Newbie question: why use non-isolated LM3875?

[tcpip]

Dear all,

I searched but couldn't find any comments on this one. I guess I'll continue searching... but:

Why would anyone use the non-isolated version of the amp chip under normal circumstances? I've read probably twenty complete threads about the Gainclone, and no one is doing anything interesting by using the heatsink as part of the supply rail. This means that no one actually seems to need this electricity flow.

On the other hand, having an isolated version of the chip could improve the thermal coupling between chip and heatsink dramatically. One of the problems of chip amps is that there's insufficient surface area for power dissipation compared to discretes, leading, in some cases, to lower reliability long-term. In that case, the use of a Sil-pad or equivalent becomes even more of a serious problem here, and the isolated version can totally eliminate this layer.

So, why do people use the non-isolating version? I guess I'm missing something obvious. Don't tell me it's the "warmth of the Sil-pad sound"?

Tarun

[joensd]

Hi Tarun,
I personally would never use the heatsink as conductor only
cause I like it properly soldered for good contact and don´t want my heatsinks which are most probable external to sit on one of the voltage rails (safety!).
Even if you do like you describe you´d have to use thermal grease which makes the contact even worse (OK, you could use silver grease).

Your arguments about the ability to dissipate are actually the other way round.
The non-isolated package has lower thermal resistance than the isolated!
With the isolated version it´s recommended not to exceed a certain power dissipation. Can´t remember the figure but it´s in either the datasheet or the calculation tool from National S.

Here is the thread about the calculation tool.


Cheers
Jens

[Peter Daniel]

Just to let you understand this better, the non insulated version has copper tab exposed, while the insulated is covered with plastic, which is not the best heat transfer material.

How I see it, the main advantage is simplified assembly, where you don't have to play with additional pads.

[Bricolo]

Originally posted by joensd
Hi Tarun,
I personally would never use the heatsink as conductor only
cause I like it properly soldered for good contact and don´t want my heatsinks which are most probable external to sit on one of the voltage rails (safety!).
Even if you do like you describe you´d have to use thermal grease which makes the contact even worse (OK, you could use silver grease).

Your arguments about the ability to dissipate are actually the other way round.
The non-isolated package has lower thermal resistance than the isolated!
With the isolated version it´s recommended not to exceed a certain power dissipation. Can´t remember the figure but it´s in either the datasheet or the calculation tool from National S.

Here is the thread about the calculation tool.


Cheers
Jens

The isolated version has a higher thermal resistance, OK
but the silicon (or everything else) pad you'll put between your non isolated chip and your heatsink isn't a perfect thermal conductor.

so, at the end, you could have the same thermal resistance with both models

[Peter Daniel]

Better pads have lower thermal resistance than case material.

[Bricolo]

Originally posted by Peter Daniel
Better pads have lower thermal resistance than case material.

I said "could"

That means that with crappy pads, you could end up with worse thermal resistance that the ready-to-use isolated version
but that you also can have a better one, with good pads


I just wanted to say that you can't compare the raw numbers of the thermal resistance of the isolated and the non isolated, you have to consider the pad too

[MWP]

Unless the pad is plastic is gonna be better for heat transfer than the TF package.

Insulator pads will either mica or silicone-impregnated-rubber.... both of which do a good job.

[roncla]

Cuz i bought a buncha non-insulated (LM3875T) when i got into this . LOL
No really just using ther,mal grease and nylon hold down screws i get no conductance between the chip and the H/S. I always measure the resistance and get an open circuit.
ron

[roncla]

Now if i could just learn ta tpye n speel i would do OK.
ron

[Peter Daniel]

No pad?