Author Topic: Need your help offset buffer  (Read 4237 times)

0 Members and 1 Guest are viewing this topic.

Offline corsoTopic starter

  • Newbie
  • Posts: 3
  • Country: ca
Need your help offset buffer
« on: July 08, 2016, 06:59:19 pm »
Hi all,

I have a strange problem with an opamp in buffer mode. At the input I have 1.9889V and the output is 1.9969V, so 8mV offset.
In the datasheet the MCP6V64 have +/-8uV offset?...
I have try with MCP6004 and a AD8630, but same problem.

I have try to remove the tension divisor at the input of the opamp (resistor 200k and 50k), the offset was just 150uV, this is perfect for what I need to do, but I need the tension divisor.

Some one have a solution or something that I can try to resolve my problem.



Thank you,
_Corso_
 

Offline Andy Watson

  • Super Contributor
  • ***
  • Posts: 2340
Re: Need your help offset buffer
« Reply #1 on: July 08, 2016, 07:05:58 pm »
Are you using a meter with 10M ohm input impedance? Consider what effect that has on the resistive input divider.
 
The following users thanked this post: corso

Offline corsoTopic starter

  • Newbie
  • Posts: 3
  • Country: ca
Re: Need your help offset buffer
« Reply #2 on: July 08, 2016, 07:11:57 pm »
Yes, I use a Fluke 45, the impedance is 10M ohm.
_Corso_
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2892
  • Country: se
  • Country: Broken Britain
Re: Need your help offset buffer
« Reply #3 on: July 08, 2016, 07:59:28 pm »
Very roughly.
You're measuring the same voltage at 2 very different impedances, ~0.5R at the opamp output and ~50K at the +Ve input.The Fluke's 10M means the measurement at the output will be low by 0.5/10M  = 1 part in 20 million, so you can forget that error.
But at the +Ve input the error is 50K/10M  =  low by 1 part in 200. As you're measuring ~2V, low by 1 part in 200 = 10mV, which is close enough to your -8mV for me!

Measure directly between the +Ve input and the output instead of Gnd and the measured offset between them 2 points should be much less than 8mV.
« Last Edit: July 08, 2016, 08:01:20 pm by StillTrying »
.  That took much longer than I thought it would.
 
The following users thanked this post: corso

Offline corsoTopic starter

  • Newbie
  • Posts: 3
  • Country: ca
Re: Need your help offset buffer
« Reply #4 on: July 08, 2016, 08:35:16 pm »
StillTrying, you are right, I have 152 uV with the +Ve input and the output  ;) . Not exactly +/-8uV but it okay for my job...

If I put the input to the ground (0V), I get 300uV at output of the opamp, so why not 8uV?
_Corso_
 

Offline RandallMcRee

  • Frequent Contributor
  • **
  • Posts: 550
  • Country: us
Re: Need your help offset buffer
« Reply #5 on: July 08, 2016, 08:48:20 pm »
Hmmm, seems like this is an error due to the input bias (offset) current of the opamp. You can compensate for it...

http://www.daenotes.com/electronics/devices-circuits/operational-amplifier-bias-current-compensation-method

Randy
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2892
  • Country: se
  • Country: Broken Britain
Re: Need your help offset buffer
« Reply #6 on: July 08, 2016, 09:05:44 pm »
If I put the input to the ground (0V), I get 300uV at output of the opamp, so why not 8uV?
The output saturation voltages come into it then. I can't see what it is at OV from the datasheet, possibly 3.5mV with a 20K load, but at 0.3mV your winning on that one. You could try a 20K from output to Gnd to keep it low, it'll depend on whether it has to sink any current from the ADC.
.  That took much longer than I thought it would.
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 6117
  • Country: au
    • send complaints here
Re: Need your help offset buffer
« Reply #7 on: July 08, 2016, 11:23:27 pm »
Hmmm, seems like this is an error due to the input bias (offset) current of the opamp. You can compensate for it...
Nope, the max input bias current of 50pA wouldn't be causing enough offset for these measurements (similar in magnitude to the 8uV input offset). Measuring between the input and output should have mostly eliminated the impedance of the multimeter, as its now at almost the same potential as the sensitive input node just like controlling leakage with guard rings.
 

Offline Jay_Diddy_B

  • Super Contributor
  • ***
  • Posts: 2994
  • Country: ca
Re: Need your help offset buffer
« Reply #8 on: July 09, 2016, 08:06:46 am »
Hi group,

I agree with members Andy Watson and StillTrying.

Here is some analysis performed with LTspice. The SPICE primitive, E, is a voltage controlled voltage source, essentially an amplifier.

First I will show the effect of having a large, but limited gain, in an op-amp.



The general formula for a servo with feedback is:

Gain = A / 1 + A*B

Where A is the open loop gain of the amplifier
and B is the gain of the feedback path

In the case of a buffer, B=1

A= Open loop gain of the amplifier, normally considered very large, but does introduce an error in ppm accuracy circuits.

If you assume the A=120dB

A = 1E6

Gain = 1E6 / (1+1E6) = 0.999999

This introduces 1ppm error


If the gain is 100dB, there is a 10ppm error



Meter Loading


The 10M input resistance of the DMM will load the 200k, 50k divider:




Thevenin Equivalent

The 10V source can be replace with a 2V and resistors of 200K and 50K in parallel (40k):




This results in 8mV of error, exactly what the OP was measuring.

I have attached the LTspice model.

Regards,

Jay_Diddy_B

 
The following users thanked this post: corso

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2892
  • Country: se
  • Country: Broken Britain
Re: Need your help offset buffer
« Reply #9 on: July 10, 2016, 12:35:07 am »
I think I prefer the "Guesstimate it as you go" method.  :)
.  That took much longer than I thought it would.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf

 

- - -
-->