LM3886 at high voltage and power?

[Eric Weitzman]

The LM3886 spec sheet shows that it's possible to get 70 watts out of it at 0.1% THD driving an 8 ohm load using a +/-35V supply. You can even push it to 85 watts with +/- 40V.

Does anyone have any impressions about the sound of the amp running at these levels?

[jackinnj]

I have found that the harder you push the LM3886 and LM4780, the lower the THD+N% --

abandon hope on the minimalist designs if you are going to push the chip really hard, for reasons dealing with the protection circuitry -- bypass the +/- as close to the power supply pins with 100nF ceramics, use polypropylene (need not be expensive) caps in the snubber circuitry.

[Eric Weitzman]

Thanks jackinnj. I'm considering building an 8 channel 60WPC amp using the LM3886. I will do a proper design following the proceedure in the spec sheet if I go ahead with this, so I should be able to keep the SPiKe circuitry from turning on.

[tlmadsen]

I have been running LM3886T at +/-42 V without any kind of problem. Only issue is that you need a lot of cooling. A 2 x 30 V transform will give you just about 40-42 V DC and I don't think you should go any higher that this.

Have fun

BR

Thomas

[macboy]

I have been running LM3886T at +/-42 V without any kind of problem. Only issue is that you need a lot of cooling.

Just a reminder to everyone reading this: the LM3886T is the non-isolated (metal tab) version of the chip. This one has much, much better die-to-heatsink thermal transfer than the isolated (all-plastic) TF version. I would never attempt to run the chips at these kinds of extreme voltages (and corresponding high heat dissipation) with the plastic package. They just won't survive it very long, and will probably end up going into thermal protection mode even if you don't realise it.

[Eric Weitzman]

Originally posted by macboy
Just a reminder to everyone reading this: the LM3886T is the non-isolated (metal tab) version of the chip. This one has much, much better die-to-heatsink thermal transfer than the isolated (all-plastic) TF version.

I have read similar statements in this forum, but I'll be a monkey's uncle if I can find anything in any of the spec sheets that show different θCS for the T and TF versions and I've looked for it. Can you point me to some documentation from NatSemi about this?

[jackinnj]

right here on there interactive design tool:
http://www.national.com/appinfo/audi...gn_Guide15.xls
rows 18, 19 -- column "N"

I will be a monkey's uncle if someone can demonstrate that the insulated version sounds different from the non-insulated version.

[Eric Weitzman]

Originally posted by jackinnj
right here on there interactive design tool:
http://www.national.com/appinfo/audi...gn_Guide15.xls
rows 18, 19 -- column "N"

I will be a monkey's uncle if someone can demonstrate that the insulated version sounds different from the non-insulated version.

THANK YOU VERY MUCH, jackinnj. I just played around with the spreadsheet and it seems that the θSA of the heatsink when using a TF package is always 1c/w lower than for the T package. So the θCS of the TF package must be exactly 1c/w worse than the plain T.

Thanks again.

[macboy]

Looks like jackinnj beat me to the response. The infamous app note AN1192 ("Overature Series High Power Solutions") also mentions the different case types in the Thermal Background section. The best reference though is the Overature design spreadsheet that jackinnj pointed to, as it actually does (approximate) thermal calculations for you based on load impedance and supply voltage.

... it seems that the èSA of the heatsink when using a TF package is always 1c/w lower than for the T package

.
That is correct. And that is a very substantial difference, not to be dismissed with a wave of the hand.