Hello there!

I have been working over high-precision mixed-signal devices, and have seen this particular requirement fulfilled through both the methods with equal frequencies/occurrences. To be frank, I am still not very much clear about the differences between a bipolar supply (with a 0V ground) and a unipolar supply with a mid-point ground!

Why would anyone even consider the required overhead in generating a bipolar supply from a unipolar input? That increases the overall component count as well as the price of the system, and certainly does not helps with noise suppression and stability!

But still, I see a lot of systems developed with the same technique, even some of the portable devices and wearable gadgets.

I would be really grateful if someone could present a comparative analysis of the two methods of designing the power supply of a mixed-signal signal.


With best regards,
Abbas.

 

How do you create the midpoint ground? How mich current can it source and sink without changing its voltage?

I am not saying you cannot use such a ground, normally called a virtual ground, but that it is not necessarily good for all situations.

Bob

 

It's a trade-off between the cost and parts count of a dual-supply versus that required to generate a mid-point ground from a single supply.
The mid-point ground is often used when there is already a single supply (such as a wall-wort) used for the circuitry requiring most of the current (generally the digital and processor circuits) and there's only a small portion of the circuitry (likely analog) that requires dual voltages.

 

"Biasing and Decoupling Op Amps in Single Supply Applications" by Charles Kitchin
Analog Devices Application Note AN-581 (2002)

This is probably still available on the internet, but I can't say for sure. I have yet to read this , but I believe it has to do with showing how the virtual ground ps can cause problems in some cases for circuits with op amps.

-Pete

 

How do you create the midpoint ground? How mich current can it source and sink without changing its voltage?

We can use any of the primitive techniques available for doing this, like using a 1:1 potential divider and then buffering the output, or use a feed-back LDO programmed to deliver output at half of its input.

I am not saying you cannot use such a ground, normally called a virtual ground, but that it is not necessarily good for all situations.

I would be grateful if you could elaborate this, especially the last phrase, with some examples.

Thanks a lot, BobTPH.

 

It's a trade-off between the cost and parts count of a dual-supply versus that required to generate a mid-point ground from a single supply.
The mid-point ground is often used when there is already a single supply (such as a wall-wort) used for the circuitry requiring most of the current (generally the digital and processor circuits) and there's only a small portion of the circuitry (likely analog) that requires dual voltages.

Thanks a lot, crutschow.

I would be grateful if you could lead me to some sources with elaborate description of this, with concrete stats, examples and observations.

 

"Biasing and Decoupling Op Amps in Single Supply Applications" by Charles Kitchin
Analog Devices Application Note AN-581 (2002)

This is probably still available on the internet, but I can't say for sure. I have yet to read this , but I believe it has to do with showing how the virtual ground ps can cause problems in some cases for circuits with op amps.

-Pete

Thanks a lot, PeteHL.

 

https://www.analog.com/media/en/technical-documentation/application-notes/AN-581.pdf

http://www.ti.com/lit/an/sloa030a/sloa030a.pdf

Above link to prior reference AD ap note and a TI ap note.

One other consideration on virtual ground, if buffered you now have
a ground that is subject to OpAmp slew rate, noise, offset, drift....
And of course as OpAmp runs out of G at frequency the output Z
starts looking inductive. Thats an analysis you can do in s plane.

Regards, Dana

 

An example of where you would not want to do it is a bipolar motor control.

If the midpoint ground is only for signal levels, I have used a divider followed by op amp buffer successfully.

Bob

 

Thanks a lot danadak and BobTPH.

http://www.ti.com/lit/an/sloa030a/sloa030a.pdf

I am not feeling very confident about this article. In this article, everytime a reference voltage is used to bias an op-amp, the reference is attenuated by a ratio equal to the gain of the op-amp circuitry, before it is fed into the op-amp. Is that really correct? An op-amp amplifier configuration amplifies the voltage difference between the two input terminals. Since the reference is common to both the inputs, what is the need to attenuate the reference voltage before feeding it into the reference terminal?

 

Yes, the Vref is gained up by the OpAmp, in other words Vref looks like
a "normal" signal. This is because the circuits shown are creating a
offset to maximize input/output dynamic range of the circuit. So the
offset is created "input referred".

Regards, Dana.

 

Thanks, danadak.

This is because the circuits shown are creating a
offset to maximize input/output dynamic range of the circuit. So the
offset is created "input referred".

Yet, I do not think I understand it clearly!
Though they are creating an offset, it is common to both the terminals and so, has to be rejected automatically!

Regards,
Abbas.

 

I remember reading an article sometime ago that used an audio amplifier ic as used in tv's etc as a rail splitter that could supply high current with good decoupling.
Maybe someone on here may know where to find more info about it.

 

Voffset is not CM to both inputs, otherwise it would result in no change in the output.,
to within the CMRR of the OpAmp.

As you can see in this inverting example -



Here is the way CMRR is setup and tested.

https://www.analog.com/media/en/training-seminars/tutorials/MT-042.pdf


Regards, Dana.

 

@danadak You are showing the non-inverting amp configuration in your post #14, aren't you? Also I believe that usually in the non-inverting configuration, the reference voltage is applied to the non-inverting input terminal of the op amp. See Fig. 1 on the first page of AN-581.

 

That's just one of several cases in the TI ap note.

Regards, Dana.

 

That's just one of several cases in the TI ap note.

Regards, Dana.

Alright, thank you, I will have to look at the TI ap note.

 

Voffset is not CM to both inputs, otherwise it would result in no change in the output.,
to within the CMRR of the OpAmp.

OOPS! My bad!

Thanks a lot, danadak.