Author Topic: Zero volt on the output of OP AMP  (Read 6087 times)

0 Members and 1 Guest are viewing this topic.

Offline Kalvin

  • Super Contributor
  • ***
  • Posts: 2145
  • Country: fi
  • Embedded SW/HW.
Re: Zero volt on the output of OP AMP
« Reply #25 on: February 18, 2018, 06:59:32 pm »
I think this is quite useful and educating project for a beginner. At least he should be getting a working instrument and also learn the limits of the real-life components and learn something about physical phenomena like thermal effects and circuit noise, compared to the clean simulated circuit. Typically it is quite easy to build an instrument with 1:1000 dynamic range, 1:10000 is much harder and 1:100000 requires good experience and knowledge (just ask any Voltnut here in the forum). As he is also adding a 24-bit ADC, there is some software development involved which won't be wasted time either, and can be reused in later projects.
 
The following users thanked this post: Tommy1984

Offline Tommy1984Topic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: se
Re: Zero volt on the output of OP AMP
« Reply #26 on: February 18, 2018, 07:01:57 pm »
I think this is quite useful and educating project for a beginner. At least he should be getting a working instrument and also learn the limits of the real-life components and learn something about physical phenomena like thermal effects and circuit noise, compared to the clean simulated circuit. Typically it is quite easy to build an instrument with 1:1000 dynamic range, 1:10000 is much harder and 1:100000 requires good experience and knowledge (just ask any Voltnut here in the forum). As he is also adding a 24-bit ADC, there is some software development involved which won't be wasted time either, and can be reused in later projects.

Thank you :-+
 

Offline Damianos

  • Frequent Contributor
  • **
  • Posts: 268
  • Country: gr
Re: Zero volt on the output of OP AMP
« Reply #27 on: February 18, 2018, 08:48:33 pm »
Very interesting project for a beginner!
An Ammeter with 7 digits resolution, made at home!
Using as a reference an instrument having an accuracy of 100 uV in the lowest range, while the requested resolution is 1 uV (accuracy?)!
How the amplifier will output 5V with 1.65V power supply is, of course, unimportant!
The Op-Amp noise figure is at least five times the requested resolution! (Signal 0.1 uV, Noise 0.5 uV typical)
... ...
After that, who cares about thermal voltages, interference...
With a little more effort, a ten or twelve digits instrument is easy to implement!

Well thank you for your reply first.
I updated the power supply to 5V so it will be +/- 2V5 so the op amp can handle max 2300 mV (which is 2.3 A).
I said at the begging that i'm trying to make this project real.
I'm a beginner, beginners make mistake and I don't feel shame of that.
But, the real shame is, to hang around and read beginners posts and problems and mention their failures in some way as ooooo I AM F%&KING GENIUS!   
I bet people like you don't dare to post a question or do anything useful in their life!

Cheers,
Tommy
Thanks for your opinion but keep it for yourself. Do you know anything about me? Do you have my CV at hand?
You are not able to understand for who was my comment but you can criticize easily!
I tried to point out the most basic things, while designing an instrument, that were ignored...

Good luck,
D.
 
The following users thanked this post: Tommy1984

Offline Tommy1984Topic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: se
Re: Zero volt on the output of OP AMP
« Reply #28 on: February 18, 2018, 09:17:16 pm »
Very interesting project for a beginner!
An Ammeter with 7 digits resolution, made at home!
Using as a reference an instrument having an accuracy of 100 uV in the lowest range, while the requested resolution is 1 uV (accuracy?)!
How the amplifier will output 5V with 1.65V power supply is, of course, unimportant!
The Op-Amp noise figure is at least five times the requested resolution! (Signal 0.1 uV, Noise 0.5 uV typical)
... ...
After that, who cares about thermal voltages, interference...
With a little more effort, a ten or twelve digits instrument is easy to implement!

Well thank you for your reply first.
I updated the power supply to 5V so it will be +/- 2V5 so the op amp can handle max 2300 mV (which is 2.3 A).
I said at the begging that i'm trying to make this project real.
I'm a beginner, beginners make mistake and I don't feel shame of that.
But, the real shame is, to hang around and read beginners posts and problems and mention their failures in some way as ooooo I AM F%&KING GENIUS!   
I bet people like you don't dare to post a question or do anything useful in their life!

Cheers,
Tommy
Thanks for your opinion but keep it for yourself. Do you know anything about me? Do you have my CV at hand?
You are not able to understand for who was my comment but you can criticize easily!
I tried to point out the most basic things, while designing an instrument, that were ignored...

Good luck,
D.

Thank you again for your help :-+

Regards,
Tommy
 

Offline StillTrying

  • Super Contributor
  • ***
  • Posts: 2850
  • Country: se
  • Country: Broken Britain
Re: Zero volt on the output of OP AMP
« Reply #29 on: February 18, 2018, 10:47:20 pm »
In practice I think your output will drift quite a bit, probably +/- 150uV or more if your full scale is ~ 5V.
So a 'Set Zero Button' will be needed at least !
.  That took much longer than I thought it would.
 
