Non standard Op-amp configuration

[opty]

Hello,

In most schematics I have seen op-amps supply lines are directly connected to power sources, where opa-amp's output carries signal.
Over time, however, I came across cases where op-amp is working in different way: output is tied to ground and op-amps power supply inputs (through resistors) pull-up / pull-down external transistors.

one example http://www.next.gr/audio/amplifiers/audio-power-amplifier-l13579.html

Could you please help me find out:
- how is this op-amp setup/configuration called
- where I can find more information about it
- advantages/disadvantages ?

Well, it is really rarely used so there must be more minuses than pluses but just out of curiosity I would like to learn more.

[Kleinstein]

The advantage of this circuit is, that one can get a rather high power amplifier with a rather simple circuit. Also a rather low voltage OP can be used to generate a rather high output voltage. Further the output can go rather close to the rails - so not much voltage lost.

On the down side, stability can be tricky  - it really depends on the type of OP and the load impedance, if the circuit is well behaved or oscillating.  So you can not easly excange the NE5534 with another OP - it really depends on the internals of the OP whether it works well or oscillates. Datasheets usually don't specify the chips for this use - so you have to try out which chip works and which does not or needs some tweeks. Also higher frequency PSRR is likely poor.

[tron9000]

This is a push-pull amp, but its not a config I've seen before but its easy to figure out.

Because the output of the op-amp is grounded, op-amp has essentially become a constant current driver, where the current being sunk to ground is proportional to the voltage into the non-inverting input of the op-amp.

The current is being pull through the op-amp from the power supply rails, which has a 130OHM resistor in series. These resistors are also across the Base-Emitter junctions of the two TIP transistors. So essentially you have to pick the resistor value so the current being pulled through the op-amp supply lines is within the linear region for a V_be of the TIP transistors.

So transistor theory will tell you that for a small change in V_be, you get a current flow, I_c+I_b into the speaker.

that's as best and brief as I can describe, if you don't know any transistor theory, I'd do some homework.

as stated before me, its not stable, but you can get a lot of power.

[Performa01]

- how is this op-amp setup/configuration called

The unusual part of this circuit is known as "Current Dumping" output stage.

[rjeberhardt]

The "current dumping" output stage patented by Quad in the 1970s used feedforward to (almost) eliminate crossover distortion in a class B amplifier.  They achieved something like 0.005% THD.  I don't see any similarity with the circuit discussed here.

Russell.

[Zero999]

You need to look at the internal schematic of the op-amp to understand what's going on.

It needs to be biased so the quiescent current drawn by the op-amp turns on both of the transistors slightly, otherwise there will be crossover distortion.

As everyone has stated above, stability will be a problem, not only due to the common emitter amplifier on the output stage but because the NE5534 is decompensated and can oscillate anyway when configured at a low gain.

I'm not sure if constant current drive results in better or worse sound quality. Normally in an audio amplifier, the output impedance needs to be as low as possible to achieve a high damping factor but this circuit, having a constant current output will have a high output impedance so there could be sharp peaks at the resonant frequency. On the other hand, the feedback loop should fix the output current fairly firmly which should get the cone to move to where it needs to be. Hopefully someone with more experience of designing audio amplifiers will know more about this.

[rjeberhardt]

I'm not sure if constant current drive results in better or worse sound quality. Normally in an audio amplifier, the output impedance needs to be as low as possible to achieve a high damping factor but this circuit, having a constant current output will have a high output impedance so there could be sharp peaks at the resonant frequency. On the other hand, the feedback loop should fix the output current fairly firmly which should get the cone to move to where it needs to be. Hopefully someone with more experience of designing audio amplifiers will know more about this.

Definitely worse quality.

The feedback coming from the resistor in series with the speaker will give a very high output impedance resulting in a completely undamped speaker resonance.  A low output impedance is essential for a smooth frequency response.   Of course some people like booming bass rather than realistic reproduction 

Russell.

[Simon]

Of course some people like booming bass rather than realistic reproduction 

Russell.

You mean the sort of people you don't call the council on because you know next thing your car will be burnt out 

[dom0]

This is basically one of two push-pull circuits to get higher power from an op amp. This one has fewer parts and worse dynamic characteristics than the other (AB-biased buffer, diamond buffer).

[Zero999]

For anyone who's struggling to understand how it works. I've simulated it in LTSpice with a basic op-amp made from discrete components. I've changed it from the current control to a non-inverting amplifier with a gain of 100. It isn't compensated so will probably oscillate if you built it in real life but it simulates fine and shows you what's going on inside the IC.

[opty]

For anyone who's struggling to understand how it works. I've simulated it in LTSpice with a basic op-amp made from discrete components.

That is great. While I understand the fundamental idea behind this circuit (I hope ) I couldn't simulate it in LTSpice with op-amp models I have in my library. This simple model really works. Thank you.

[rjeberhardt]

That's all very well but the speaker isn't an 8 ohm resistor.  It looks something like this:

The current drive amplifier will give an big peak at the bass resonant frequency due to the Les Ces combination.

Russell.

[babysitter]

If I ever start a audiophool supply company, I gonna call it Les Ces.

[f5r5e5d]

few free op amp models get power pin currents right - and even those often don't get power pin V variation coupling to internal compensation right

Spice op amp models really shouldn't use node 0 internally

really annoying is that AD, TI have decades old whitepapers explaining the issues, show better op amp macromodeling techniques - and they still release the flawed models for new parts

[dom0]

Op amp models are sometimes not very useful anyway, since they usually have specs far exceeding even their real typical specs, especially when it comes to input currents and offset. Output current and slew rate, too. Basic linear parameters (frequency response) are usually about right - if they weren't the model wouldn't be good for literally nothing.

[tszaboo]

The opamp/resistor transistor configuration is called "opamp bootsrap" or bootstrapping. It is used even by Linear tech, in their reference designs, so it is not all that bad.
The benefit is that you can bias and control two transistors by one opamp, and it is cheap. But the feedback on this circuit is a little bit messed up. Speakers should be driven by voltage source, and this is driven by a current source.

[Zero999]

No it's not bootstrapping. The output stage inside the op-amp (Q1 and Q2 in my schematic) is configured as two common emitter amplifiers rather than the usual complementary emitter follower. Common emitter amplifiers invert the signal and move it to the opposite supply rail so the output now appears across R8 and R9 and another pair of common emitter amplifiers (Q11 and Q12) invert the signal again to drive the load.

[Zero999]

I find this configuration interesting so thought I'd revisit it.

In the original schematic it will operate in class B so there will be crossover distortion. The NE5534 only draws 4mA so the quiescent voltage across te 130R resistors will only be 520mV which will mean they're both turned off. The eliminate crossover distortion the resistor values needs to be increased to something more sensible such as 220R or more.

Like many people I found it won't simulate in LTSpice with the standard models so I made my own op-amp. I have found the LT1037 model works which is imiular to the OP37 but more expensive than the NE5534.

With a gain of 100 it does oscillate but this can be avoided by reducing the gain of the output stage by adding R8 which introduces negative feedback.