Heat Dissapation Problem LME49720/OPA1632 Preamp

[peranders]

Just glue it. Time will tell if the joint will break up. You can test with super glue to start with.

[BWRX]

Originally posted by artquake
It's somehow logic that heat transfer is better when the pad has a direct connection to a bigger materialarea or mass with a high heat capacity. How can I solder/realize this ? Is it not problematic that the IC get's to hot while soldering a bigger mass to it ?

Originally posted by BWRX
It could be something as simple/cheap as a thin strip of aluminum from a soda can that you place under the part with a little thermal paste before soldering it down. That will help much more than any heat sink you put on the top of the plastic package.

The part is meant to be soldered in a reflow oven but it could be done by hand since the part is rather small. It does take some skill and a proper layout though.

Like I said above, try placing a thin strip of some kind of aluminum or copper under the part BEFORE soldering it down. I.E. cut the thin strip of aluminum to size and place it in between the pads, place a little thermal grease on the thermal pad of the part, place the part down onto the board, then carefully solder it into place. You can bend up the extra aluminum sticking out the sides, curl it, cut it, or whatever.

Here's a picture of what I'm talking about: http://www.diyaudio.com/forums/attac...amp=1113442445
(from this thread: http://www.diyaudio.com/forums/showt...288#post619288 )

The chip in the picture has a thermal pad on the bottom of its package and it didn't come soldered to the PCB like it was supposed to be.

[jackinnj]

PowerPad Made Easy http://focus.ti.com/lit/an/slma004b/slma004b.pdf

I'm going to assume that the bottom of the adapter has no "wiring on it" -- if it has a ground plane so much the better.

If you can rework the adapter by removing the IC and make a VIA with a 37 mil drill (I'm thinking in inches/pounds/seconds so you have the conversion to mm.) Use a small piece of wire of the same diameter through the VIA to the other side. To this attach a piece of copper tape to spread the heat.

With SOIC, MSSOP etc it really is better to design the PCB yourself -- I learned this making a bunch of class D guitar amplifiers with the Texas Instruments chips. You punch a bunch of vias through to a copper plane and spread the heat out. It's easier to get it right in the first place.

wrt Peranders recommendations of SuperGlue some of these cyanoacrylates won't stand up to a lot of heat and some are used in very stressful industrial applications -- i couldn't find from their website the point at which it will break down. time and heat are not necessarily your friends.

[jcx]

a heavy Cu (or maybe Al) foil strip under the package could be glued to the adapter for stability, avoiding shorting the inner edges of the pcb pads, and a small dot of heatsink compound applied to the pwr pad, then the chip could be soldered to the adapter in the normal orientation with a little bending of the leads – a miniature hobby C or spring clamp could be helpful

@~1mm length the pins should easily bend down 0.1-0.2mm to accommodate the foil thickness – much easier than milling the adapter pcb at the desired tolerances on parallelism

the foil strip could be longer than the width of the adapter, and wider than the chip except where it threads under the chip, bend the ends up in a “U” shape

I would use nonconductive ceramic/oxide filled thermal compound – the silver filled stuff could give capacitive and possibly DC electrical shorting if it spreads to the pcb pins

another option is belly-up mounting, then the heat sink can be attached to the pad, I've used heatsink compound and clamping but the PC video memory heatsink people have made thermal conductive epoxies readily available in small quantity
I bent the leads down on a TPA6120 one at a time as I soldered them but that was a standard SO pitch, not the MSOP – the TPA is actually totally symmetric so I didn't have to change the wiring like you'd have to with the 1632 mounted upside down

[theAnonymous1]

I know I plug this stuff way too much, but J-B KWIK WELD epoxy works wonders for holding heatsinks on. It's a whole lot cheaper than "real" thermal epoxy too.

http://jbweld.net/products/jbkwik.php

And yes, those little OPA1632 get blazing hot . I cut a fin off of an old CPU heatsink and used some KWIK WELD to hold it on.

[artquake]

Hi all,

thanks again for your constructive informations.