The following users thanked this post: Tommy1984

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Zero volt on the output of OP AMP
« Reply #30 on: February 18, 2018, 11:24:38 pm »
Easiest would be to use a composite autozero amp and a DPDT push button switch you push before you start using it to zero out the last few uV with the microcontroller.

The drift will be more than small enough and the low noise BJT opamp will let you measure something useful at the uV range.
« Last Edit: February 18, 2018, 11:26:59 pm by Marco »
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 16548
  • Country: us
  • DavidH
Re: Zero volt on the output of OP AMP
« Reply #31 on: February 19, 2018, 12:17:18 am »
Actually, I tried an OPA277 which has an offset trimmer (and very low input offset which is around 10uV), but, as the output is very low (tens of uV) and swinging a lot, it was so hard to trim it!

Usually the circuit configuration has DC gain so any input offset voltage is multiplied before measurement.  So for instance a non-inverting amplifier with a gain of 1000 will show 10 millivolts for 10 microvolts of input offset which is much easier to measure.

Input offset voltage can be adjusted at the input, output, or through the offset null pins.

If the offset null pins are used, then they should be used *only* to adjust the input offset voltage of the operational amplifier itself and not offsets generated elsewhere, like offset from the input bias current through the input resistance, because on precision operational amplifiers, trimming the input offset away from zero will increase the input offset voltage drift.

So across just one shunt resistor, voltage drop range is between 500 mV and 0.1uV.

This is going to be a problem no matter how good the operational amplifier is.  At low current shunt voltages, the thermocouple voltages generated by the dissimilar metal junctions inside and connecting to the current shunt will overwhelm the output from the current shunt.

Low value 4-wire precision resistance measurement gets around this by reversing the excitation current and voltage measurement "chopping" the output from the resistor in much the same way that a chopper amplifier removes its own input offset voltage.  The same thing could be done with a current shunt by placing it inside of a transistor H-bridge which is switched with overlapping drive to reverse the current through the shunt but I do not remember ever seeing this done in practice.
 
The following users thanked this post: Tommy1984

Offline Marco

  • Super Contributor
  • ***
  • Posts: 6694
  • Country: nl
Re: Zero volt on the output of OP AMP
« Reply #32 on: February 19, 2018, 02:57:17 am »
The input resistance is 0.1 Ohm, so offset current isn't relevant.

Thermal EMF for the resistance wire to copper lead connections, it seems so. Vishay has nice high power bulk metal foil current shunt, the CSM3637P, nowhere in stock though. They also have the CSM3637 which is 2W at 200 mOhm, they are even "guaranteed in stock", unfortunately Vishay seems to speak a language almost like English except the meaning of guaranteed is the opposite.
« Last Edit: February 19, 2018, 03:31:22 am by Marco »
 
The following users thanked this post: Tommy1984

Offline forrestc

  • Supporter
  • ****
  • Posts: 646
  • Country: us
Re: Zero volt on the output of OP AMP
« Reply #33 on: February 19, 2018, 11:12:56 am »
Instead of quoting any particular post, I figure I should start clean here:

What some of us are trying to explain is that the dynamic range of what you are trying to do is way too high.   

Let's start with the shunt resistor.

Let's assume at the bottom of your range you're willing to accept a 1uV input from your shunt.   This is probably way too low to be useful, but let's go there anyways.   Getting 1uV at 1uA takes a 1 \$\Omega\$ resistor.  At this end of the range the power dissipated by the resistor is negligible.   1e-8 watts.

Now, let's go to the other end of your range.   Putting 5A through that same 1 \$\Omega\$ resistor will now result in a 5V drop (burden voltage), and 25W of power dissipation.   That much power being dissipated in the shunt resistor is going to cause serious heating and will cause serious drift just due to the temperature coefficient of the resistor.  Not to mention that I'm not even sure you can get a 25W resistor with good enough accuracy/tempco effects (someone will probably let me know of some esoteric part which does).

Let's move on from the shunt.

Let's say that you instead just want to be able to measure from 1uV to 5V.   This is 5,000,000 counts, aka 7.5 digits.  In the ADC world, this does require somewhere between 22 and 23 bits of accuracy.  24 bits will give you some reasonable room to move.   One would think you could just slap a 24 bit adc on there and it would work fine.   But if this were the case, you'd have rigol and the like selling 7.5 digit dmm's for $99.00.   There's a reason that even the least expensive reasonable quality 5.5 digit DMM's cost multiple times that.

