Bug on Op Amp LM358 10x non-inverting amplification with Virtual Ground

[ruiseixas]

Hi,

I'm trying to avoid any bias tunning with a LM358 throughout a Virtual Ground in a single Power Supply. So the idea is to supply +10v and split by 2 to have +5V and -5v supplying the Op Amp LM358 with the help of the second Op Amp in the same Chip.

The data is:

Code: [Select]

Power Supply:
Vcc+ = +10v;
Virtual Ground = +5v;
Vcc- = 0v.
Function generator:
Amplitude = 0.30v;
Frequency = 10kHz.
In the Proteus Simulator (A try the LTSpice first but it hasn't the LM358 Op Amp!), where U1:B serves to create the Virtual Ground and the U1:A does the amplification. On the simulator everything looks nice like this:



However, in the real world this is the result after trying with two different LM358 chips:



Is this due to some sort of polarity transition on the chip, does this means that I absolutely need a cap in the output. This is odd because with the traditional inverting amplifier I haven't this kind of output shifting!

In the breadboard I always used the Virtual Ground as ground when connecting the oscilloscope and function generator. I also used a portable multimeter to see any changes on the +5v of the Virtual Ground with no changes at all seen!

Any ideas? 

[Kalvin]

Any ideas? 

- Add a 10uF bypass capacitor between the +10V power supply (VCC+) and the negative power supply (VCC-).
- Add a 100nF ceramic bypass capacitor between the op amp V+ pin and the negative power supply (VCC-).
- Add a 1uF filtering/bypass capacitor between the U1:B non-inverting pin and the negative power supply (VCC-).

[Andy Watson]

Is this due to some sort of polarity transition on the chip,

The output stage of this op-amp is odd. This video gives a good explanation:

does this means that I absolutely need a cap in the output.

No. You need to either bias the output in one direction (as per video), or choose a different op-amp.

In the Proteus Simulator ...

The perils of simulation   You need to know what model Mr Proteus has used - in this case it looks like it is a generic op-amp that does not model the output peculiarity of the LM358. You should be able to download a more accurate model, possibly from TI, and incorporate it in your spice program.

[ruiseixas]

No. You need to either bias the output in one direction (as per video), or choose a different op-amp.

If you have to bias the signal I don't see any advantage between an Op Amp and a Transistor to amplify a given signal!

In the video he speaks about with and without a load, maybe with load the crossover distortion isn't an issue...

Thanks

[mikerj]

If you have to bias the signal I don't see any advantage between an Op Amp and a Transistor to amplify a given signal!

You can't perform proper op-amp functions with a single transistor, e.g. sum & difference.  The LM358 has a push pull output stage that will both sink and source current with negligible quiescent current, something not true of a single transistor amplifier.   This op-amp works perfectly well for most DC or low frequency applications where the slew rate is sufficient to compensate for crossover distortion.  It's only when you need higher frequencies that you need to resort to tricks like biasing the output into Class A, or alternatively you could use an amplifier more suited to the application (e.g. the LMV358 is almost the same apart from the output stage).

In the video he speaks about with and without a load, maybe with load the crossover distortion isn't an issue...

Only if you have a DC coupled load and one end of the load is connected to the negative rail (or zero volt in a single rail circuit).  The load will then bias the output into Class A.

For an AC coupled load, or a load returned to 0v on a split rail supply, then crossover distortion won't magically fix itself by loading the output more.

[Audioguru]

I agree that you should use a newer and better opamp.