LM3886T

hi I have 2 LM3886T chips around and have been reading in the forums that I need to isolate the chip when I connect it to a heat sink I just wanted to know what types of pads I should use in-between the heat sink and the chip and also what kind of thermal compound would you recommend (is normal cpu thermal compound good to use?)


Thanks
 
Member
Joined 2009
Paid Member
If you do indeed have the LM3886T - the one with the metal back - then you do need to isolate it from the heat sink as the metal back is connected to VEE of the chip.

If you search the usual suspects (Mouser, Digikey, RS, Farnell, etc.) for thermal pads and narrow your search to those that fit a TO-247, you can usually find something that fits. It'll take a little searching to find a pad that is large enough and has the mounting hole in the right spot, but it can be done. You can find the mechanical dimensions of the IC on one of the last pages of the data sheet (available at TI.com).

Rather than messing with the thermal grease, I suggest getting a silicone pad, such as SilPad 400, 800, or 1500ST. Make sure that the pad you get is non-conductive.
If you do decide to go for mica + grease, the CPU grease will likely be fine.

Tom
 
If you do indeed have the LM3886T - the one with the metal back - then you do need to isolate it from the heat sink as the metal back is connected to VEE of the chip.

If you search the usual suspects (Mouser, Digikey, RS, Farnell, etc.) for thermal pads and narrow your search to those that fit a TO-247, you can usually find something that fits. It'll take a little searching to find a pad that is large enough and has the mounting hole in the right spot, but it can be done. You can find the mechanical dimensions of the IC on one of the last pages of the data sheet (available at TI.com).

Rather than messing with the thermal grease, I suggest getting a silicone pad, such as SilPad 400, 800, or 1500ST. Make sure that the pad you get is non-conductive.
If you do decide to go for mica + grease, the CPU grease will likely be fine.

Tom

is this the kind of stuff your on about
http://www.digikey.co.uk/product-search/en?mpart=SPK10-0.006-00-90&vendor=211
Bergquist SP 400AC-54 SIL-Pad 400 Adhesive Pads TO220 (Pack of 22) | Rapid Online
it says its thermally conductive is that fine?
 
Last edited:
Member
Joined 2009
Paid Member
is this the kind of stuff your on about

Yes. That's SilPad 400. You'll need something larger than the 19.05x12.7 mm TO-220 size to fit the LM3886, however.

it says its thermally conductive is that fine?

I suggest taking a look at what you are aiming to do. You're trying to form a good thermal connection between the LM3886 and the heat sink while keeping them electrically isolated, correct?

To be more precise: Beware that some of the pads are electrically conductive. Avoid those.

Tom
 
DIY guru Rod Elliott says that SilPads are not nearly as effective as a piece of Kapton and some thermal grease.
SilPads come in a variety of types and there are many other manufacturers who make many types that get referred to as silpads even though they are not.

They all perform differently.
Bad versions are worse than thick mica
Good versions are better than thin mica.
That covers an enormous range in Thermal performance.
 
Yes.

Grease fills in voids/air pockets which*destroy* good heat transmission, and since surfaces never ever are perfect (a microscope picture will show pits and throughs all over the place), you need "something" to fill them, hence the grease, which to boot is loaded with metal oxide dust which turns it white.

It´s not a problem of mica which can be quite flat, or kapton which is flatter, but of the metal surfaces in contact (transistor case and heat sink surface).

Metal oxide dust is also electrically insulating.

Beware that some CPU cooler thermal grease (Arctic, etc.) is loaded with metal dust (silver - aluminum - etc.) which is DANGEROUS in a transistor or chipamp situation, it shorts everything.

Silpads and such are made of elastic rubber which fills in the tiny voids, almost "as if" it were grease.

Important:
  • on the TOP example, screw touches chipamp flange, it does not touch heatsink or chassis
  • on the BOTTOM example, screw touches chassis but does not touch chipamp metal.

Both are valid, pick the most convenient to you.

As an extra precaution, check with multimeter low ohm or continuity scale that chip amp flange does NO TOUCH heatsink/chassis/ground.

3-s2.0-B9780081017647000098-f09-16-9780081017647.jpg


this is s TO3 mica and nipple image, only to show what nipples look like .

TO3 mica nipple.png
Important: nipple goes through both transistor/chipamp flange, AND through mica AND often into a 4mm or so hole drilled/punched through heatsink or chassis, and screw goes through it.

A common screw size is 3mm metric, maybe there is some crazy Imperial or American size matching that but you can´t go wrong with Metric.

Notice nipple acts both as a shoulder and a sleeve which always separates screw metal from heatsink and chassis.
 
Last edited:
If the heatsink is also connected to VEE, then it may not be necessary to isolate the thermal tabs. Though the thermal paste would still be required (for the voids/gaps), this arrangement could potentially give a better interface, as the mica/silpad is no longer in between the IC and the heatsink.
 
Is it necessary then to connect heatsink to VEE via extra wire? Or just leave it floating?

Dear archeo,

Do not connect the heatsink to any potential other than Earth and keep it isolated. Now I just feel like I shouldn't have written post #12. #12 was intended in a particular case where the amplifier chip is to be operated on a single isolated supply, whose VEE would then be already grounded.

And, even if one chooses to do otherwise (for whatever reason), power entry must not be made by means of an electrical connection to the heatsink (as in the inverter board shown below), as any HF noise present could then transmit into the air using the heatsink as an antenna.
 

Attachments

  • sukam inverter.jpg
    sukam inverter.jpg
    90.8 KB · Views: 58
Last edited:
Another reason you don't want to conduct power through the heat sink is that the IC likely does not have a bond wire from the die to the DAP (die-attach paddle). So while the DAP (that metal back of the package) is electrically connected to VEE, it's not likely to be a very good connection. It may simply be the die substrate connected by a dab of conductive epoxy. Conducting power through a power pin is much better.

Tom