Comparing noise specs in precision Op-Amps

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
Hello there!

When it comes to designing high-precision electronics, even the tiniest of noise, in any of the input or output or intermediate stages, plays a significant role in defining the quality of the device, and thus, cannot be ignored.

There are several specifications and plots concerning noise in a precision amplifier, such as input noise voltage, input noise voltage density, intermodulation noise, etc. It gets a bit challenging when it comes to prioritising these specifications, especially if one gets to comparing different devices, making the selection of "best-fit" a colossal task.

I would be really grateful if someone could guide me through, especially in properly weighing these noise specs for any particular application.
Once the weight of each of these noise specs and plots is specified, making the proper selection of devices would become a lot easier.
 

Daniel Sala

Joined May 28, 2015
65
Hi,

I couldn't delude myself or you that I can explain all that, especially as it sounds as though you know more than I do. I can point you in the direction of a little helpful reading, 'though, in case you haven't seen them yet:

Low Noise Amplifier Selection Guide for Optimal Noise Performance

Op Amps for Everyone

Op Amp Selection Guide for Optimum Noise Performance

There are quite a few app notes on this topic by different semiconductor manufacturers if you feel like doing the legwork to hunt them out using a few "variations on a theme" search terms, I really don't have the time to spend hours doing the search(es) right now, sorry.

TI Precision Labs - Op Amps has some nice, well-explained (short, e.g. +- 15 to 20 mins each) video tutorials with optional exercises worth looking at that also deal with this (/these) issue(s). I "Netflixed" my way through these over a few evenings and it was lots of fun and well worth the time spent on it, same as Op Amps for Everyone book, and the Analog Devices Op Amp Applications Handbook by Walt Jung may be worth perusing, direct pdf link on another website.


.
 

BR-549

Joined Sep 22, 2013
4,928
I think current is a stream of noise. It would take zero degrees to quiet down, but then there would be perfect tones. The only noiseless song, nature can sing, is at zero.
 

crutschow

Joined Mar 14, 2008
34,280
There are several specifications and plots concerning noise in a precision amplifier, such as input noise voltage, input noise voltage density, intermodulation noise, etc.
Intermodulation noise involves the noise from two frequencies generating additional noise, but that's not normally a problem with op amps.

Noise voltage density is the input noise in a given bandwidth (typically 1Hz) and is typically given in nV/√Hz for opamps.
You get the total input noise voltage by multiplying the noise voltage density by the square-root of the circuit bandwidth.
To minimize the total noise, you typically add a filter so that the circuit bandwidth is no larger than the desired signal bandwidth.

So picking the best low-noise opamp is mainly determined by the maximum noise you can tolerate, and the bandwidth of the signal, which determines the required opamp bandwidth.

Another factor is the signal source impedance.
If the source impedance is high, than the opamp input noise current can also be a factor, so you have to calculate how much input voltage noise the current-noise will generate as it goes through the input source resistance.
The total input noise is the square-root of the sum of the squares of the two input noise voltages.
For high source impedances, an opamp will a lower input noise current may give you a lower overall noise, even if its voltage noise is higher.
 
Then there is that wierd concept of "Noise Gain" that I don't understand.

Some X-ray detectors need to be cryogenically cooled and they degrade within days of lack of cooling.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
Intermodulation noise involves the noise from two frequencies generating additional noise, but that's not normally a problem with op amps.

Noise voltage density is the input noise in a given bandwidth (typically 1Hz) and is typically given in nV/√Hz for opamps.
You get the total input noise voltage by multiplying the noise voltage density by the square-root of the circuit bandwidth.
To minimize the total noise, you typically add a filter so that the circuit bandwidth is no larger than the desired signal bandwidth.

So picking the best low-noise opamp is mainly determined by the maximum noise you can tolerate, and the bandwidth of the signal, which determines the required opamp bandwidth.

