Yes... it "seems" bomb proof, but I have seen a several LM3886 chips with the the full rail voltage (+ or -) at the output. I don't know neither the origin of the failure nor the OP design because I just measured them after the failure, but it is a real risk if you don't protect your loudspeakers.
And of course, it has nothing to do with Tom's design... it is about the chip
And of course, it has nothing to do with Tom's design... it is about the chip
Last edited:
Read here:
http://www.diyaudio.com/forums/chip-amps/257069-lm3886-full-negative-rail-output.html
AudioKarma.org Home Audio Stereo Discussion Forums - View Single Post - Question about LM3886 based amp
And remember: If you have fuses in each rail and one of them blows, the output will probably get stuck against the remaining rail.
http://www.diyaudio.com/forums/chip-amps/257069-lm3886-full-negative-rail-output.html
AudioKarma.org Home Audio Stereo Discussion Forums - View Single Post - Question about LM3886 based amp
And remember: If you have fuses in each rail and one of them blows, the output will probably get stuck against the remaining rail.
First link was a poor implementation with no output zoebel network that was likely oscillating madly.
Second link was just some opinion with no data on the failure modes but related to PA applications so may be the same issue as the first link.
I remain unconvinced there is an issue in a domestic environment with Tom's design. And one of you posted links agrees with me on that.
Second link was just some opinion with no data on the failure modes but related to PA applications so may be the same issue as the first link.
I remain unconvinced there is an issue in a domestic environment with Tom's design. And one of you posted links agrees with me on that.
Once the chip has blown there is nothing you can do to find out the cause of the failure other than to put another chip and try to replicate a mostly unknown situation.
Anyway, you are free to do whatever you want. My comment was in reply to the OP who posted the related question.
And I insist: if somebody cares a little about his speakers it is a good idea to protect them, even in domestic or PA applications.
And I insist again: there is no fault in Tom's design. It is all about the chip..any chipamp really...
Anyway, you are free to do whatever you want. My comment was in reply to the OP who posted the related question.
And I insist: if somebody cares a little about his speakers it is a good idea to protect them, even in domestic or PA applications.
And I insist again: there is no fault in Tom's design. It is all about the chip..any chipamp really...
Bonjour Dominique,
I was preparing to ask the question on your behalf, further to our exchange on Melaudia, but I see that you did it yourself!
This is also my understanding that the OPA277 brings a kind of protection. Tom will confirm.
I will post a photo of the amplifier later today, with the enclosure now completly finished.
Edit: i have not seen the last messages before posting, sorry, the situation looks clearer now. It is just a matter a failure mode and probability. If we get 10-9, I am happy!
Cheers,
Dominique T
I was preparing to ask the question on your behalf, further to our exchange on Melaudia, but I see that you did it yourself!
This is also my understanding that the OPA277 brings a kind of protection. Tom will confirm.
I will post a photo of the amplifier later today, with the enclosure now completly finished.
Edit: i have not seen the last messages before posting, sorry, the situation looks clearer now. It is just a matter a failure mode and probability. If we get 10-9, I am happy!
Cheers,
Dominique T
Last edited:
Correct. The DC servo will also to some extent correct for DC offsets elsewhere in the signal path leading up to the power amp. The DC servo in MOD86 Rev. 2.0 can correct for up to about ±100 mV of DC offset.
If the LM3886 fails and one of its output devices shorts to the supply rail, your speaker will see the full rail voltage. The odds of an LM3886 failure are extremely low, but not zero. If you want to protect your speakers from an LM3886 failure, you'll have to add a protection circuit. I haven't looked real hard, but it may be a challenge to find a protection circuit that protects your speakers well without impacting the performance of the MOD86.
If you do go the route of protection circuits, I also suggest that you buy or design a circuit which turns off the amplifier in the event of a failure.
~Tom
Last edited:
Hi Dominique,
I have taken time to read the entire thread and I agree with the reply of ezavalla.
DC servo only insures a proper working (i.e. offset = 0) and is not intended to protect loudspeaker from DC due to a failure in the circuit.
And once again, most of the LM3886 chinese cards are fitted with protection chip and relays against failure or misworking.
Tom's reply is wellcome.
I have taken time to read the entire thread and I agree with the reply of ezavalla.
DC servo only insures a proper working (i.e. offset = 0) and is not intended to protect loudspeaker from DC due to a failure in the circuit.
And once again, most of the LM3886 chinese cards are fitted with protection chip and relays against failure or misworking.
Tom's reply is wellcome.
Hi Tom,
- Did you heard about the Upc1237 protection chip and what do you think about its implementation in the project ?
- Did you heard about the Upc1237 protection chip and what do you think about its implementation in the project ?
MTBF depends also on environmental parameters.
- I presume that DIY decreases drasticaly MTBF...
I'll repeat my advice:
A DC coupled amplifier should have DC detection and detection triggered protection for the speakers.
The detection could come in several stages:
a.)
detecting the output offset of the servo opamp. When this exceeds the set value, you get a warning (light?) to tell you there is an error that exceeds your set level.
b.)
detecting the output offset of the servo opamp just before it reaches opamp's supply rails. As this set point is exceeded, the input to the amplifier is muted and a warning of why the amp has gone quiet.
c.)
detecting amplifier's output offset. As this set level & time duration is exceeded, either the supply rails are interrupted, or the output line is interrupted. Both these need a solid state (SS) relay, since the DC error current can be very difficult to break and for the mechanical relay to survive, or even operate as a DC current break.
There is a Thread discussing the SS relay.
A DC coupled amplifier should have DC detection and detection triggered protection for the speakers.
The detection could come in several stages:
a.)
detecting the output offset of the servo opamp. When this exceeds the set value, you get a warning (light?) to tell you there is an error that exceeds your set level.
b.)
detecting the output offset of the servo opamp just before it reaches opamp's supply rails. As this set point is exceeded, the input to the amplifier is muted and a warning of why the amp has gone quiet.
c.)
detecting amplifier's output offset. As this set level & time duration is exceeded, either the supply rails are interrupted, or the output line is interrupted. Both these need a solid state (SS) relay, since the DC error current can be very difficult to break and for the mechanical relay to survive, or even operate as a DC current break.
There is a Thread discussing the SS relay.
Last edited:
I still remain unconvinced that there is a failure mode I need to worry about. The only failure that any analysis has been done on it a poorly implemented design that lead to oscillation. If I was building something for the road crew to use I might think differently. A DC rail failure would be costly for me as well, but as I control every step of my chain I am comfortable that my FMEA is good enough.
Problems will not occur necessarily only during DIY workshop, but also during set-up of you listening system due to a unwanted event on input signal.
what sort of unwanted event? Find me a failure case that I have not considered. My FMEA is a little rusty but I used to do some work on mission critical stuff so have a very good feel on what is likely and what is not. So far the most likely cause of failure is one of my kids sticking their finger through a ribbon, and in 18 years that hasn't happened.
If the MTBFs are >10 years I can live with that, as tells me I will need to upgrade.
If the MTBFs are >10 years I can live with that, as tells me I will need to upgrade.