LM3875 murder my speaker

[Ipanema]

Guys,

I have a LM3875TF build up with my own PCB. It works fine for few months. But out sudden, the right channel chip just went dead. No physical damage or burn mark on the chip. Worst thing, the dead chip also output +32V to my speaker, which murdered it as well. My supply voltage is +/-34V.

Does anyone experience this before? Do you know why this happen? There is just no forewarning to this failure mechanism. I have cried till all my tears is gone.

Thanks.

[ozonek]

Maybe need failure analysis on the silicon die to find out the failure mechanism...

[Ipanema]

Hi Ozonek,

How to do that? Can national Semi help out in this forensik investigation? Need more evidence to charge this murderer.

Thanks.

[carlmart]

Well, that's the way it is. Even commercial equipment used to blow speakers in the past.

Nowadays they use a DC protection circuit, that senses any danger and puts a relay in the way to avoid destroying the speaker.

There's absolutely no way you can claim anything here. Sorry. That's how things are.

Get yourself a DC protection kit and use it from now on. Or take a chance and get along, as most people do in this hobby.

[ashok]

A short term voltage surge might have raised power supply voltages enough to kill the chip . Maybe . Both chips can have different absolute Vmax before conking out.
There could be many other causes of course.
Using a relay to protect the speaker is essential. Not good for sound ( debatable ) but certainly a very great feeling when the chip dies and your speaker lives another day !

My LM3886 has relay protection . I thought a lot about this and finally added it on a separate board .

Strangely I also have a TDA2050 chip amp and one day I accidentally reversed the power supply leads from the transformer . There was a loud crack and snapping noises and the two chips ( channels ) exploded but nothing happened to the speakers which were directly connected . The fuse was intact. I replaced the chips and everything worked normally.

Cheers.

[Ipanema]

Thanks for sharing your experience here.

Agree that DC protection helps here as an insurance. I just don't bother previously because most of the guys here who build GC does not care about this. I only saw Mauro Penasa design using it. A lesson learnt in a hard way. :-(

But I'm still very intested to investigate further on why this happen. I also have an experience , where the damage chip just exploded/crack open when I touch the back of the chip to GND. This is understandable.

For this one, I cannot see any physical damage to the chip. It is just like a time bomb that exploded when I least expected it to happen. Most GC builder here pls take note of this. It might happen to you as well. That's why I would like to investigate further on why this happen. Pls help to list down your experience or possibility that the chip might fail this way :

1) voltage surge (Ashok)
2)

Regards.

[ashok]

If you have a 2 way speaker the tweeter should be safe.
The woofer should be inspected to see what is wrong.
Fully burnt coil ( it will smell!) or something else like torn off lead in wire.
You could have a rewound coil which would work well though not as good as the original ( maybe ).
I have heard woofers that had their coils wound locally ( at a screw driver shop !) and I couldn't say if they sounded any different. They would probably measure different from the stock driver , but does it really matter now ?
Make sure the spider and surround are removed carefully by using the right solvent . Remember to desolder the leadin wires from the tags on the frame before attempting to remove the cone assembly.
Will your speaker manufacturer sell you a spare cone coil assembly ? Some companies do . Cheers.

[AndrewT]

Hi Ip,
did you have supply rail fuses after the main smoothing caps?
These would normally blow if the output stage shorts a rail to the speaker.

I have found that you can get full power to the speaker without the fuses blowing even when the output rms current is two times the fuse rating. I have even managed three times the fuse rating and not blown the fuses.

Using your 32Vpk and 8ohm speakers indicates that the fuses could be F1.6A on each supply rail (if you blow these too regularly then try F2A fuses for 8ohm speakers, you might get away with F1.25A).
They will blow within a very few seconds if the output goes to 32V into the speaker DC resistance and should provide sufficient protection.

To help blow the fuses more quickly the DC detect could operate a crowbar to force the Vrail to ground and blow the fuses within a few hundred mS.

[justblair]

Originally posted by Ipanema
Guys,

I have a LM3875TF build up with my own PCB. It works fine for few months. But out sudden, the right channel chip just went dead. No physical damage or burn mark on the chip. Worst thing, the dead chip also output +32V to my speaker, which murdered it as well. My supply voltage is +/-34V.

Does anyone experience this before? Do you know why this happen? There is just no forewarning to this failure mechanism. I have cried till all my tears is gone.

Thanks.

I have had a similar experiance with an Lm3875.

Except on mine, the voltage in went straight to the output and to ground burning out a rc filter there (well several filters). The only damage I recieved was a couple of pounds worth of resistors/caps.

My PS was fused, however the fuses did not blow, The rc network took it instead.

It was my first amp build, and it confused the hell out of me, I was measuring 0dc output from the amp yet it didn't work. Other chips on the same power supply were going haywire.

When I removed the lm3875, it had blown its package apart.

[I_Forgot]

Originally posted by Ipanema
I have a LM3875TF build up with my own PCB. It works fine for few months. But out sudden, the right channel chip just went dead. No physical damage or burn mark on the chip. Worst thing, the dead chip also output +32V to my speaker, which murdered it as well. My supply voltage is +/-34V.

How do you know the 32VDC is what killed the speaker? The chip may be shorting one rail to output now, but it is a damaged chip. What caused the damage to the chip? Does your amp have some stability protection/BW limiting or did you leave out those parts because they don't "sound good"? If the amp was oscillating it could have killed the speaker then burned up the chip. Power transistors die when they are driven too hard- i.e. crazy volume with inadequate heatsink, oscillation, etc. When they die, they short C to E. The chip has a bunch of protection built in, but if it oscillates and overheats, all bets are off. Were you present when it killed the speaker? What did it sound like? Is the tweeter dead? Just woofer?

Voltage surge seems unlikely. The chip is OK up to 84V, you're running at 68V. Maybe lightning could make that big a surge, but you'd have a bunch of other dead equipment too.

I_F