1
\$\begingroup\$

Regarding my previous question and this paper, I wanted to simulate the circuit below to see whether it is rejecting common mode voltage as claimed. The paper says that sharing DAC's Vref with non-inverting node, the common mode voltage variation at Vref will not effect the output:

enter image description here

To try that, I first modeled the DAC and tried the following circuit in LTspice:

enter image description here

But I get the following plots for (Voutp - Voutn) and Vcm; showing that Vcm appears at the output:

enter image description here

Another similar variant of this circuit is as follows(with voltage divider):

enter image description here

And now I removed the ground of the DAC model and voltage divider to the non-inverting input as follows:

enter image description here

And now the output is quiet small:

enter image description here

I couldn't figure out what I am doing wrong. Or is my last try correct way to do this? How can this circuit be simulated in SPICE so that we can see the common mode error is eliminated as claimed?

\$\endgroup\$
7
  • 1
    \$\begingroup\$ Removing the ground will leave an open node, which works sometimes but sometimes introduces numerical instability. Technically there is a fp cap on those nodes but best to also put some resistance on it. \$\endgroup\$
    – Voltage Spike
    Apr 22, 2022 at 18:08
  • 2
    \$\begingroup\$ Does anything change if you use [Opamps]->UniversalOpamp2 instead of the LM324? \$\endgroup\$
    – Ste Kulov
    Apr 22, 2022 at 20:35
  • 1
    \$\begingroup\$ The opamp you're using for the DAC is nothing but a G+(R||C). \$\endgroup\$ Apr 23, 2022 at 8:22
  • \$\begingroup\$ @SteKulov No it didn't make any difference. Im looking for someone to simulate what that paper claims. \$\endgroup\$
    – ty_1917
    Apr 23, 2022 at 19:33
  • \$\begingroup\$ @ty_1917. I'd like to kindly make a suggestion: why not providing a quote, print or formula from the app. note, where it specifically claims that "the common mode voltage variation at Vref will not effect the output:". It may help contributors to focus on that (proving or disproving, maybe). \$\endgroup\$
    – devnull
    Apr 23, 2022 at 19:37

1 Answer 1

0
\$\begingroup\$

I simulated this using LT1007A for all three opamps, using the voltage divider. I got 237uV P-P with the dividers grounded and 470uV P-P without them grounded.

My output is also centered on 0 V.

enter image description here

I suspect the problem may be in the opamp model you're using for your DAC model.

\$\endgroup\$
6
  • \$\begingroup\$ You set R1 R2 and DAC resistors equal. But if you vary them you will see huge common mode voltage appear at the output. But in reality it shouldn't be the case. That should be independent of R1 and R2 and also DAC output. I doubt the models we tried is correct so far. \$\endgroup\$
    – ty_1917
    Apr 22, 2022 at 18:18
  • 1
    \$\begingroup\$ The whole point is to cancel out the common mode voltage by adding Vref in the inverting amplifier, if you change the voltage ratio of course you'll get high common mode voltage at the output. Without the second divider you get 1V + (-0.5V) = 0.5V, with it you get 0.5V + (-0.5V) = 0V. \$\endgroup\$
    – GodJihyo
    Apr 22, 2022 at 18:31
  • \$\begingroup\$ Yes but that means the model is wrong. R2 can be a poti to change the range of the output. And imagine even R1 and R2 is equal when the DAC will vary(means in my DAC model its resistors will vary). That means this model would be useless unless it outputs zero always. I need a model that can replicate the real scenario. \$\endgroup\$
    – ty_1917
    Apr 22, 2022 at 18:37
  • \$\begingroup\$ When the DAC varies the common mode should not vary that much. DAC model like mine can be wrong as well. \$\endgroup\$
    – ty_1917
    Apr 22, 2022 at 18:40
  • \$\begingroup\$ Shouldn't U3 source come from U1? \$\endgroup\$ Apr 22, 2022 at 18:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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