Another factor is the signal source impedance.
If the source impedance is high, than the opamp input noise current can also be a factor, so you have to calculate how much input voltage noise the current-noise will generate as it goes through the input source resistance.
The total input noise is the square-root of the sum of the squares of the two input noise voltages.
For high source impedances, an opamp will a lower input noise current may give you a lower overall noise, even if its voltage noise is higher.
Thanks a lot, crutschow.
However, I am aware of the fundamental principles of all these specs. What I wanted to ask is how to weigh these specs relative to each other. As we all might have noticed when it comes to comparing op-amps, some have lesser input noise voltage density, while some other have lesser intermodulation distortion, and so on.
 

crutschow

Joined Mar 14, 2008
34,280
What I wanted to ask is how to weigh these specs relative to each other. As we all might have noticed when it comes to comparing op-amps, some have lesser input noise voltage density, while some other have lesser intermodulation distortion, and so on.
You have to look at your specific application and calculate which op amp gives the lowest total noise from all the noise factors.
You do know that random noise from various sources adds as the square-root of the sum-of-the-squares?
There's no one-size-fits-all.
 

ronsoy2

Joined Sep 25, 2013
71
The best thing to do is stick with a good manufacturer like Analog Devices who makes state of the art components. Simply describe your application to the applications engineer and they will give you the recommendation for the best amp to use in that application. They have engineers experienced in all fields so you can get great help. There are other manufacturers that are also great. I listed Analog Devices simply because they are my favorite.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
Hi,

I couldn't delude myself or you that I can explain all that, especially as it sounds as though you know more than I do. I can point you in the direction of a little helpful reading, 'though, in case you haven't seen them yet:

Low Noise Amplifier Selection Guide for Optimal Noise Performance

Op Amps for Everyone

Op Amp Selection Guide for Optimum Noise Performance

There are quite a few app notes on this topic by different semiconductor manufacturers if you feel like doing the legwork to hunt them out using a few "variations on a theme" search terms, I really don't have the time to spend hours doing the search(es) right now, sorry.

TI Precision Labs - Op Amps has some nice, well-explained (short, e.g. +- 15 to 20 mins each) video tutorials with optional exercises worth looking at that also deal with this (/these) issue(s). I "Netflixed" my way through these over a few evenings and it was lots of fun and well worth the time spent on it, same as Op Amps for Everyone book, and the Analog Devices Op Amp Applications Handbook by Walt Jung may be worth perusing, direct pdf link on another website.


.

Thank a ton, Daniel Sala.
All these links provided me with much deeper insights. Now, I can say I have a clear image of the various noise sources in an active circuitry, especially Op-Amps, how they influence the input and the output of the circuitry, and how to minimize the same.

But there is one thing I am still not very clear about. All these articles talk about the interaction of input voltage and current noises with the (external) op-amp circuitry and the source impedance. However, none of these, or any other, articles give any hint on the influence of the voltage and current noise sources interacting with the input impedance of the op-amp.

Could anyone please provide some more insights into the same?
 

OBW0549

Joined Mar 2, 2015
3,566
However, none of these, or any other, articles give any hint on the influence of the voltage and current noise sources interacting with the input impedance of the op-amp. Could anyone please provide some more insights into the same?
The influence is negligible, because whatever the op amp's input impedance is, it inevitably is in parallel with the impedance of the external circuit elements which are several-- often, many-- orders of magnitude lower. As a result, the combined impedance (which is what the op amp's input current noise interacts with) is essentially unchanged from that of those external elements alone.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
The influence is negligible, because whatever the op amp's input impedance is, it inevitably is in parallel with the impedance of the external circuit elements which are several-- often, many-- orders of magnitude lower. As a result, the combined impedance (which is what the op amp's input current noise interacts with) is essentially unchanged from that of those external elements alone.
OOPS! I missed that!
Thanks, OBW0549.

Now, I have one more question.
Why is input current noise standardized as two equal current sources, each sourcing out of one of the two inputs of an Op-Amp? If this is how the input current noise is modeled, then it won't ever interact with the input impedance, since the effective current through the input impedance is zero!

Where did I take the wrong turn?
 

OBW0549

Joined Mar 2, 2015
3,566
Why is input current noise standardized as two equal current sources, each sourcing out of one of the two inputs of an Op-Amp? If this is how the input current noise is modeled, then it won't ever interact with the input impedance, since the effective current through the input impedance is zero!

