Ultra low voltage amplifier

[Tommy1984]

Hi,
I tried to amplifier an ultra low voltage (0.05 mV) with help of AD8628, but, without any luck!
As it's shown in the picture, I have gain of 1000 or (1001), so the output must be something close to 45 mV. But all I got was 150mV.
Isn't so strange? Can anybody tell me where's my fault?

Tommy

[David Hess]

It should have worked.  You might need to add a feedback capacitor across R3 and another capacitor from the non-inverting input to ground to limit noise.

[Audioguru]

I hope the input signal is DC because it is not biased properly to be an AC amplifier.

[Zero999]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

[Tommy1984]

It should have worked.  You might need to add a feedback capacitor across R3 and another capacitor from the non-inverting input to ground to limit noise.

Thanks for your reply!
Any suggestions for capacitors values?

[Tommy1984]

I hope the input signal is DC because it is not biased properly to be an AC amplifier.

Thanks for your reply!
Ya it’s DC!

[Tommy1984]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

Thanks for your reply!
The output is around 150 mV.
How do you mean by shorting the probs? The input terminals?

[Zero999]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

Thanks for your reply!
The output is around 150 mV.
How do you mean by shorting the probs? The input terminals?

Presumably you were using a multimeter? If it's a really cheap model, it's possible or has wandered off calibration it's possible it has an offset voltage. See what it reads, when it isn't connected to your circuit and the probes are shorted together.

[Tommy1984]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

Thanks for your reply!
The output is around 150 mV.
How do you mean by shorting the probs? The input terminals?

Presumably you were using a multimeter? If it's a really cheap model, it's possible or has wandered off calibration it's possible it has an offset voltage. See what it reads, when it isn't connected to your circuit and the probes are shorted together.

Aha ok! Now I understand! No it’s not a cheap one and yes it’s shows ZERO when I short the probs!
I’m using VC 880
https://www.conrad.com/ce/en/product/124609/Handheld-multimeter-digital-VOLTCRAFT-VC880-Calibrated-to-Manufacturers-standards-no-certificate-Data-logger-CAT-III

[Zero999]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

Thanks for your reply!
The output is around 150 mV.
How do you mean by shorting the probs? The input terminals?

Presumably you were using a multimeter? If it's a really cheap model, it's possible or has wandered off calibration it's possible it has an offset voltage. See what it reads, when it isn't connected to your circuit and the probes are shorted together.

Aha ok! Now I understand! No it’s not a cheap one and yes it’s shows ZERO when I short the probs!
I’m using VC 880
https://www.conrad.com/ce/en/product/124609/Handheld-multimeter-digital-VOLTCRAFT-VC880-Calibrated-to-Manufacturers-standards-no-certificate-Data-logger-CAT-III

What do you get when you short the input of your amplifier to 0V?

Where did you get the AD8628 from? Could it be fake?

[daqq]

The voltage is VERY close to the negative power rail. Are you sure that it would work at those levels? While it is true that it's a rail to rail op amp, that just means that it can operate closer to those rails than a non rail to rail op amp.

[tszaboo]

1.) Try powering the amplifier from +/- 2.5V. It might be a rail to rail problem.
2.) Try decreasing the 1M resistor to 100K and the other one also. It could be an offset voltage due to bias and offset current.

[2N3055]

That opamp has less than 100 pA input bias current... Non inverting input of opamp is "in the air" with 10s of gigaohms of input loading and will function as a nice electrometer....
You need to provide guarding for the input and also clean board and area around the input very, very good..

[Tommy1984]

That opamp has less than 100 pA input bias current... Non inverting input of opamp is "in the air" with 10s of gigaohms of input loading and will function as a nice electrometer....
You need to provide guarding for the input and also clean board and area around the input very, very good..

Thanks for your reply!
I'm thinking of using a 100 ohm resistor for the input, do you think it's ok?

[2N3055]

It depends what do you connect to the input.
It will be correlated to output impedance of what you connect to input.

What do you plan to connect to input?

[Tommy1984]

It depends what do you connect to the input.
It will be correlated to output impedance of what you connect to input.

What do you plan to connect to input?

As shown in the picture, it's 0.045 mVDC

[2N3055]

I read that, and didn't mean that..

I meant: what, as in is shunt resistor, output of some sensor, NTC, strain gauge... What?
So we can determine interaction of that and opamp input...

[Tommy1984]

I read that, and didn't mean that..

I meant: what, as in is shunt resistor, output of some sensor, NTC, strain gauge... What?
So we can determine interaction of that and opamp input...

A shunt resistor!

[tszaboo]

That opamp has less than 100 pA input bias current... Non inverting input of opamp is "in the air" with 10s of gigaohms of input loading and will function as a nice electrometer....
You need to provide guarding for the input and also clean board and area around the input very, very good..

