Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Understanding opamp noise specs

Status
Not open for further replies.

Mosaic

Well-Known Member
The LM358 has a noise spec of 40nv/root(Hz). As a difference amplifier for DC does this mean that I can expect a 40nV noise on the output? Does the noise increase with gain?

Therefore for a 100Hz waveform is the noise getting less by a factor of 10?
 
The noise is amplified by the gain of the amp and will indeed appear on the output. You multiply 40nV/√Hz by the square-root of the bandwidth of your circuit. Thus if you had a circuit with a gain of 10 and a 10KHz bandwidth, the output noise would be 10 ×√10kHz × 40nV = 40µV.

The noise has nothing to do with the frequency of any signals going through the circuit, only the circuit bandwidth frequency. And it does not reduce with frequency (although 1/f noise, which is a type of low frequency random noise also present in amplifiers, is inversely proportional to frequency).

Therefore, to minimize noise you want to add a filter to the circuit to limit the bandwidth to no greater than you need.
 
Carl, can u help me out on this? Sample circuit, perhaps a diff opamp of 5X with bandwidth limitation?
 
Carl, can u help me out on this? Sample circuit, perhaps a diff opamp of 5X with bandwidth limitation?
What are the signal and power requirements for the circuit?

What are the differential requirements (common-mode rejection, common-mode range, etc.)?

What is the frequency response you need?
 
specs

Well, I am reading the voltage impressed on a .05 ohm current sense resistor, with either a constant 4 or 8 amp load. The intention is to calc. the int. resistance of a lead acid batt. which can be from under .01 ohm to perhaps .1 ohm.

The diff amp is fed by a 12.288 V offset and is measuring a 1.024V range up to 13.312V as a battery can vary in charged voltage depending on temps by a few tenths of a V. Nominal resting full charge is about 12.8Vdc.

As the input range of the diff amp is 1.024V max diff. it does a 4x gain to deliver a 0-4.096 V to the MCU ADC (Ref'd to 4.096V) . The MCU can calc the internal resistance of the lead acid based on V=IR given the open loop Voltage of the Batt vs the loaded voltage of the batt. You can see that at a .01 ohm Int res. the batt will only drop 40mV under a 4 Amp load. So I need to control the 'noise' voltage by limiting the bandwidth as you have described. Perhaps to something like 2mV max? I can do multiple ADC samples to avg out small noise.

I included an attempt at filtering the Opamp gain The bode plot shows a big spike just before the gain collapses around 1Khz. Otherwise the gain is 1% stable from under around 80Hz which is good as we are looking at DC loading over the sense resistor. Is this how to approach it?
 

Attachments

  • opampsD.png
    opampsD.png
    107.3 KB · Views: 146
  • opamp-bwlimit.asc
    2.3 KB · Views: 112
Now that I finally understand what you are trying to do I can better help you. (Why does everyone always ask for help to try to solve the design problem the think they have rather than solving the application problem they do have?):rolleyes:

First, you can't control the frequency response by adding a capacitor directly to the output of an op amp. That's whats causing the large spike in the frequency response. You need to add a resistor in series with the output to the capacitor.

Second, a high side current monitor circuit should work better for you than an op amp configured as a differential op amp. They are optimized to do what you want.

The common-mode rejection of an op amp circuit is no better than the matching of the resistors, thus with 1% resistors the CM rejection is only 40dB. That will give large errors as the battery voltage changes.
 
Thanks for the advice:

I see an interesting opamp design in fig 9.3 in this link:
**broken link removed**

It seems to get around having to use matched resistors.

BTW is this bandwidth limiter as u describe...it seems to work now.
 

Attachments

  • opamp-bwlimit.asc
    2.4 KB · Views: 112
Yes, those are typical high side current measurement circuits. They need an instrumentation amplifier or a rail-to-rail input and output type op amp, depending upon the circuit..
 
Thanks for the advice:

I see an interesting opamp design in fig 9.3 in this link:
**broken link removed**




It seems to get around having to use matched resistors.

BTW is this bandwidth limiter as u describe...it seems to work now.


Hi,

One more little thing here...

If the op amp is spec'd with a given input resistance then the noise could increase if the resistance is increased. For example, if it is spec'd at 100 ohms and you raise the input resistor to 1k the noise could increase by as much as 10.

The amp you have now doesnt look too bad really. You could improve a little by increasing that 100 ohm resistor on the output to 470 or something like that so the op amp output isnt loaded as much.
 
Last edited:
If the op amp is spec'd with a given input resistance then the noise could increase if the resistance is increased. For example, if it is spec'd at 100 ohms and you raise the input resistor to 1k the noise could increase by as much as 10.
That's only true for the input current noise. It has no effect on the voltage noise, which is usually the dominant noise source for typical values of input resistor values.
 
Hi Carl,

That's fine, i just wanted to mention this possibility. I dont have any spec's in front of me right now so i cant compare.
 
Guys, thanks for the input....

Can you suggest a suitable opamp? The 358 has a -1.5V on the Vcc rail, the mc33072 I have has a +.3V on the Vee rail. I can supply the opamps with 24VDC or less rail to rail. 4.096V is max output.

BTW what is the prob with using the 358 if the output is 1.5VDC below VCC once Vcc is much greater than the expected output voltage swing? The pdf link didn't really go into it...they just said rail to rail, I'd like to understand why. Seems to me both of those amps I mentioned meets the more detailed spec .6V>Vout>Vin (sense).
 
Last edited:
BTW what is the prob with using the 358 if the output is 1.5VDC below VCC once Vcc is much greater than the expected output voltage swing? The pdf link didn't really go into it...they just said rail to rail, I'd like to understand why. Seems to me both of those amps I mentioned meets the more detailed spec .6V>Vout>Vin (sense).
Yes, the 358 should work for that.

You need rail-to-rail if either the input or output need to go near the power rails.

But if you only need to go from 0V to within 1.5V of the supply voltage or less then the 358 will work.
 
Carl , your assistance has been outstanding!!

I attach a simulation of the requisite current sense as well as battery voltage sense with comments on the integer MCU calcs...which give a best accuracy of .04% incl the opamp offset V. Now, it's just about getting decent resistors or manually matching them up, or using trimmers and calibrating them. I'm looking at using arrays to help with consistency.
https://www.newark.com/bourns/4816p-t01-103lf/resistor-iso-res-n-w-8-10kohm-2/dp/33K0647

I am using the LM258 amp as I have a few around. Hopefully they're a bit better than the 358s.
 

Attachments

  • Curr-sense.asc
    5.1 KB · Views: 119
  • LM258.sub.txt
    2.9 KB · Views: 103
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top