Bootstrapping LM3886

[xplod1236]

I want to bootstrap the LM3886 to run off +/- 75v rails.

The ground pin has to be 9v above the negative rail, but since the opamp's rails will be chaning with the signal, is it ok to connect the gnd pin to the output?

Where would the mute pin be connected? Would I connect it to the fixed supply negative rail?

What is common mode input voltage and differential input voltage?

[memorex]

are you designing for variable supply rails (like class-h) or rails that switch from 9volts to 70volts and back?

I'm not an expert with supply bootstraping for op amps, but my understanding is that you don't connect the output to ground. The mute pin would have to be connected to the -70volt supply rail. My understanding of common mode is when the signal connected to the non-inverting opamp input cancels with the signal connected to the inverting opamp input. My understanding of differential input is the difference in voltage between the non-inverting lead and the inverting lead of the opamp. Could someone confirm or correct me?

[xplod1236]

I'm designing for variable supply rails. But I thought that class H was switching between 2 different supply voltages.

As for the ground pin, after looking thru a document on bootstrapping and doing some simulations, I figured out that a voltage divider should work (resistor from v+ to gnd pin and resistor from v- to gnd pin). Can someone confirm this?

[memorex]

Class-G is switching the supply rails to different voltages. Class-H is a variable supply rail voltage.

What are you trying to do with this voltage divider circuit? The ground pin on the lm3886 op amps should be kept at ground.

The voltage divider circuit created by rf and rg in the document you posted is used with the inverting input and it sets the gain of the opamp.

[xplod1236]

The LM3886 had undervoltage protection. The only way to disable it is to keep the gnd pin 9v above the negative rail. With varying rails, the negative rail (-37v) would be varying, and going close to 0v (simulations). Then the gnd pin would not be 9v above v-, and the undervoltage protection would shut off the chip.

With this voltage divider, I'm trying to keep the gnd pin voltage halfway between the varying rails.

[memorex]

just bias the rail controlling transistors to +-9 volts. In reference to that document you posted, they are biased with R1 and R2 for the positive rail and R3 and R4 for the negative rail.

[xplod1236]

How about this...

[xplod1236]

Which transistors would be good for the supplies? They don't need to handle much current.

Which transistors would be good for the output?

[janneman]

xplod,

Your basic topology is sound, but a warning is in order. The way this is set up means a lot of extra (current) gain in the loop, also because you use a minimum (10x) gain overall. This is a strong invitation to instability, which will vary with the type of output transistors.

Also, the 100k input resistor (I can see where it comes from) just generates a lot of noise, and is unnessecary.

The resistor-diode string for the supply offset transistors is quite high impedance; remember that if the output swings toward a supply rail, the diodes still need enough current to be biased. Maybe a current source is better here.

In fact, if you want to use this topology, why use the LM3886? This is a power amp, and in this way, you just use it as a driver. There are good opamps that will give far better performance here and also avoid the instability problems.

Jan Didden

[xplod1236]

you use a minimum (10x) gain overall

Changed the gain to 25x.

Also, the 100k input resistor (I can see where it comes from) just generates a lot of noise, and is unnessecary.

Took that out.

The resistor-diode string for the supply offset transistors is quite high impedance; remember that if the output swings toward a supply rail, the diodes still need enough current to be biased. Maybe a current source is better here.

What do you mean by current source? (What is a current source?)

In fact, if you want to use this topology, why use the LM3886?

I have 20 of them (free samples--I was surprised that National sent me 4 packs of 5 of them in a row) This will be used as part of a bridgeable sub amp, so performance doesn't matter that much. And it will be easier to do with the LM3886, because I can keep the rails at +/- 35v, instead of +/- 15 as with a regular opamp. (A regular opamp wouldn't swing to 75v, would it?)