Yes. 100pA x 1Mohm x 1000 gain is 100mV

[2N3055]

AAh now we're talking... :-)

in that case, if you can be sure that opamp input will be always connected to shunt, you don't need anything, except shunt.

Did you try to ground noniverting input and check output then? Any changes? What is output then?

Regards,

Sinisa

[Tommy1984]

AAh now we're talking... :-)

in that case, if you can be sure that opamp input will be always connected to shunt, you don't need anything, except shunt.

Did you try to ground noniverting input and check output then? Any changes? What is output then?

Regards,

Sinisa

Hi Sinisa,
can you explain the wiring more please? Grounding the non-inverting and?
Regards, Tommy

[Tommy1984]

That opamp has less than 100 pA input bias current... Non inverting input of opamp is "in the air" with 10s of gigaohms of input loading and will function as a nice electrometer....
You need to provide guarding for the input and also clean board and area around the input very, very good..

Yes. 100pA x 1Mohm x 1000 gain is 100mV

I didn't get your formula well!
I have 0.045 mV input and the gain is about 1000 so the output must be something close to 45mV.... What's the relation between input voltage, Rf, output voltage and the current?

[2N3055]

He is talking about feedback resistors... It is all about currents... and voltage drops over resistors..

But I think input offset caused by input offset current would be 100pA over 1000 Ohms.. That would be 100 nV multiplied with 1000 times amplification that gives error of 100 uV on output... Not 100 mV....

As for wiring, I meant grounding input, meaning connecting + input to ground. That should give you close to zero on input and close to zero on output... I said close to zero on output because there will be internal opamp voltage offset multiplied with amplification.. But it should be less then or about 1mV max....

[Tommy1984]

He is talking about feedback resistors... It is all about currents... and voltage drops over resistors..

But I think input offset caused by input offset current would be 100pA over 1000 Ohms.. That would be 100 nV multiplied with 1000 times amplification that gives error of 100 uV on output... Not 100 mV....

As for wiring, I meant grounding input, meaning connecting + input to ground. That should give you close to zero on input and close to zero on output... I said close to zero on output because there will be internal opamp voltage offset multiplied with amplification.. But it should be less then or about 1mV max....

I understand it, non-inverting input (the + input) to ground. And then the input voltage to the inverting input through Rf and Rg?

[2N3055]

No. Your schematic is OK.  Something is wrong with circuit as soldered....

So I recommend to short input to ground so you have approx. zero on input  and then measure output.
You should have less than 1 mV in that config... If you have more, something is wrong...

Look, my advice to you is to download LTSpice and play with that.
I attached schematic of your circuit... It works fine...

There are many good free resources for opamp theory... And many more books on the topic... It's light reading time... :-)

Good luck!

Sinisa

[tszaboo]

That opamp has less than 100 pA input bias current... Non inverting input of opamp is "in the air" with 10s of gigaohms of input loading and will function as a nice electrometer....
You need to provide guarding for the input and also clean board and area around the input very, very good..

Yes. 100pA x 1Mohm x 1000 gain is 100mV

I didn't get your formula well!
I have 0.045 mV input and the gain is about 1000 so the output must be something close to 45mV.... What's the relation between input voltage, Rf, output voltage and the current?

Sorry, ignore that, it is wrong of course. The source impedance for the amplifier is about 1K not 1M. But the offset current is also more that 100pA, up to 250pA
So the error is much smaller. Anyway, time to go to bed.
Anyway, read this:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5693

[Tommy1984]

No. Your schematic is OK.  Something is wrong with circuit as soldered....

So I recommend to short input to ground so you have approx. zero on input  and then measure output.
You should have less than 1 mV in that config... If you have more, something is wrong...

Look, my advice to you is to download LTSpice and play with that.
I attached schematic of your circuit... It works fine...

There are many good free resources for opamp theory... And many more books on the topic... It's light reading time... :-)

Good luck!

Sinisa

Wow! Fantastic!
I'll try it tomorrow, thank you very much!
Regards,
Tommy

[2N3055]

Sorry, ignore that, it is wrong of course. The source impedance for the amplifier is about 1K not 1M. But the offset current is also more that 100pA, up to 250pA
So the error is much smaller. Anyway, time to go to bed.
Anyway, read this:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5693

That's a good read NAND... Right on the topic...

And yeah it's bed time... Good night!

[Tommy1984]

Yes, it should work.

What did you measure the output? Did you check it reads zero when the probes are shorted together?

Thanks for your reply!
The output is around 150 mV.
How do you mean by shorting the probs? The input terminals?

