..generate the most heated debates. Trick question, what is the common mode voltage of the op-amp buffers in the attachment?

1) Most op-amp literature tells you the common mode voltage is the average of the algebraic sum of V+ and V-. This is the same as saying the common mode voltage is the voltage on the V+ pin. Using this definition, the common mode voltage for both buffers is 7.5V.

2) If I told you the common mode voltage is referenced to (Vcc+Vee)/2 or otherwise the mid-point between the supplies, then the left buffer's common mode voltage is 7.5V since the mid-point is 0V and the right buffer's common mode voltage is 0V.

3) If I told you the common mode voltage is referenced to Vee then the left buffer's common mode voltage is 22.5V and the right buffer's common mode voltage is 7.5V.


I have used the first definition almost exclusively until I entered the semiconductor field. Many test and evaluation engineers (ATE/Bench) have used the second or third definition.

Do you agree or disagree?

 

Stop stirring up trouble, before I roundhouse-kick you into the year 573,526, where you will be instantly vaporized by our then ever-expanding sun.



Have a nice day.

 

Being a technician, I would say both op amps are running single-ended, so there is no common mode element present. I only worry about common mode voltages when trying to catch the difference voltage on a strain gauge bridge or something similar.

And you thought they were kidding when they said ignorance is bliss?

 

Being a technician, I would say both op amps are running single-ended, so there is no common mode element present. I only worry about common mode voltages when trying to catch the difference voltage on a strain gauge bridge or something similar.

And you thought they were kidding when they said ignorance is bliss?

The ultimate concern is the offset error associated with the common mode voltage. An op-amp datasheet will list a common-mode voltage range spec. Regardless of it is an AC or DC signal, one would operate somewhere within the range of the op-amp supplies (really the input differential pairs) and the offset changes depending on where one is in this range (common-mode range).

Given the first definition, both amps have a Vcm of 7.5V. If the op-amp had a CMRR of -60dB then that would be an additional 7.5mV of offset.

If Vcm is referenced at the mid-point between supplies then one amp has 7.5mV additional offset and the other 0mV.

If Vcm is referenced to Vee then one amp has 22.5mV additional offset and the other 7.5mV.

If I were a customer and wanted to calculate my error budget, what definition would I choose?

 

Whatever definition the opamp manufacturer used when he specified the CMRR as -60dB.

 

Whatever definition the opamp manufacturer used when he specified the CMRR as -60dB.

So if the conditions listed for CMRR were 0V - 4.2V that would mean what?

 

So if the conditions listed for CMRR were 0V - 4.2V that would mean what?

An adequate definition of the conditions under which the CMRR applies would have to say more than "0v - 4.2v". It would have to tell me what those two numbers refer to.

In your point No. 1), you didn't say what the reference was, but it's easy to infer that it's ground. In 2) and 3), you say explicitly what the reference is.

The really essential point you're getting at, is that the data sheets rarely say what the common mode reference is, and that's all important.

If the numbers "0v - 4.2v" refer to the common mode input voltage range, that would suggest that it's a single supply opamp with a supply of 5 volts. But, it doesn't tell me what the common mode reference is.

I notice that the data sheet for the Texas Instruments TLC272, for example, says in the electrical characteristics chart for a Vdd of 5 volts that the common mode input voltage range is -.2 to 4 (minimum) volts, and at the top of the chart, it says that these parameters apply at a supply voltage of 5 volts, and that under these conditions the minimum CMRR is 60 dB.

But they don't give a CMRR test circuit, or tell me what the common mode reference is, so a call to the manufacturer would be in order. Maybe you would answer the phone.

 

Finally, a good conversation.

If the numbers "0v - 4.2v" refer to the common mode input voltage range, that would suggest that it's a single supply opamp with a supply of 5 volts. But, it doesn't tell me what the common mode reference is.

Yes and the reference is Vee in my opinion. For a single supply op-amp a reference using the power supply mid-point (2.5V) would suggest Vicr would be -2.5V to +1.5V.

However, contrast this to the OPA656 for a +/-5V supply which lists +/-0.5Vcm for CMRR. The typical graph on pg. 9 actually shows a wider common-mode range, -4.2V to +2.5V, while maintaining >90dB for CMRR for the same +/-5V supply. So this suggests to me that they are referencing to the mid-point which, conveniently, would be 0V and not Vee which is -5V.

Perhaps it is arbitrary, but it causes different calculations for an error budget depending on what reference one chooses.

But they don't give a CMRR test circuit, or tell me what the common mode reference is...

Practically all manufactures use a DC servo loop, but regardless of the circuit the reference seems to be arbitrary.

so a call to the manufacturer would be in order. Maybe you would answer the phone.

That would be hard to do since I do not work there.

 

but regardless of the circuit the reference seems to be arbitrary

They almost always reference ground. In a single-supply config you always (well almost always) have a CMV present.

However, contrast this to the OPA656 for a +/-5V supply which lists +/-0.5Vcm for CMRR. The typical graph on pg. 9 actually shows a wider common-mode range, -4.2V to +2.5V, while maintaining >90dB for CMRR for the same +/-5V supply.

They tested it with Vcm at that voltage so you can add/sub that Vcm to get the CMIR @ CMRR see notes 5 and 6. The CMIR is limited by the CMR of the device. When you start to get too close to the rails the amp does funny things with the input like the "CM gain" starts to become significant thus the ratio gets smaller. Of course that is not the whole picture -- other params are affected like bias currents and

Also CMRR as listed is a DC parameter but it is frequency dependent -- lower as frequency goes up.

 

They almost always reference ground. In a single-supply config you always (well almost always) have a CMV present.

Under what conditions would you argue that a single supply op-amp does not have a common-mode voltage?

If the reference is the mid-point between supplies, then a single supply op-amp's common-mode voltage is 0 when the input is Vdd/2. If the reference for the same op-amp with the same input voltage is Vee like you said then the common-mode voltage is +2.5V.

The input structures;i.e., the differential pair, active load, and current mirror are biased the same way in both of the above examples. However, if you were told to do a budget analysis for Vos using the CMRR then you come away with two different answers due to an arbitrary reference point.

They tested it with Vcm at that voltage so you can add/sub that Vcm to get the CMIR @ CMRR see notes 5 and 6. The CMIR is limited by the CMR of the device. When you start to get too close to the rails the amp does funny things with the input like the "CM gain" starts to become significant thus the ratio gets smaller.

Notes 5 and 6 do not apply to the CMRR specification. They only apply to the min and max input voltage ranges and they are telling you what conditions occur at these times. These do not hold for the CMRR specification; i.e., MINIMUM 80dB at 25°C. This is a typical 'trick.'