Where did I take the wrong turn?
I'm not sure, but it sounds to me like you're considering only the differential input impedance (that is, the impedance between the two inputs) in arriving at the above conclusion. There's also a common-mode input impedance, the impedance between each input and the op amp's supply terminals.

In any case, as I said above, it's simply not a factor and can be ignored.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
I'm not sure, but it sounds to me like you're considering only the differential input impedance (that is, the impedance between the two inputs) in arriving at the above conclusion. There's also a common-mode input impedance, the impedance between each input and the op amp's supply terminals.

In any case, as I said above, it's simply not a factor and can be ignored.
Yes, I am considering only the differential impedance for this question.
I am aware of the two types of input impedance specifications. Common-mode impedances are a piece of cake. But when it comes to dealing with differential-mode calculations, both common-mode and differential-mode impedances have their influence on the same.
 

OBW0549

Joined Mar 2, 2015
3,566
But when it comes to dealing with differential-mode calculations, both common-mode and differential-mode impedances have their influence on the same.
For some kinds of calculations these impedances are relevant; but I've never needed to consider them when doing noise analyses.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
For some kinds of calculations these impedances are relevant; but I've never needed to consider them when doing noise analyses.
Me too!

But this particular application I am dealing in right now requires me to pick up signals from a very high impedance source, of the order of tens of MegaOhms. So, It's a bit challenging to make decisions!
 

OBW0549

Joined Mar 2, 2015
3,566
But this particular application I am dealing in right now requires me to pick up signals from a very high impedance source, of the order of tens of MegaOhms. So, It's a bit challenging to make decisions!
The fact that you're dealing with a source having extremely high impedance would suggest that for lowest noise, you might need to select an op amp with very low input current noise.

Over the years I've evolved a simple (but crude) method for determining whether opamp input voltage noise or input current noise is likely to be more important in selecting an opamp for a low-noise application. I simply divide the part's specified input noise voltage by its specified input noise current to yield what I call the "noise impedance." This represents the value of source impedance at which the noise contributions from the opamp's input voltage noise and its input current noise, acting through that source impedance, will be equal. If that calculated value is less than the impedance of my source, I know I need to be looking for an opamp with less input current noise; if that calculated value is greater than my source impedance, I know that I might want an opamp with lower input voltage noise.

I say the method is "crude" because it takes no account of 1/f voltage or current noise; the two may have very different spectral characteristics. Nor does it take into account the Johnson noise of the source itself, which can be significant for a high-impedance input signal source.

But most of the time it works, and makes op amp selection a bit simpler.
 

Thread Starter

Abbas_BrainAlive

Joined Feb 21, 2018
113
The fact that you're dealing with a source having extremely high impedance would suggest that for lowest noise, you might need to select an op amp with very low input current noise.

Over the years I've evolved a simple (but crude) method for determining whether opamp input voltage noise or input current noise is likely to be more important in selecting an opamp for a low-noise application. I simply divide the part's specified input noise voltage by its specified input noise current to yield what I call the "noise impedance." This represents the value of source impedance at which the noise contributions from the opamp's input voltage noise and its input current noise, acting through that source impedance, will be equal. If that calculated value is less than the impedance of my source, I know I need to be looking for an opamp with less input current noise; if that calculated value is greater than my source impedance, I know that I might want an opamp with lower input voltage noise.

I say the method is "crude" because it takes no account of 1/f voltage or current noise; the two may have very different spectral characteristics. Nor does it take into account the Johnson noise of the source itself, which can be significant for a high-impedance input signal source.

But most of the time it works, and makes op amp selection a bit simpler.

I get your approach quite well. That's what is mentioned in the links shared by Daniel Sala in #2. It also takes into account the Johnson noise of the source impedance and the Op-Amp circuitry as well. I would suggest you have a look at it, you might find something useful.

Although, these articles do not take into account the 1/f noise. If anyone finds something on how to tackle it, please let us all know.
 

crutschow

Joined Mar 14, 2008
34,280
Although, these articles do not take into account the 1/f noise. If anyone finds something on how to tackle it, please let us all know.
1/f noise is generated by the op amp, so if low frequency noise is important, than you want to pick an op amp that has low 1/f noise as well as low voltage noise.

What is you signal's lowest frequency of interest?
 
Top