Presumably you were using a multimeter? If it's a really cheap model, it's possible or has wandered off calibration it's possible it has an offset voltage. See what it reads, when it isn't connected to your circuit and the probes are shorted together.

Aha ok! Now I understand! No it’s not a cheap one and yes it’s shows ZERO when I short the probs!
I’m using VC 880
https://www.conrad.com/ce/en/product/124609/Handheld-multimeter-digital-VOLTCRAFT-VC880-Calibrated-to-Manufacturers-standards-no-certificate-Data-logger-CAT-III

What do you get when you short the input of your amplifier to 0V?

Where did you get the AD8628 from? Could it be fake?

I got about 6.3 mV when I shorted it to ground  Isn't so weird?
I must get roughly  0 V !!!
And where did I get it, it's a good question. From ebay (China!) is it possible that it's fake?
I'll try another one (from the same source) and give a feedback

Regards,
Tommy

[daqq]

From ebay (China!) is it possible that it's fake?

Yes. There's a problem with fake parts from ebay, you're taking your chances every time you buy something.

[Tommy1984]

From ebay (China!) is it possible that it's fake?

Yes. There's a problem with fake parts from ebay, you're taking your chances every time you buy something.

Sorry, I meant 6.3 mV !

[Zero999]

From ebay (China!) is it possible that it's fake?

Yes. There's a problem with fake parts from ebay, you're taking your chances every time you buy something.

Sorry, I meant 6.3 mV !

It's highly likely it's fake. What often happens is the fraudster changes the marking on a cheaper op-amp IC, to a part number of a more expensive part. What you have is a working op-amp, but it's quite likely a much cheaper part than the AD8628. Try ordering one from a reputable supplier, such as: RS Components, Farnell, Digikey, Mouser etc.

[Tommy1984]

1.) Try powering the amplifier from +/- 2.5V. It might be a rail to rail problem.
2.) Try decreasing the 1M resistor to 100K and the other one also. It could be an offset voltage due to bias and offset current.

It works finally! 
After powering the op amp with  +/- 2V5, it workes like a charm!
I used LM358 op amp to split the 5V voltage to +/- 2V5, but I noted something which was very strange!
After powering the AD8629 op amp, and before connecting anything to the non-inverted input, the voltage over -V and +V were -2V5 respectively +2V5, but after connecting to it, the voltage became 0V respectively +5V!!!
Is that normal?

Tommy

[Zero999]

1.) Try powering the amplifier from +/- 2.5V. It might be a rail to rail problem.
2.) Try decreasing the 1M resistor to 100K and the other one also. It could be an offset voltage due to bias and offset current.

It works finally! 
After powering the op amp with  +/- 2V5, it workes like a charm!
I used LM358 op amp to split the 5V voltage to +/- 2V5, but I noted something which was very strange!
After powering the AD8629 op amp, and before connecting anything to the non-inverted input, the voltage over -V and +V were -2V5 respectively +2V5, but after connecting to it, the voltage became 0V respectively +5V!!!
Is that normal?

Tommy

That does sound odd. Please post a schematic of your set-up.

[Tommy1984]

That does sound odd. Please post a schematic of your set-up.

Here's my schismatic!
Before I connect anything to SV1, voltage over -V and +V is -2.5 V respectively +2.5 V, after connecting, becomes 0 V respectively +5 V

Regards,
Tommy

[Zero999]

That does sound odd. Please post a schematic of your set-up.

Here's my schismatic!
Before I connect anything to SV1, voltage over -V and +V is -2.5 V respectively +2.5 V, after connecting, becomes 0 V respectively +5 V

Regards,
Tommy

You're shorting your virtual ground node out. By the way, R6 is not a good idea, because it increases the impedance of the virtual ground.

[danadak]

If you are measuring a shunt, to measure current, might make sense to use an IA,
to get rid of any common mode errors in the layout and wiring.


Take a look at this -


http://www.analog.com/en/education/education-library/dh-designers-guide-to-instrumentation-amps.html


Regards, Dana.

[Tommy1984]

You're shorting your virtual ground node out. By the way, R6 is not a good idea, because it increases the impedance of the virtual ground.

You are right!
I was powering my circuit with the same source which I used to measure thorough the shunt resistor!
I simply powered the circuit from another source and it worked fine

Regards,
Tommy

[Tommy1984]

If you are measuring a shunt, to measure current, might make sense to use an IA,
to get rid of any common mode errors in the layout and wiring.


Take a look at this -


http://www.analog.com/en/education/education-library/dh-designers-guide-to-instrumentation-amps.html


Regards, Dana.

Thank you for your reply!
Actually, I'm not familiar with this IA, lots of new information to read! I'll read it and try to give a feedback!
But, isn't as an op amp with a buffer-built in?

Regards,
Tommy