LMC6462AIM:- voltage offset issue

I was test a board containing LMC6462AIM as high impedance buffer, +ve
pin to 1K to 0V (for test), wire link between -ve to output
(gain=1V/V). Output load 100K. The power is +/-5V from linear bench
PSU. The room temp is 25C and the PCB is very clean, no residual flux
or dirts.

Now using precision DVM, I measure 0.0mV at the +ve pin. I then measure
the output and found to be -2.1mV.

According to the datasheet the maximum offset should be 1.2mV and
typical being 0.6mV, which is why I choose this device. I was
surprised to measure this. I tried different DVM and the result is the
same, the voltage offset exceed the manufacture.

I tried other op-amp within the same batch and it has similar results.
It must be bad batch but it leave me wonder how they occurs, why
National fails to rejects them and how to avoid this happening again.

 

Put a scope on it, see if there is a little fuzzy
oscillation that the DVM is averaging to a dc.

Conversely, if the dc offset gets better when you
put the scope on it then you *had* an oscillation.

If there was an oscillation, try a 100uA to 1mA
pulldown current on the output stage.

 

Riscy,
The input offset voltage Vos is defined as the differential voltage
that must be applied to the input to achieve 0V output. However, the
Output voltage that appears with 0V input is equal to:
~
Vo = Vos(1 + Acl)
~
Where Acl is the closed loop gain of the amplifier. In your case, Acl
= 1. So the maximum output voltage that you can expect is:
~
Vo = 1.2(1+1) = 2.4.
~
This is in the range that you would expect for this device.
Regards,
Jon

 

Riscy,
Oops! My previous reply was wrong. The equation does not apply to the
unity gain buffer.
Regards,
Jon

 

Darn, I was getting ready to jump all over that one.

John

 

John,

You'll probably get your chance after my next post

Regards,
Jon

 

Jon,
Just as happen when I read you 1st message...I nearly dropped by coffe
and missed my heart beat until I saw the 2nd message(!)...whew, you
really had me knotted.

By the way, I found there is no ac on waveform, it pure DC.

 

Okay...no response since the question, I try AD822 or similar device
with lower offset and see the difference AD may have better quality
control than NSC but that only the guess.

 

In your last post you said there was no AC, only DC. Were you measuring the
DC output with the precision voltmeter, at the same time that you were
looking at it with the scope? Pay attention to Tony Williams' post,
"Conversely, if the dc offset gets better when you put the scope on it then
you *had* an oscillation."

You're powering it from a bench supply, presumably several feet of test lead
away from the circuit. Do you have bypass capacitors around the opamp? The
LMC6462 has a CMOS output stage; the AD822 is bipolar. That will tend to
make the LMC6462 less stable, I think, especially if the load is at all
capacitive.

Despite your claim that there is no AC, there must be at least some, because
of noise. The LMC6462 is quite noisy, around 100nV/rt-Hz between DC and 1k.
The AD822 is quieter, at around 20nV/rt-Hz in that range. But even the
AD822 is specified to have 2uV of broadband noise between 0.1Hz and 10Hz, so
the LMC6462 is probably worse; how long did you average your measurement
over?

Did you wait for the opamp to temp-stabilize before taking your
measurements?

"There is no such thing as DC."