LM3886 GC - "Simplest Ever Amplifier Bridging"

[jackinnj]

the LM4780 is basically a pair of LM3886's -- the design by National Semi on the LM4780 PDF works very nicely indeed if you have to bridge -- I see no reason that a pair of LM3886's shouldnt do as nicely providing that they are properly heatsinked.

[theChris]

yeah, i'm not conviced the distortion would cancel as opposed to summing. if the distortion was in phase on both amplifers then yes, but as it stands, i'm not conviced this will happen.

[pinkmouse]

OK, this sounds like an ideal job for someone with one of those spice simulator things. Sim the two designs, (they can be any old op amps in the sim, it's the principal of the thing that matters ), and see how the two distortion figures come out. Or, of course, someone with the two amp configurations could just measure them I suppose!

[arcom]

I think that this is the common bridging technique for most chip amps. I have used it in my TDA7294 amp and it works great!

[jackinnj]

in theory, it works perfectly -- i put it on Multisim8 and there is no distortion to speak of:

in the real world I have found distortion of 0.01 to 0.07% with the capacitor coupled, bridged LM4780

[pinkmouse]

Originally posted by jackinnj
iin the real world I have found distortion of 0.01 to 0.07% with the capacitor coupled, bridged LM4780

I wonder if one of the differences between sim and real life comes down to tolerance in other components between the chips. Could this be reduced by more exact gain matching I wonder?

[jackinnj]

Originally posted by pinkmouse


I wonder if one of the differences between sim and real life comes down to tolerance in other components between the chips. Could this be reduced by more exact gain matching I wonder?

I guess I should put up the picture -- with the resistors matched to 0.1% there is a slight difference in gain between the A and B channels -- Gain on A is 19.67dB, and on B it is 19.77dB. I am averaging 64 samples of a 1kHz signal.

[ingrast]

Originally posted by jackinnj


I guess I should put up the picture -- with the resistors matched to 0.1% there is a slight difference in gain between the A and B channels -- Gain on A is 19.67dB, and on B it is 19.77dB. I am averaging 64 samples of a 1kHz signal.

Gain matching is not an issue in a bridge configuration except for efficiency (max available output at exact matching). If one side has 100% gain and the other 50%, then the output voltage will be 150% compared with the single ended case.

Below 0.01% THD, other issues like PSU and layout are much more relevant.

Rodolfo

[pinkmouse]

Thanks for clearing that up Rudolfo, little tips like that are often the things you miss when you are teaching yourself.

[TwoSpoons]

The problem with the sim is that the model generally does not include the non-linear elements that create the distortion. Op-amp macro models are pretty simplistic.