How to parallel multiple power amps?

[kokoriantz]

I selected thumb up, yes I got it on the post but not the thread. It is assigned to my opinion by the moderator.

[RussellKinder]

The short answer is this is not the way.

Ultimately, this would work kinda OK, but you have to delete Rin, because it is positive feedback across the amps 2, 3, and 4. Ultimately, just put all 4 amps in parallel with 0.22 in series with each output. You can even put the 0.22 ohm resistors inside the feedback.

This can work very oddly, however. Each amplifier is loaded by the other amplifiers, so any mismatch between amps can strain them all, and possibly cause distortion and or small-signal instabilities. You won't see that in the sims, because all the amps are perfectly matched. You have to introduce differences in offset, bandwidth, gain, etc.

Originally Posted by kokoriantz

Each paralleled amp equates the voltage of its resistor to be equal to the master's resistor. By this the output current is multiplied here by four. the main amp sees the load as Rload×4. The load sees the impedance of the main amp as Z/4.
Hayk

[kokoriantz]

It is the Rin that transforms the voltage low impedance amplifiers to become high impedance current multipliers. This is the novelty. Of course you can always get ride of your engine and attach a pair of horses to drive your car, if it is the way things should do.

[Kay Pirinha]

If you swap the LM1875's for TDA7293's, you could parallel two or more of them even without the need of current balancing resistors. Have a look at the respective datasheet.

Best regards!

[RussellKinder]

Originally Posted by kokoriantz

It is the Rin that transforms the voltage low impedance amplifiers to become high impedance current multipliers. This is the novelty. Of course you can always get ride of your engine and attach a pair of horses to drive your car, if it is the way things should do.

It is the Rin that will cause this little experiment to melt into a puddle under either clamping to a supply rail, or rail-to-rail oscillation.

As I said, this is not the way.

[kokoriantz]

The LM1875 for best
See post 106 Bode curves.

[JMFahey]

Originally Posted by kokoriantz

"How to parallel multiple power amps?"
....
Hayk

Simple answers:

1) don´t.
You simply do NOT put voltage sources in parallel.
Trying to do so demands kludgy solutions.

2) What for?
Very poor outcome for increasingly complex design.

A single chipamp is a thing of beauty, created by GENIUS Italian designer Bruno Murari , who invented the now common (for us) TO220 5 pin chipamp.
It was revoutionary way back then and still a very valid concept today .

Now, once you start parallelling them, power increase does not justify added complexity and cost.

Practical example with current Mouser prices:

* Single chip amp: 20W for $3.15 plus a few other parts, simple, stable , compact, clean, what´s not to like?

* 4 chipamps , bridged parallel for 80W into same impedance , kludgy, have to be very careful so parallel outputs don´t suicide fighting each other, 5 or 6 times as complex (not just 4X because you have to add *precision* current balancing elements) , 4 devices to bolt to heatsink, 4 micas plus greasing them or Silpads, more complex PCB because chips must touch heatsink, now you spend $12.60 on "actives" plus a couple bucks on extra passives ..... it quickly loses its charm.

* compare that to discrete alternative: TIP142/147 pair can easily provide 80W into 8 or 4 ohm, needs only 2 wirewound resistors, can be as stable and protected as designer skills allow, only 2 devices to bolt on heatsink (6 legs in total instead of 20), all for $5.50 in semiconductors plus a few assorted parts.

* or a chipamp alternative: LM3886 , good for 60W at $6.82 or TDA7294, good for 80W at $3.18 , 11 or 15 chip pins to solder, still less than 20

My point?: chipamps are great if used as intended, not that much if you make them play out of their league.