Power Transistor compatibility question (pics)

[billbyrd1945]

Please see attached photo. I ordered the one on the right to replace the one on the left. It attaches to a substantial ribbed heat sink. The new one seems to be all plastic while the old one has a metallic back plate. Will this work? Thank you

[commongrounder]

Isolated transistor packages have less thermal conductivity than the metal tab versions. This means less maximum power dissipation. The data sheet says max 75 watts for the metal tab, and max 28 watts for the plastic version. If the circuit parameters are unknown, it would be prudent to obtain a replacement with the same type package.

[billbyrd1945]

Thank you. I will do just that.

[coromonadalix]

+1  had the same problems recently at my job,  had to temporarly put the plastic ones to substitue an metal one who's going obsolete, and slightly redesign a circuit to accomodate a new type,   the plastics ones operate near their thermal limits.

[xavier60]

A metal tab would still be present under the  epoxy and could be exposed by sanding if desperately needed.

[David Hess]

If the metal package was used with insulating hardware, then the difference in performance is closer because the plastic package does not require an additional insulator.

[billbyrd1945]

Re: sanding them down...
Being a neophyte to electronics, I'd feel reckless modifying anything. Some engineer with a fully functional brain twice the size of mine, wanted plastic. I don't understand it. But what else is new?

[David Hess]

The plastic packaged part is not entirely plastic.  It is a metal tab TO-220 part with a plastic overmold so the thin plastic on the back replaces the insulator which would normally be used with a TO-220 metal tab part.  This makes assembly much easier.

[billbyrd1945]

I'm confused (as usual). Doesn't the transistor require heat sinking? I'm trying to understand why the plastic would ever be a good thing (ordered it by mistake).

[David Hess]

I'm confused (as usual). Doesn't the transistor require heat sinking? I'm trying to understand why the plastic would ever be a good thing (ordered it by mistake).

It is still a metal tab transistors but the tab is overmolded with plastic which serves to replace the separate insulator and flange washer which would normally be used with a standard metal tab TO-220 part.  Typically a mica insulator would be used and that adds significantly to the thermal resistance of the metal tab package so the plastic overmold package is not as terrible compared to metal TO-220 package plus mica insulator as the datasheet specifications indicate.

The advantage is increased ease of assembly and greater reliability.  There are a lot of ways to screw up installing a mica insulator and associated hardware.

[wraper]

If original transistor is without or with a small heatsink, basically there is no difference. If it's a heavily stressed part on a largish heatsink, transistor on the right should not be used as replacement.

[coromonadalix]

THERE IS A DIFFERENCE 

The plastic transistor may have 1/3 of the metal tab transistor power dissipation ALWAYS check the datasheets ... you could have big surprises, do not take that for granted.

[David Hess]

And then do the calculations for the metal package transistor with appropriate insulating hardware.

[wraper]

THERE IS A DIFFERENCE 

The plastic transistor may have 1/3 of the metal tab transistor power dissipation ALWAYS check the datasheets ... you could have big surprises, do not take that for granted.

Did you freaking actually read the post? 

If original transistor is without or with a small heatsink

With small heat dissipation it does not matter how much additional thermal resistance plastic layer gives. When without a heatsink, plastic package may actually dissipate heat better due to better thermal radiation of black plastic surface compared to shiny metal.

[billbyrd1945]

I thought it might be helpful to this conversation if I posted a pic of the actual heat sinks. Going purely on intuition (that's about all I have), these sinks are telling me that the transistors will or may get quite hot and need a metal to metal interface for bleeding off the heat. I hear some of you talking about insulation, and the benefits of it, and I'm totally baffled as to the benefit of insulating a hot component. But-- there are tons of things that work contrary to my intuition. There. Just wanted to remind the crowd how close to the ground my position is on the totem pole. I'm really appreciating all the input. 

[glarsson]

That's a tiny heatsink. You might need to isolate the transistor from the heatsink if the heatsink us attached to something conductive or can be touched by someone. The metal tab of a TO220 is connected to the middle pin...

[David Hess]

I agree that heat sink is small enough that I doubt the increase of thermal resistance from the overmold plastic package is significant.

[SeanB]

It will work there, just screw it on with some new thermal compound. I would also check the 4 mains diodes D13 to D16, one looks kind of odd looking, and bears checking out if it is shorted or open. I see you took C15 out, was that because it was bulged or just to get the transistors out. If it was bulging then change the other one there as well, 120uF 200V, as they really need to be close in value and ESR for best results.

[billbyrd1945]

I replaced all caps. The diode looks suspicious because of hot glue that spilled on to it when I was stabilizing the cap. All of the diodes check good on diode mode of my Fluke 15b+ (flow in one direction only). I'm surprised to hear that these heat sinks are considered small. And now I'm surprised that one or more of you see the arrangement as a shock hazard (if I understand you correctly)! Why would they design it to be dangerous? I'm not doubting you, just trying to understand how it would serve a Chinese manufacturer to make something like that.

[xavier60]

I hope that the thermal paste isn't metal loaded? 
It's likely that the heat sinks are tied to live side ground to reduce EMI. I mean HVDC negative rail.