The real challenge is that electronics are really not that stable.   We've all been programmed to believe so with modern electronic design, but underneath it all, things drift, things aren't as accurate as you'd like.  Everything is affected by temperature.   At the lowest voltages even the temperature gradients on the circuit boards matter since the resistance of the traces change causing voltage offsets.   And on and on.   Getting a 7.5 digit anything done takes a lot of work and a lot of patience.   There's a reason why the 7.5 digit DMM's are well over a thousand dollars, even when bought used.

The real killer is dynamic range.   Although it is definitely quite difficult to read down to 1uV, it is even harder to read both 1uV and 5V without a range switch.   Most electronics like their signal to be bounded in a tight little range.  This is any thing but tight.  So you get to not only deal with the challenges of dealing with voltages down at the microvolt level, but you also have to deal with the challenges of high dynamic range, which is another challenge at least as difficult as the "micro voltage" issues.

I agree with others that a high precision project like this is an interesting thing for a beginner.  But be aware that a project like this takes a *lot* of attention to detail, and a lot of frustration, and a lot of research and most importantly a lot of experimentation and learning.   You may find that it is even impossible as stated, and so your goal may have to change, or even be abandoned.   As long as you are treating this as a project you may never finish but instead learn from the journey, then I'd say go for it.   But please be aware that every problem solved will likely make you discover a new problem.  And another and another.   But each of these is a learning experience.

P.S. I'm a big fan of nearly impossible goals, as it tends to prevent you from reaching a goal and then quitting.   The trick is understanding how to gain satisfaction from each step, instead of from reaching the goal.  And if by chance you get to what you thought was an nearly impossible goal, by all means pick another and proceed on.
 

Offline Tommy1984Topic starter

  • Regular Contributor
  • *
  • Posts: 57
  • Country: se
Re: Zero volt on the output of OP AMP
« Reply #34 on: February 21, 2018, 12:21:57 pm »
Instead of quoting any particular post, I figure I should start clean here:

What some of us are trying to explain is that the dynamic range of what you are trying to do is way too high.   

Let's start with the shunt resistor.

Let's assume at the bottom of your range you're willing to accept a 1uV input from your shunt.   This is probably way too low to be useful, but let's go there anyways.   Getting 1uV at 1uA takes a 1 \$\Omega\$ resistor.  At this end of the range the power dissipated by the resistor is negligible.   1e-8 watts.

Now, let's go to the other end of your range.   Putting 5A through that same 1 \$\Omega\$ resistor will now result in a 5V drop (burden voltage), and 25W of power dissipation.   That much power being dissipated in the shunt resistor is going to cause serious heating and will cause serious drift just due to the temperature coefficient of the resistor.  Not to mention that I'm not even sure you can get a 25W resistor with good enough accuracy/tempco effects (someone will probably let me know of some esoteric part which does).

Let's move on from the shunt.

Let's say that you instead just want to be able to measure from 1uV to 5V.   This is 5,000,000 counts, aka 7.5 digits.  In the ADC world, this does require somewhere between 22 and 23 bits of accuracy.  24 bits will give you some reasonable room to move.   One would think you could just slap a 24 bit adc on there and it would work fine.   But if this were the case, you'd have rigol and the like selling 7.5 digit dmm's for $99.00.   There's a reason that even the least expensive reasonable quality 5.5 digit DMM's cost multiple times that.

The real challenge is that electronics are really not that stable.   We've all been programmed to believe so with modern electronic design, but underneath it all, things drift, things aren't as accurate as you'd like.  Everything is affected by temperature.   At the lowest voltages even the temperature gradients on the circuit boards matter since the resistance of the traces change causing voltage offsets.   And on and on.   Getting a 7.5 digit anything done takes a lot of work and a lot of patience.   There's a reason why the 7.5 digit DMM's are well over a thousand dollars, even when bought used.

The real killer is dynamic range.   Although it is definitely quite difficult to read down to 1uV, it is even harder to read both 1uV and 5V without a range switch.   Most electronics like their signal to be bounded in a tight little range.  This is any thing but tight.  So you get to not only deal with the challenges of dealing with voltages down at the microvolt level, but you also have to deal with the challenges of high dynamic range, which is another challenge at least as difficult as the "micro voltage" issues.

I agree with others that a high precision project like this is an interesting thing for a beginner.  But be aware that a project like this takes a *lot* of attention to detail, and a lot of frustration, and a lot of research and most importantly a lot of experimentation and learning.   You may find that it is even impossible as stated, and so your goal may have to change, or even be abandoned.   As long as you are treating this as a project you may never finish but instead learn from the journey, then I'd say go for it.   But please be aware that every problem solved will likely make you discover a new problem.  And another and another.   But each of these is a learning experience.

P.S. I'm a big fan of nearly impossible goals, as it tends to prevent you from reaching a goal and then quitting.   The trick is understanding how to gain satisfaction from each step, instead of from reaching the goal.  And if by chance you get to what you thought was an nearly impossible goal, by all means pick another and proceed on.

You are right in every single word!  :-+

Regards,
Tommy
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf