Hot LM4702

[Dxvideo]

Hi all,

I am just finished my last amplifier with LM4702B;

Everything is ok. The sound is excellent. Noise and distortion is uncomparable. I havent make a test with my big speakers and I cannot say something about on basses but mids and trebles are very detailed.
I am using Renesans 2SK1058/2SJ162 MOSFETs. And feeding with +/-55v snubberized (4 x MUR860 2 x 10,000uF+ 470nF + 1R2-100n) power suplly.
My problem is (I m not sure that if its a real problem or not) LM4702 is warming like crazy (about 65°C)! I didnot connect it a heatsink but as I know its unnecessary.
Am I wrong? Should I connect it to a heatsink?
Thanks in advance for helps.

Best regards,
Ozgur

[gmikol]

With 110V total rails, the 25mA supply current translates into about 2.75W dissipation. The data sheet has a junction to ambient resistance of 30 C/W. This translates into a temperature rise of 82.5C over ambient, so perhaps National is over-estimating the Theta-JA of this package.

With a Tj_max of 150C, you're well within the operating parameters of the chip. But even a 4 or 5 C/W passive CPU heatsink could make a big difference.

Greg

[Dxvideo]

I already put a heatsink behind her. And now shes under comfort. But thats interesting I didnt expect it will be warmed.
Does that affect the sound quality?
Thx

[AndrewT]

Hi,
Yes, it affects the sound quality.
There are numerous mechanisms inside that will deteriorate when temperatures rise.
A few will improve with increasing temperature, but the majority will degrade with increasing temperature to the point that some component will fail.

High temp will also activate the protection circuitry much earlier (if National have designed it correctly to take account of reducing SOA with higher temperature).

You MUST fit an adequate heatsink and for best sound I suggest you keep Tc below 50degC. This will require a sink that never exceeds 40degC under any and all operating conditions.

BTW,
National specify max Iq=30mA and that equates to 3.3W on +-55Vdc supplies.
Rth j-a =30C/W results in delta T of 99Cdeg.
Add on ambient conditions inside the case and the junction temperature is up at about 130degC before you even feed a signal through the chip.
You must read the datasheet before you start designing/procurement/construction.

[glennb]

Originally posted by Dxvideo
I already put a heatsink behind her. And now shes under comfort. But thats interesting I didnt expect it will be warmed.
Does that affect the sound quality? Thx

I have plotted the LM4702 temperature with a heatsink and without : http://home.pacific.net.au/~gnb/audi...ml#heatsinking

A small heatsink is definitely recommended. It won't overheat without a heatsink, but it gets damn hot. I don't know if it affects sound quality.

[Dxvideo]

I see.
As mentioned I noticed that in first trying. And surprised a lot. Because as my understanding it works like an opamp and if its not a power op-amp then it will never need a heatsink.
Anyway,
mathematic is an universal language and a simple calculation gives an exact result:
LM4702 needs a heatsink!
Thx my friends.

[Dxvideo]

Dear Glenn,

I want to ask a question about your design (which is on your pages);
You recommend that pin 7 should connect to power ground point not on signal gorund.
However until now; I connect the other power opamps (LM3886 like) gnd pin to signal ground.
Whats that for? Does any current is sinking from this pin? I am a bit confused.
Thx. again.

[glennb]

Originally posted by Dxvideo
Dear Glenn,
I want to ask a question about your design (which is on your pages); You recommend that pin 7 should connect to power ground point not on signal gorund.
However until now; I connect the other power opamps (LM3886 like) gnd pin to signal ground.
Whats that for? Does any current is sinking from this pin? I am a bit confused.
Thx. again.

You are referring to http://home.pacific.net.au/~gnb/audi...html#groundpin ?

My research has shown that LM4702 pin 7 (GND) is used for biasing and current sources which are internal to the chip. It does not carry any audio signal currents. Therefore in theory it should be directly connected to the 'power supply' star earth rather than the 'signal' clean earth point. I haven't actually compared the sound of connecting it to power supply or signal earths, there probably isn't any audible difference. I can't speak for the GND pin on the LM3886, 3875, 1875 etc because I haven't measured them.