1) Can someone here be so kind to explain to me why so many chips are designed with POSITIVE voltage frames????!! I cannot understand the logic.

2) Also, since they choose to make the frames electrically hot, how do you insulate them from the heat sink? A thin piece of acetate or dielectric I get, but what about the screw hole in a TO-220?

3) Is it undesirable to make the sinks grounded? (Aside from the obvious short that would occur if uninsulated).

 

#1. not sure.. I'm sure there is some logical reason.
#2. insulative thermal pad and shoulder washer for the screw.
#3. ground if required for safety,etc..

 

Can you list some examples? Voltage regulators usually have grounded tabs in the TO-220 package. With a transistor, it hardly matters since a transistor can be configured in numerous ways to have positive, negative, or ground on the tab.

 

1) The package polarity is not arbitrary. For most power devices the semiconductor chip substrate is bonded directly to the metal case or tab of its package for minimum thermal resistance between the chip and package. That means the substrate is also electrically connected to the metal and the case polarity is whatever the substrate polarity happens to be for the process used to fabricate the device. For some devices that's a positive voltage and for some, a negative voltage.

 

PB... LM350T for one... I see it often enough.

crutschow, I didn't say it was "arbitrary." That's why I asked. I figured there was some reason. Still, I can't help wonder why they would do it. The added cost to manufacture couldn't possibly cost more than a fraction of what dealing with unfavorable polarity of the package would be. Someone, somewhere makes the decision about this or that parameter to feature in a chip. One such parameter, Pd, leads to a choice to mount a metal region. The metal is surrounded by plastic or some kind of dielectric. Is it really difficult to put a thin electrically insulative layer between the electrically conductive elements? Whatever cost in heat dissipation would undoubtedly be less than whatever was utilized in installing the device for end use.

Chip manufacturers manufacture 1 chip and produce millions. Often times thousands or even millions of products are designed using those chips. Each product has to incorporate measures dealing with polarity unfriendly packages. Seems bass akwards to me. If the only reason is it's inconvenient, then the reason is as good as "arbitrary."

 

some chips are now made with their emitters connected to the tabs. at first, it was that the collectors were the underlying layer, so were connected to the case for heatsinking purposes.
chips also have a substrate, and usually the substrate is insulated from the componants on the chip by a reverse junction. this is why it is very bad for chips if they are connected backwards. not all voltage regulators have grounded tabs, the lm317 tab is not grounded.

 

.................................
Chip manufacturers manufacture 1 chip and produce millions. Often times thousands or even millions of products are designed using those chips. Each product has to incorporate measures dealing with polarity unfriendly packages. Seems bass akwards to me. If the only reason is it's inconvenient, then the reason is as good as "arbitrary."

There are some chips available with isolation to the metal tab or package but they cost more since they have to add a thin insulator between the chip and the case. So it's not inconvenient, it just requires an extra manufacturing step and insulator part which adds to the cost. The insulator also increase the thermal resistance from the chip to the package and that may be undesirable in some high power applications. It many cases it may be cheaper to isolate the heat sink (by being mounted on a PC board or using a few isolation washers to the chassis) then to pay for the isolated chip.

 

There are some chips available with isolation to the metal tab or package but they cost more since they have to add a thin insulator between the chip and the case. So it's not inconvenient, it just requires an extra manufacturing step and insulator part which adds to the cost. The insulator also increase the thermal resistance from the chip to the package and that may be undesirable in some high power applications. It many cases it may be cheaper to isolate the heat sink (by being mounted on a PC board or using a few isolation washers to the chassis) then to pay for the isolated chip.

Although there might well be less thermal resistance in manufactured in insulation than the mica or silicone-rubber insulators.

Some of the extra cost is offset by savings during equipment manufacture, often done by relatively low skilled assemblers - and subsequent costs rectifying mis-applied insulator kits.

 

PB... LM350T for one... I see it often enough.
....

In 5 decades I've never seen or used one. Hmmm...must be an oddball part.

 

That was a mistake.... The "T" is a plastic case version. I'm actually using the standard LM350. I just checked Fairchild's Datasheet. I don't see a mention of the case being positive. But, I tested mine and it is live.

 

The Motorola datasheet for the LM350 states 'heatsink surface connected to pin-2' (pin-2 is Vout)