AD623 in single power - output saturates

[ericgibbs]

I'm lost. What is our bone of contention here?

hi Ron,
No bones, I just want to clear my thinking on this, if I am wrong about about something I like to know why.
I use these AD623 for other jobs, so if there is some limitation I want to know.

The OP is asking if an Vref will solve his problem, I dont think it will it, but I want to be sure thats all.

Tried it again with 100mVref, it now clamps at about 0.6V.

If I have caused offence, I'm sorry, that was not my intention.

 

[Roff]

hi Ron,
No bones, I just want to clear my thinking on this, if I am wrong about about something I like to know why.
I use these AD623 for other jobs, so if there is some limitation I want to know.

The OP is asking if an Vref will solve his problem, I dont think it will it, but I want to be sure thats all.

Tried it again with 100mVref, it now clamps at about 0.6V.

If I have caused offence, I'm sorry, that was not my intention.

No offense (offence), I'm just not sure what we disagree on, if anything.
Attached is the sim I ran.

 

[ericgibbs]

No offense (offence), I'm just not sure what we disagree on, if anything.
Attached is the sim I ran.

hi,
Looked at the zip, waves.png... the graph you got with the Sim does not reflect the Vout results from the actual circuit when constructed.
It still clamps at lower Vout, even with an REF of 100mV.

I would ask 'krmt' if he could use a REF of 100mV, with pin 2 not connected to 0V and see what he gets on Vout

The only difference, 'I think' we may have is the meaning of the datasheet.
I wonder if anyone else has experience of this 'problem'.

BTW: The Britway of spelling is 'offence' , quite often we use a 'c' where the States use an 's'.

 

[audioguru]

The datasheet shows a hex and a triangle that shows how low input voltages result in low clamped output voltages.

 

[ericgibbs]

The datasheet shows a hex and a triangle that shows how low input voltages result in low clamped output voltages.

I agree and the simple tests I have done confirm this.

I have advised the OP that he should change from AD623 to MCP6291/2, they are also much cheaper.

 

[Roff]

hi,
Looked at the zip, waves.png... the graph you got with the Sim does not reflect the Vout results from the actual circuit when constructed.
It still clamps at lower Vout, even with an REF of 100mV.

Did you try it with the inputs biased up to 1.8V, as in my sim? I'm having trouble believing their spice model would be that wrong. Am I interpreting the Fig. 23 (below) wrong?

I would ask 'krmt' if he could use a REF of 100mV, with pin 2 not connected to 0V and see what he gets on Vout

The only difference, 'I think' we may have is the meaning of the datasheet.
I wonder if anyone else has experience of this 'problem'.

BTW: The Britway of spelling is 'offence' , quite often we use a 'c' where the States use an 's'.

I know, that's why I spelled it both ways. Just funnin'.

 

[ericgibbs]

Did you try it with the inputs biased up to 1.8V, as in my sim? I'm having trouble believing their spice model would be that wrong. Am I interpreting the Fig. 23 (below) wrong?

I know, that's why I spelled it both ways. Just funnin'.

hi Ron,
No, I didnt bias the inputs by 1.8V, to be honest, I missed seeing that in the data you posted.!

But looking at the datasheet it would give the results you got.

The English language has much to answer for.

 

[Roff]

I agree and the simple tests I have done confirm this.

I have advised the OP that he should change from AD623 to MCP6291/2, they are also much cheaper.

He may still have to offset the output to avoid clipping when the input is 10uV. He will also need to have an input offset voltage adjustment, and he will lose the common mode rejection that an IA will provide (if he needs it).

 

[krmt]

He may still have to offset the output to avoid clipping when the input is 10uV. He will also need to have an input offset voltage adjustment, and he will lose the common mode rejection that an IA will provide (if he needs it).

Thanks once more,
you are all helping me a lot! Unfortunately, we came to a point, where I cannot follow anymore :-/

I switched to a 2 stage design, so that i could decrease the amplification for the input at the first stage from 1000 to 50. Do you advise doing so?
ok, I set the REF to 100mV as suggested and do get quite good results for inputs higher than 300µV. Unfortunately, not for my desired lowest input of 10µV. Apparently the REF offset helps, i dont understand why, though. (i dont have negative inputs, what do i need the offset for?!)

I couldn't find a multisim model for the MCP6291 you proposed, i try to import it the spice model...

The more you help, the more I think I should redesign from scratch. I'm not very experienced in that field... So how would you amplify low level signals (10µV...4mV) by a factor of 1000?! Can I stick with parts of my ideas or is switching to the MCP6291 the overall key? (Could you find time to have a look at the AD8223?)

Sorry, for bringing more questions than light into the problem...

 

[Roff]

Thanks once more,
you are all helping me a lot! Unfortunately, we came to a point, where I cannot follow anymore :-/

I switched to a 2 stage design, so that i could decrease the amplification for the input at the first stage from 1000 to 50. Do you advise doing so?
ok, I set the REF to 100mV as suggested and do get quite good results for inputs higher than 300µV. Unfortunately, not for my desired lowest input of 10µV. Apparently the REF offset helps, i dont understand why, though. (i dont have negative inputs, what do i need the offset for?!)

I couldn't find a multisim model for the MCP6291 you proposed, i try to import it the spice model...

The more you help, the more I think I should redesign from scratch. I'm not very experienced in that field... So how would you amplify low level signals (10µV...4mV) by a factor of 1000?! Can I stick with parts of my ideas or is switching to the MCP6291 the overall key? (Could you find time to have a look at the AD8223?)

Sorry, for bringing more questions than light into the problem...

If you really want to get an output of 10mV for an input of 10uV, you need dual (positive and negative) supplies. Can you do that?
What sort of sensor (or whatever) are you using? How long are the wires from the sensor to the amplifier?