LM3886T?

[Arn C.]

Howdy,
I am replacing a LM3886T (Not isolated -metal tab), can I use a LM3886TF (isolated
no metal tab) in its place?

Thanks!
Arn C.

[R.G.]

I am replacing a LM3886T (Not isolated -metal tab), can I use a LM3886TF (isolated
no metal tab) in its place?

Maybe. It fits the same holes, acts the same electrically. The insulating coating is not as good at getting heat out of it as heat goo plus mica film or one of Sil-pad heat transfer wafers.

The further you are away from thermal shutdown, the more the answer is yes. If you're right on the edge (max supply voltage, 4 ohm speakers) the answer becomes no because it will shut down from thermal overload as it gets hot. This isn't destructive to the device, but it blows a heck of a hole in a smoking hot solo.

[Arn C.]

Thanks RG!

I guess I should stick with the original this time.  But if I was to use it would I still have to put the heatsink goop on it before I mount it?

Thanks again,

Arn C.

[anchovie]

Yes, you want to get the best thermal transfer possible whichever IC variant you use.

[R.G.]

Yes, you want to get the best thermal transfer possible whichever IC variant you use.

Absolutely correct. Waste heat is always an enemy of power amps. Whatever you're using, you have to get enough heat out to keep the chip cool. In practice, that means keeping the microscopic roughness of the IC and the heat sink surface from trapping a layer of air between the two surfaces. You do this by filling the layer with a thin film of compliant stuff that fills the hills and valleys with something. Almost anything that will stay in the crevices is a better heat conductor than air. But stuff that's filled with heat conductive material can do even better. And that is the recipe for heat sink goo - a carrier filled with very finely divided particles of conductive stuff. The stuff can be one of a number of oxides or ceramics, or metal particles; some use graphite.

Incidentally, the best heat conductor ever found is an insulator. Diamonds are significantly more heat conductive than solid copper. They'd make a great thermal material if only they could be shaped as needed and didn't cost so much.

How's that for an engineer's view of the world? 

[PRR]

> I am replacing a LM3886T

Maybe we should ask "Why?"?

The 3886 is nearly blow-up proof.

If you know you fed it 100V, or got the + and - swapped, or dropped a hammer through the box, OK.

If a commercial product is SO cost-cut that the poor chip overheated millions of times (it CAN be killed but you have to work at it), then you probably want to improve the heat sinking first. More case-holes. Fan. Even a sink-swap.

Check that the V- pin is wired on the PCB, that they didn't save one trace by wiring the V- through the heatsink and tab. That's mandatory on TO-3 pack, uncommon but possible on 3-pin TO-220, unlikely when there's 13 other pins.... but check.

The "heat penalty" for using the plastic-wrap chip is small. The main obstacle to heat is usually the heatsink's grasp of the air, unless the chip-sink joint is badly made up. If a well mounted TF isn't happy, then the heatsink is designed too close to the edge of catastrophe. Especially if the previous chip overheated so many times it quit.

You need the goo. Two "flat" surfaces are only really touching on their 1% high-spots, 99% air gap. Air is very bad for heat. Anything is better. A drop of sewing machine oil fills the 99% airgap with something not as good as metal, but a lot better than air. At midnight in Lodi, you can take a drop off the van's dipstick. But these are runny. Firm zinc-loaded grease is readily available. Radio Shed may still have a 99-cent tube. Your neighborhood PC hotrodder has a tub of magic nano-silver polarbear CPU goo (the Shed does now but it's hi-profit). Unless you are tweaking a Sexium two more MHz past 150% of rating, you don't need the magic stuff, plain heat goo is fine.

Inspect the mounting for burrs and warp.

Gus posted a link to screw-torques. Good plan. My feeling is that on small screws there isn't much in-between too-loose and stripped-out. So I just tighten until TIGHT but not in danger of stripping.

Diamond may be R.G.'s best friend, but there will still be gaps. Diamond dust goo is still a lot of grease. A diamond heat-plate might be wonderful soldered to the silicon chip. Also eliminates an insulator and one gap to goo (assuming diamond can take screw pressure). Might be great for over-speed CPUs which need small die size and high current for speed. But in audio we rarely want the kind of heat density that can't be handled with a careful joint and ample heatsink.

[Arn C.]

First, let me thank you all for your assistance.  Much appreciated!

You ask why?  Well, I am working on a digital crate amp (DXJ112) for a friend and there is no output.
I tore it apart and most everything is surface mount, but on the output board there are a couple of regular components
and the Lm3886T.  I just figured that I would through one that I know is new in there and see what happens.
Unless, anyone else has had this problem with one of these amps and knows where to look for the problem.

Thanks again gents!

Arn C.

[PRR]

If you are blind-guessing.... the 3886 may not be the place to start. It's hard to kill. It has a lot of pins and goop.

Basics: reasonable supply voltageS at the supply pins. OUTput pin either at zero or at half-rail (for single-rail cap-output).

Then use a 0.1u cap from the INput pin to a guitar cord (shell to Crate chassis) to a small Turned-Down amp. Fire up the Crate and play into it. If all that surface-mount junk does its job, there will be a BIG (relative to guitar) signal at the 3886 input pin. It may overload the monitor amp even at "1"... but if you got signal that's something.

If no signal here, leave the 3886 alone. Look through the SMT junk for cracks, missing parts, loose connectors, burst caps, junk pots and switches... the usual stuff.

If there's signal IN to the 3886, move your cap-amp to the 3886 OUTput. Play gently. This should be a HUGE signal. If you got that, but nothing in the speaker, follow the path, inspect the wires and jacks.

[R.G.]

Diamond may be R.G.'s best friend, but there will still be gaps.

We do all like sparkly thangs, don't we? 8-)

NASA used diamond heat spreaders on some critical components at least at times. Removing heat in a vacuum is much harder than in an atmosphere. You can't conduct it away and you can't convect it away; all you have is radiation. So getting the heat from the hot part to the radiating fins requires the lowest possible thermal resistance to get the heat where it needs to go.

Lapping is a common process for supercritical heat transfer jobs. You use a flat surface like float-process plate glass or a surface plate, an abrasive, and the thing you want to be flat. You abrade, holding the thing you're trying to flatten against the surface, moving in alternating circular motions and figure 8 patterns until the surface appears evenly frosted, then wash away the abrasive and star again with finer abrasive. As the scratches left on the surface get smaller than half a wavelength of light, the surface becomes reflective. This is how they make lenses and optical mirrors. You can get an optically flat surface. If you lap the top of a 125W CPU and its heatsink together - and do a good job! - heatsink goo becomes not only useless but counterproductive. But no one has time to spend a couple of hours per device lapping surfaces optically smooth.

A diamond heat-plate might be wonderful soldered to the silicon chip. Also eliminates an insulator and one gap to goo (assuming diamond can take screw pressure).

That's what NASA did. Diamond sheet was (at the time) so expensive even for NASA it was used only where super- critical. That may be changing. Vacuum deposition techniques have advanced to where diamond is commonly deposited as a film on some things. I've seen it proposed as an anti-scratch coating for lenses. Boy is that a made-in-heaven match!

[Gus]

Check things like PRR posted.  The input to the amp chip will tell a lot.

You will need a eye loop etc or a good set of eyes look at every solder connection at the SMD parts.  Thermal cycling and/or vibration can break SMD solder or thur hole solder connections.

I have "fixed" a few laptops by re soldering connections at SMD ICs.