0
\$\begingroup\$

I am a complete beginner in op-amps. The signal I want to amplify is a triangular signal that goes from -50mV to 50mV. For the amplification I have chosen the LTC6228 because of the low noise level.

The first op-amp is an impedance converter and the next three should do the gain (planned was 3x a gain of 3) and the offset. At the output (Vout4) I want to have my signal from 0V to 3V.

My circuit is shown below.

Now I have the following problem: why does the gain shift to the negative range and not fluctuate around the 0V line?

Circuit:

enter image description here

Vout:

enter image description here

\$\endgroup\$
2
  • \$\begingroup\$ Is there a reason you've cascaded three op amps with a gain of 3, for a total of 27, rather than one opamp with a gain of 27? \$\endgroup\$
    – LordTeddy
    Commented May 15, 2023 at 16:22
  • \$\begingroup\$ If you want low noise and high input impedance and good gain, I'd go with an instrumentation amplifier. They can have GigΩ input impedance. \$\endgroup\$
    – Aaron
    Commented May 15, 2023 at 17:24

1 Answer 1

3
\$\begingroup\$

Low noise op-amps often have rather high bias currents, for reasons I won't go into here.

Further, you've left the shutdown pin open, which allows the op-amps to function but disables bias current cancellation.

Bias current can be as much as 44uA, and multiplied by the 20kΩ feedback resistance gives us as much as 0.88V input-referred offset in the first amplifier stage alone. Multiply that by the gain.. and add the contribution from each stage. Typical at room temperature is 16uA.

One might also question the accuracy of the simulation which seems to be outputting considerably more than the supply voltage.


Where noise is concerned you really need to include the current noise from the input. The amazing low voltage noise may not look as good when that is considered, unless your source impedance and feedback resistances are very low. Bias current cancellation reduces the bias current but does not reduce the noise, it increases it (think of two noisy current sources fighting each other and you see the difference). At source (and feedback) resistances over a kΩ or two the current noise will dominate.

I suggest you quantify the noise level you can tolerate and attempt to design to that number. Picking op-amps that are highly optimized for noise (or any other single parameter) because you have a concern about noise may result in unnecessary problems and cost/availability issues. If you are measuring a 50mV signal with a 10 or 12-bit ADC and moderate bandwidth, noise is not going to be a huge issue, for example.

Similarly, if you were to say that you want an op-amp with the lowest available bias current since you were burnt with it in this case, you'd end up with an expensive drifty high-offset noisy op-amp with fA of input bias current. Nice if you're making an electrometer, but not required in most applications, and lots of performance drawbacks.

The main reason to cascade multiple stages (as you've done) is to get higher bandwidth with a given op-amp, otherwise you are adding drift, cost and power consumption unnecessarily. If you can do it with a single stage it's generally better.

\$\endgroup\$
3
  • \$\begingroup\$ Thank you for your detailed answer! " Further, you've left the shutdown pin open, which allows the op-amps to function but disables bias current cancellation." So I have to supply the shutdown pin with 350mV to activate bias current cancellation? " One might also question the accuracy of the simulation which seems to be outputting considerably more than the supply voltage." You are right, I wanted to show the shift with it. \$\endgroup\$
    – Hally
    Commented May 16, 2023 at 16:10
  • \$\begingroup\$ "I suggest you quantify the noise level you can tolerate and attempt to design to that number." Quantifying the noise is difficult because I have no data on my input other than the -50mV to 50mV. The input is still under development. "The main reason to cascade multiple stages (as you've done) is to get higher bandwidth with a given op-amp, otherwise you are adding drift, cost and power consumption unnecessarily. If you can do it with a single stage it's generally better." Good point, I reduced the resistances to Ω and there is alsmost no drift. \$\endgroup\$
    – Hally
    Commented May 16, 2023 at 16:13
  • \$\begingroup\$ I think you're using the wrong op-amp. Unless you have very good reasons use one with lower inherent input bias current and fewer stages. You must demand or develop specifications or you can't make a competent design. \$\endgroup\$ Commented May 16, 2023 at 16:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.