Trying to attach the SMD heatsink with superglue failed. The contact area is to small and the surface of the heatsink or even the chip is to flat, that when I press (wich i really should, to achieve a high conductivity) it never will be friction-fitted. So i forget glueing.

BWRX all ok, good solution, but i can't do it in case of aesthetics. Thank you nevertheless.

Jackinnj, that would be also a solution. If I understand you correct, you ment drilling holes from the adapters downside to top (or vice versa ), while placing the holes in the thermal pad area mentioned in the TI-Spec. Sheet of OPA1632. And then sticking wires thru the frilled holes and attaching them to a peace of copperfoil. Right ?
How can achieve good contact pressure with the wire from the bottom side to the PowerPad plane ? Designing a pcb in the beginning for this would be the best. Nice for the one who can do this.

Well if alexw88 would join and make an offer for the gerber files that would be great !

On the other side i found a Thread of Stef1777 wich designed an outputboard for the DEQ2496 from Behringer which i own too. He uses the OPA1632 and considered also thoughts relating to high temperatures. TEXT



So i quote the statement of janneman:

„Originally posted by janneman



Steff,

I think the OPA1632's are fine at 75degrees. If you feed them from +/- 15V, they dissipate about 450mW. They are designed for that. Even if you cannot touch them, 75degr is not hot by ss standards. With 450mW and at 170degree/Watt thermal resistance, the die will be about 76degr above ambient. If the case air temp is 40degr, the chip will be internally, the die, at 40+76 is 116 degr. The max is 150degree, with recommended operating max of 125degree. So, no problem.
You should stop thinking about this with your 'human' feelings of heat

Jan Didden“

Do i have also have some human feelings of heat ? Or is this temperature not critical ? If so that would be really good. Anyway i could not improve sound while using a p2p connection board. It really has to be a well done pcb, which alexw88 has done really well.

best regards
artQuake

[jackinnj]

Originally posted by artquake

Do i have also have some human feelings of heat ?

this will be the case until you get married.

[artquake]

and yes that I am ... Sorry for my english. Incisive diction isn't often easy.

[jan.didden]

Originally posted by artquake
[snip]Do i have also have some human feelings of heat ? Or is this temperature not critical ? [snip]best regards
artQuake

What I mean, at 50degr Celcius, you get unbearable pain and blisters on your fingers. The opamp just gets cozy at that temp. So we tend to judge the chip temp with our 'human' fears of heat.

If you worry about the OPA1632 temperature (for instance if you expect the ambient temp in your case to grow above 50degr Celcius), you have two options:

- lower the supply voltage to, say, +/-9V. That gives you 1/3 less dissipation. In my DCX2496 replacement board I run them at +/-9VDC (because also the CS3318 runs on +/-9VDC) for about 5V RMS max undistorted output. Very good sound.

- change to the MSOP PowerPad package and design the pcb as shown in the data sheet for the PowerPad package. That gives you more heat removal capacity.

Jan Didden

[artquake]

Thanks Jan for reply,

if you expect the ambient temp in your case to grow above 50degr Celcius

I don't, as the enclosure will be vented thru the bottom plate, and so heat will flow in natural convection thru the enclosure out of the top of the enclosure to ambient. So the temperature of the air (not including the radiation of parts) inside of enclosure will get max. to 40 °C (dependig on the position inside the enclosure). Let's estimate Radiation for 70% and convection 30% of entire dissapated power. But the p2p-Board at present is on free air, so when mounting the board into a enclosure these parts will get basically hotter. How hot who knows. To simulate that with FEM is to complex for that dead time it needs, and for, in fact results that are far away from reality. No way but:

Asked in short words some further questions:
Do you think now that a surface temperature on the MSOP OPA1632 of 70...75°C has to be minded critical or not ? Still cozy for the OPA ?

and:

lower the supply voltage to, say, +/-9V

Wouldn't this action increase the THD + Noise ? In my case it's allready noted as average @ -75.5 db(A) / +0.017% and i think this preamp could perform even better @ 15V ...