Class AB amp using TIP102. Seeking Critique

[Gusser]

Yes. This design is primarily meant for use with a single-rail 48Vdc switching supply.

I'm not sure, but I think I could reverse the design for a positive ground if needed by swapping all the NPNs for PNPs and flipping all the capacitor polarities. I don't even think I would need to change my board layout.

Actually it's easier than that!

Just connect the input and speaker ground to the VCC rail which will now be grounded and the amp gnd is now -48v.

Remember that the audio signal is AC. And the power supply is ideally a dead short at AC. So it doesn't matter if the input and output grounds are connected to the amp gnd or VCC in a single ended design. Of course you can't have the gnd buss of the PC board ties to a metal chassis in this case.

 

[jlny]

Right well there's that too, lol. What we call ground and +48V or -48V is pretty much a matter of semantics, but I see your point. I didn't realize that enterprise equipment considered it -48V.

 

[Gusser]

I didn't realize that enterprise equipment considered it -48V.

There's a very interesting reason for that. It has to do with the way metal conduits corrode.

Here's a good quote I found:

If two metal objects which are in contact with water are electrically connected to different potentials, material will tend to be removed from the one which is more positive and deposited on the one which is more negative. Having a small amount of material removed from a grounding stake and deposited on some buried wires is apt to be much less of a problem than having material removed from the wires. Using AC instead of DC would avoid electrically-caused corrosion, but having a DC bias in the "right direction" helps avoid even the corrosion that would occur without current flow

They figured this out in the early 1900s I believe.

 

[bimo]

You can add Miller Input Compensation (MIC), a capacitor from VAS's colector to inverting input of LTP. You can use around 5pF. The you can reduce the value of Miller compensation. It will increase the slew rate.

 

[indianajo]

That would be 5 pf from the collector of Q12 to the base of Q6? If I understand the functions correctly. Then decrease C5 to 82 or 68 pf?
I read somewhere that 7 pf is an inch of two 30 ga wires twisted together. So no need to buy anything IMHO. Twist 3/4".

 

[bimo]

That would be 5 pf from the collector of Q12 to the base of Q6? If I understand the functions correctly.

Yes. For correct value of C5, use simulator to get decent phase margin and gain margin. Or if you can not use simulator, please decrease it slowly, try 680pf, 470pf, 330pf, etc. and watch sign of instability (oscillation).