LM4780 and cooling


diyAudio Moderator Emeritus
2003-01-18 9:50 am
Hi John and welcome

I fixed your links, if you happen to correct them on your site let me know and I'll redo the job.

40 to 51° above ambient. Of course it's hot in France these days.
What is the room temp?
Is that with no signal or at normal volume?
I see that your heatsinks look fairly heavy so perhaps the amp is oscillating.
Do you have a scope at hand?



diyAudio Moderator Emeritus
2003-01-18 9:50 am
Without looking at the datasheet, 10°C above ambient looks fairly normal to me for a 60W amp and 21°C at full power also. A lot depends on the idle current of the chip, and the heat sink properties.
Only a scope could tell you if the amp oscillates.
Also, a little fan can do wonders these days.

/Hugo :)
Are you running the LM4780 in parallel? If so make sure both side have their resistors very closely matched, otherwise your amp will run hot because the two halves are fighting each other. I had that problem, and resistor matching fixed it.

I also have a LM4780 amp that powers my mains, and for added safety just to be sure I never hit the SPIKE protection I added a 12V fan, which is powered by about 4V so it runs very slow, it keeps the chips very cool, and is silient.

I hope that helps.
Hi John,
in my opinion the heatsink for two (you used two chips each in parallel-mode, right?) chips is toooooo small! also i think this heatsink isn't good for cooling without a fan: the rips are too close for passive cooling...:bigeyes:

In my 4780er i used a heatsink with about 0.8 K/W thermal resistence. Because auf the dimensions of such a heatsink you have to spread the active area of the chip by adding an additional heatspreder. I put a copperplate with about 4 or 5 times the chip area and 6mm thickness between chip and heatsink so i increased the active area which transferes the heat from chip to the heatsink.

BTW: don't forget to mount the chip isolated on the heatspreader/ heatsink...

PS:I place a photo of my solution if i'm back at home...
hope this small image shows the idea of heatspreading...



  • heatsink.jpg
    48.8 KB · Views: 682
No, I don't use the chip in any parallel/bridge mode.

I use it in the basic configuration: 2*60W. I utilized the datasheet of National Semi to make my pcb (figure 7 in the datasheet ).

I mounted one package on the heatsink. The heatsink results of a 2*100W old amplifier (maybe a STK**** , but I don't know).

I made this amplifier for a very low price. I baugh only resistors and capa (exept the 2*6800µf).

For the package I utilized a mika isolation and thermal paste.

Considering your remarks, it is normal that heats much.

To Slakman: you wrote "BTW: don't forget to mount the chip isolated on the heatspreader/ heatsink..." . What is the consequence of no?
The backside of the chip holds a metall area, right? this is for better thermal flow to the heasink. This Area is connected to V-!!

So if you mount this unisolated on the heatsink and the heatsink is mounted directly on the metall case ALL metall areas are connected to V-!!

And: if you have grounded your chassis with the earth-connection there will be a shortcut between earth and V-, so normally the FI should switch power down (in your house/appartement).

Boy, boy - the fundamentals of electricity!


(Maybe this is the reason why your amp runs hot...)
Well, temperature in my room is 30°C.

40°C-> with no signal.
Your amp is definitely, absolutely oscillating.. Without any doubt whatsover. When attached to a heatsink, and with no signal present, the amps should not get warm. The fact that they are getting warm indicates that they are not just sitting there idle. They are doing something. They are oscillating. You need to fix that, not your heatsink. Your heatsink is fine.
I tend to agree with Macboy. At idle the 4780 typically draws 110 mA, or around 8W with your rails. With that sink you should notice that the chip is warmer than ambient, but barely with no signal.

Put an ammeter in line with the V+ supply (disconnect the v+ lead from the supply, connect one ammeter lead to the wire and the other to the supply where it ws removed) and measure the current - if over 175 mA you have something wrong.

Are you sure that the board is right? Are there any solder bridges? All pins in the proper holes and no legs touching another? Use a magnifying glass to check. Move the signal leads around - try to keep them as far from the output leads as possible.
this all can be but without isolation of the chip <-> heatsink this circuit is definatly a DANGER!

It also seems to me that the output cables are very close to the back of the chassis. Would be better to solder them at the end of the jacks or to use an additional screw to fix the soldered contact (sorry, don't know the matching words :-/ ).