1
\$\begingroup\$

The circuit below is from a TI reference design, but my question is generic. I'm generating a voltage reference to add a DC bias to an AC signal for ADC purpose. Could I use the output of the voltage reference IC, in this case LM4041, directly instead of making it passing through an op amp buffer? How do I decide that?

enter image description here

\$\endgroup\$
4
  • 4
    \$\begingroup\$ It depends on how much current you want it to supply \$\endgroup\$
    – DKNguyen
    Commented Mar 27, 2020 at 17:59
  • 3
    \$\begingroup\$ Generally you want to use buffers if loading would cause an issue. The ADC input may present a dynamic load when sampling. A high speed buffer will be able to satisfy the transient current spikes the sampling action causes. The reference IC may not be able to adjust it's output current quickly. Additionally, if you have the reference physically far away, a buffer close to the ADC pins will not have as much inductive parasitics, again, allowing for high bandwidth transients current draws to be met . \$\endgroup\$
    – Yet Another Michael
    Commented Mar 27, 2020 at 17:59
  • 1
    \$\begingroup\$ @Michael - You should be putting your comment here into an answer box. Comments here are meant to help to clarify the question, not answer it. \$\endgroup\$
    – Michael Karas
    Commented Mar 27, 2020 at 20:17
  • \$\begingroup\$ @MichaelKaras I can't convert a comment to an answer, the best thing to do is to ask the user of the comment to do so. \$\endgroup\$
    – Voltage Spike ♦
    Commented Jun 3, 2020 at 22:53

1 Answer 1

Reset to default
2
\$\begingroup\$

Could I use the output of the voltage reference IC, in this case LM4041, directly instead of making it passing through an op amp buffer?

Not if you want low noise and some decent degree of accuracy. For low noise note the RC on the output of the op-amp - it has a low pass frequency of about 16 kHz so, any noise above that coming from the op-amp or voltage reference is progressively attenuated. There's no reason why the 100 nF couldn't be 10 uF (if the op-amp can handle the RC without oscillating) and get a 3 dB noise point of 160 Hz (even better for noise).

So, given that the LM4041 produces circa 20 uV RMS of wide-band noise it seems sensible to use an RC circuit.

I'm not a big fan of op-amp buffers because they can bring all sorts of errors to the party but the RC is a must in my book because it simultaneously keeps noise down and offers a low impedance to the ADC at frequencies above where the raw output from the 4041 can't deliver low impedance (look at figures 7 and 8 in the TI data sheet).

\$\endgroup\$

Your Answer

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