Op-amp stacking

[PRR]

> are the Op-amps supposed to heat up when stacked?

As Oskar said: You aren't supposed to stack them. They WILL heat-up.

No two opamps have the same gain. Inside the same feedback loop, they will try to force slightly different output voltages. Being tied together, neither one can hit the voltage it wants to be at. This voltage difference forces a current to flow; ideally infinite though really limited by stray resistance and chip protections.

So they tend to run as hot as the designer thought was worst-case safe.

> a lowering of noise floor due to random cancellation

Hiss VOLTAGE or hiss CURRENT?

As R.G. said many months ago, paralleling "may change the source/load impedances in a manner which is a better fit..."

Voltage sources in parallel is NOT the same as adding voltages.

Parallel voltage sources load each other.

Put two car batteries in parallel. One 12.1V, one 12.8V. Clearly they do not sum to 24.9V, nor RMS to 17V. A huge current flows (hot jumper-cables just like those hot chips) and it settles somewhere near 12.5V.

Hiss currents add (RMS). But in practice there are other resistances being paralleled.

Flip a coin. You may get heads or tails. Flip a million coins. You will almost certainly get a 50:50 mix of heads and tails. The randomness averages away.

So the answer is: IT DEPENDS.

Look at LM394. This is/was two sets of _50_ parallel transistors. It gives lovely low hiss VOLTAGE. It also gives high hiss current.

The magic is not in the number of devices. It is the total cathode (emitter) area. There's reasons to keep transistors small and parallel a bunch. In tubes, for low hiss VOLTAGE, you use a large-cathode device like 6DJ8 instead of a skinny 12AX7. (And about the only place you have to go that far is FM/TV antenna stages.)

> If simply doubling the amplifying circuits in an opamp could reduce noise, all low noise opamps would already be double/quadruple/octuple/etc inside.

1) that would cost more. The LM393 die is as big as some quad amps.

2) Many-many-many applications are not noise-sensitive; why pay more?

3) 98% of applications work in 1K-50K noise resistances and do fine with single normal-size devices. Let the 2% users buy something else.

As Merlin says: nearly all audio circuits are scaled into the zone where single transistors work fine. The low-Z microphone is a marginal case (200 ohms is awful low) yet most mike preamps give low hiss with a single pair of large ('4401) transistors.

[Lurco]

http://www.elektor.com/magazines/2010/october/the-5532-opamplifier-part-1.1529745.lynkx

I have heard tight bass too when stackin semiconductors

http://www.diystompboxes.com/smfforum/index.php?topic=16784.msg731880#msg731880