and tracks for very high currents.

Thanks, I have to say I agree, and sometimes I wonder why there are not some TO220's that are sold with legs pre-bent into the tripod shape, since it is actually a fiddley assembly process for someone to have to do.

Was the pcb routed out between the pins

no routing between pins.

 

[Warpspeed]

`The only problem I can see for the manufacturers supplying pre bent pins would be the requirement for special plastic tubes in which to pack them.

No doubt someone has come up with a very expensive tool to bend pins on the production line, but I have never seen such a tool.

Pre bent pins are far from an original idea.

 

[FlapJack]

The only other trick besides bending the pin forward and routing between pins is to shave the pads, make them oval.

Hmm, i just thought of another sneaky way to do it.

The inner layers have a different voltage spacing rule than the outer layers, much less spacing required. The top and bottom pads maybe nonexistent and the holes plated through. The connections to the device would be made on the inner layers.

 

[treez]

Thanks, I see what you mean, though of course, we'd really like some decent copper of the pad on the top and bottom layers for soldering purpose.

 

[kathmandu]

For that matter, is there a problem if one could (only) use the heatsink as a drain connection? The heatsink is further connected to the PCB by screws. Is there any downside for not using the pin at all?

 

[andre_luis]

is there a problem if one could (only) use the heatsink as a drain connection

The biggest drawback of this deed is that you would turn the heatsink with a live electric potential, becoming dangerous to someone who could get an electric shock. This also imply in 3-dimensional limitations, having to provide now sufficient keep out distance from other devices of the board ( as well enclosure, cablings, etc... ) which could be inadvertently mounted too close.

 

[Warpspeed]

The inner layers have a different voltage spacing rule than the outer layers, much less spacing required. The top and bottom pads maybe non existent and the holes plated through.

This is quite true, and it is a very curious thing.

Two electrodes in air have a much higher breakdown voltage than if a flat surface is introduced over which an arc can travel. Even very good insulators like fibreglass, mylar, or real glass cause this increased surface arcing phenomena.
But a very thin sheet of the same material placed at right angles blocking the path between the same two electrodes, hugely increases the breakdown voltage.

Inner PCB layers increase the voltage rating between pads, and so will milled slots which break this troublesome surface arcing effect.
And surprisingly so will an ordinary solder mask.

For high voltages what you can do, is use relatively large area pads which may be required with very wide tracks to carry high current.
Then edit the solder mask layer for a smaller pad size than the real pad underneath.
Just enough exposed copper area for a neat tight solder fillet, and no more, and cover up the outer edges of the pad with solder mask.

This is quite effective, I have tried it and it works.

 

[crutschow]

You could use the transistor tab as an electrical connection on a heatsink if you isolate the tab from the heatsink using a insulating washer, and use a terminal under the (insulated) mounting screw to make the electrical connection.
That would avoid having the heatsink at a high voltage.