Max supply voltage for LM3886 @ 4 Ohms

[madscience]

Hi fellow DIYers!

I got lucky recently and acquired a huge heatsink. Cut in half each heatsink has a K/W value of 0.4! I'm already building a very nice aluminum case and I'm thinking of getting as much power as possible from a single LM3886 (per channel). I still want to have the option of feeding 4 ohm speakers. Now, the datasheet says that a supply voltage of +/- 30 VDC should not be exceeded for a 4 ohm load. I recently found a spreadsheet file called the "overture design guide" which says that a supply voltage of +/- 35 VDC is totally ok if the "T" package is used (roughly 1 K/W). I'll have to isolate the package from the heatsink since it is part if the case (thus connected to ground). This will of course cause a worse K/W value. Now if I assume that the insulator has a K/W value of 0.5, the required heatsink should be (1-0.5) K/W = 0.5 K/W! Mine has 0.4 which should be big enough even for full power operation @ 4 ohms, right?
Any opinions and warnings are greatly appreciated! I will post pics of my case as soon as it is finished!

[peranders]

The tab is connected to V- (In 99.9% connected to the lowest potential, valid for all IC's)

You could have V- in the heatsink but then again it is all dependent of which music you play and also at what level. Don't forget that you have overheat protection so you will hear when the cooling isn't enough and it won't harm the amp in the short run.

[Mick_F]

The limitation of the rail voltage for 4 Ohm loads is not a heat issue. The output power steeply drops at low loads and this occurs at higher loads for higher rail voltage.
Have a look at the plot "Output power vs. load resistance" on p. 13 of the datasheet.

Mick

[Stabist]

Hmm - why needing heatsink with K/W 0,4 for one small LM3886 that doesn't heat so much at all??? I mean - mine stereo GC has both!! chips mounted on one copper bar - 85 x 44 x 20mm.
And I have +/-28V PSU voltage ...
And the whole thing gets to some 15°C over ambient ....

With 2 of yours heatsink you could easilly build something like Aleph30 or Zen v4 with 100W disipation per channel - just to ilustrate what those heatsinks are able to cool down!!

[bogicp]

hello madscience,

You must use overture design guide xls table very carefully.
You must type ambient temperature, but this isn't room temperature
but temperature near the chip, in your very nice aluminium case
In case, temperature may be from 30-50 Celsius degrees, in normal
usage, even if temperature in your room isn't greater than 24 degrees.

Try to measure temp in your computer box, and you will find what is
"normal ambient" temperature


Best regards
and good luck.

-boggy

[madscience]

Thanks to everyone for your advice! Yes, determining the ambient temperature could be the key to my problem. I can only "guesstimate" that it shouldn't be too high since there are no other components inside the case that will be generating a lot of heat. This is because I plan to use an unregulated supply, two big rectifiers and a 300 VA toroidal transformer. Considering that the case is 100% alu, heat should be removed quite quickly. Hmm...I think if I'm especially careful about mounting the ICs to the heatsinks I should be on the safe side. By the way: Does anyone have experience with different types of insulators like mica, silicon, etc.? Which one is best for heat transfer?

to Stabist: Yeah, I was considering a class A amp but I must admit that I don't like the concept of having a huge heater that puts out some 25 W of music...

[bogicp]

No problem, but note that you can hold some other equipment above or below your amp, or any other heat source, like sun, for example...

Best regards,

-boggy

[macboy]

Does anyone have experience with different types of insulators like mica, silicon, etc.? Which one is best for heat transfer?

Apparently, berilium ceramic insulators are the best, but they are both expensive and highly toxic to both you and the environment. Mica are better than silicone rubber, but you must use heatsink compound (special silicone-based white greasy substance) between the Mica and the metal surfaces. Mica is terrible when used dry. Silicone rubber insulators are only slightly worse, and are often, but not always, designed to be used dry. They can be used dry because the rubber will conform to the imperfect surfaces of the heatsink and device, filling in the microscopic air gaps.

Even better of course is to isolate the heatsinks from the chassis, instead of isolating the devices from the heatsinks. Then you can connect your devices directly to the heatsinks with some heatsink compound, gaining a few tenths of a deg.C/W. This is relatively safe for you to do, since your heatsinks will be enclosed inside the chassis, protected from curious fingers. Those using external heatsinks are not advised to connect them to the devices directly. Just make sure that you remember that those heatsinks are "live" when you open up the chassis to poke around inside.

[madscience]

Thanks for the advice . I think I'll go for the silicon types since I've used them before and they are quite a clean solution. There's no chance I can isolate the heatsinks from the chassis since they are part of the chassis, all alu plates are screwed into the heatsinks.
Now, time to go back to