LM3886 PCB layout questions

[zAAm]

I cover the mute circuit in some detail on my Taming the LM3886 - Grounding page.

The LM3886 has two input diff pairs. One is used during normal operation and one is use only during mute. The mute function itself is accomplished by a multiplexer that routes the bias current to either the input pair connected to the outside world (normal operation) or the mute pair (mute operation).
The Pin 7 ground is used as a reference ("input voltage") for the mute pair, reference for the multiplexer, and for a pair of clamp diodes. These diodes are not for ESD. They form a clamp that limits the differential voltage across the mute differential pair. This allows the mute pair to work faster. If the clamp wasn't there, the mute pair would likely have to recover from saturation (carrier recombination process ... sloooooowwwww) to mute or un-mute the IC.

Oh wow, I'm sorry. I really need to learn to read. Your article gives exactly the answer to my question and I ended up glossing over the muting part altogether the first time.

Therefore (after I've properly read your article ), it would probably be erring on the side of caution to route it to power ground, since there is still 0.5mA~1mA going through that loop. Even though that's still very low (and just DC), we might as well remove that from the signal ground...

Nice layout! You get plus points in my book for including Rf2, Cf, Cc, and both Zobel and Thiele networks from the start. Nice work.
The only room for improvement I see at this point is to move the power input connector to the left side of the board. Put the mute components elsewhere. This would allow a bit cleaner supply routing and decoupling. I don't know if you're emotionally tied to having all connectors along one edge of the board, though. You don't need two mute resistors, by the way. Rmute + Cmute is all you need. The layout is nice the way it is, though. We're approaching the nit-picking stage in the review here...

Thanks! I'll make a mock-up of the side mounted terminals to decide if I'm willing to go that far

In terms of the muting circuit, I used the one described in the LM4780 datasheet mostly since it will allow a turn on mute condition while Cm is charged. Just thought it would give some more time for the power rails to stabilize before un-muting the amplifier. Although, since I'm considering a delayed speaker connection circuit, this might be overkill...

 

[tomchr]

I've measured the amp output offset voltage vs time with various Cmute caps. You actually don't get any advantage of muting for longer. The amp recovery once it comes out of mute is the same regardless of the size of the mute cap (at least within the range of 100-1000 uF).

Some people like elaborate mute circuits. I guess it's just another way to geek out. There's really no practical need for anything more than an RC circuit. The resistor should be sized to ensure Imute > 500 uA under worst case operating conditions.

Tom

 

[zAAm]

I've measured the amp output offset voltage vs time with various Cmute caps. You actually don't get any advantage of muting for longer. The amp recovery once it comes out of mute is the same regardless of the size of the mute cap (at least within the range of 100-1000 uF).

Some people like elaborate mute circuits. I guess it's just another way to geek out. There's really no practical need for anything more than an RC circuit. The resistor should be sized to ensure Imute > 500 uA under worst case operating conditions.

Tom

Yeah, I suppose since the LM3886 circuit is pretty standardised people want to add their own bit of flair to it Btw, you made some kickass amps with your Modulus-86 and Parallel-86! Thinking of making a parallel version with the LM4780, but I think I'd probably need to add DC servos to match them. Will maybe add a DC servo to the LM3886 circuit later as well just for kicks.

Thanks for all the advice! Appreciate it.

 

[tomchr]

Btw, you made some kickass amps with your Modulus-86 and Parallel-86!

Thank you.

Thinking of making a parallel version with the LM4780, but I think I'd probably need to add DC servos to match them.

Getting two LM3886es or two channels of the LM4780 to play nicely and share the output current evenly is a bit of a challenge. The DC servo can muck up the THD performance at 20 Hz, so some design effort is needed there as well. Both are solvable problems, though. Good luck.

Tom