LM4780 4ohm @ +/-40V rails

[alexcd]

I have a 60VCT transformer laying around and was thinking about using it for a high-power LM4780 project. I want to drive two 8" midbass per chip in parallel mode. By my calculations I should get +/- 40V rails give or take a few volts. This is within the chips capability but seems awful high compared to other designs. What will happen? Just get too hot to use? Can anyone figure out how hot it would get at various power levels?

[pinkmouse]

With 4ohm speakers, at 40V rails, the amp will eventually go bang, even with big fan cooled heatsinks. I know, I've been there...

[alexcd]

So basically it'll be the "warmest" amplifier ever... hmm, how much power do you think it would be good for before it went "bang"?

[Stuart Easson]

Depending on how the rails hold up under load you might get away with it...

This is one of those cases where less is more, if your transformer and caps are big enough to stay close to 40v under load you are going to pop the chip...if on the other hand your rails sag into the low 30s under load you may have no problem at all...

Not often you hear underbuilding advocated on diyaudio...

Of course you could simply drop a few volts using a CRC or CLC power supply in which case the problem will go away and you'll have a better power supply to boot.

You should download the spreadsheet from NatSemi, it gives you power output from the chips as a function of load and voltage, it's a very handy thing for those that want to build gainclones...

HTH

Stuart

[alexcd]

CRC/CLC? Can you explain?

[alexcd]

I just checked the output power vs supply voltage on the datasheet and 40V is at the very end of the chart at around 160W... wow, that would be nice but I'm sure I can't deal with that much heat.

[Stuart Easson]

Considering just one of the the 2 rails (the other being a mirror image): a CRC power supply uses a capacitor, then a resistor, then another capacitor to generate a smoother rail than could be generated using the caps alone.

A CLC replaces the R with an inductor, typically a milliHenry or two which is a little more efficient than the CRC, more voltage for the same ripple, or less ripple for the same voltage...

As an example, simulated using PSUD2:

With 30v secondary, 2.5 Amp load, 10000uF caps

CC: ~38v, ~1.5v ripple
CRC: ~35v, 600mv ripple, 0.5 ohm, 5W resistor
CLC: ~37.5v, 600mv ripple, 1.1mH (0.1r)

The caps are the same in the 4 cases.

The inductors can be expensive if you get high end (air core etc) ones, but since you are looking to lose a few volts you probably don't want the very lowest resistance inductors anyway...

If you have the time and interest I'd recommend trying PSUD2, it's an awesome tool for power supply design:

http://www.duncanamps.com/psud2/index.html

Stuart

[alexcd]

I was thinking far too complicated when you said CRC/CLC and trying to think of some DC/DC switching scheme. Yes, a C-R-C or C-L-C would drop a few volts but I think I need to drop 8 or 9 to be in the comfort zone of the chip. Thanks for the tool. I'll check it out.

[MBK]

The bigger problem might be mains voltage variations. I use 24 V transformers and get 35-36 V rails, which is consistent with calculated expectations. To my big surprise I have measured 40 V rails on several occasions though... considering the 35 V rated caps I was not too amused.

[alexcd]

I have measured the rails. 43V-OFF and down to about 37V-CLIPPING. Overall I think I have a strong power supply (enough caps and large enough transformer.) I just think I went a little off reading the first page of the datasheet and thinking that anything under 84V(on) would be good... stupid marketing hype got me again!