0
\$\begingroup\$

When looking at INA topologies (lets take AD8220 for example) I see examples of integrator or buffered resistor dividers used. What are the disadvantages of using an integrator? Both typologies consume 1 op-amp. The only advantage of the buffer seems to be if you want to preserve DC information. If you are only looking at AC signals it seems like the integrator is the way to go with no disadvantages. Am I missing anything? Are there any design challenges with integrator if you are running off a single rail supply with a virtual ground?

Integrator (Datasheet Reference)

enter image description here

Buffer (Reference)

enter image description here

\$\endgroup\$

2 Answers 2

0
\$\begingroup\$

This INA has the unity gain Diff Amp Output stage with laser trimmed R's for high CMRR*.

Thus for single supply REF input = +Vs/2*

Since REF needs to go to a Zout=0 the AC coupled Op Amp in Figure 64 meets that criteria, thus +Vs/2 goes to the external OA +in = Vref.*

AC coupling for INA is mandatory for ECG , EEG applications due to galvanic skin response which create a pressure sensitive DC voltage. *

enter image description here

Did you catch all 4 points*?

\$\endgroup\$
5
  • \$\begingroup\$ Not sure I understand the first point. Would having non-unity gain at dif amp stage be a problem for integrator? \$\endgroup\$
    – EasyOhm
    Nov 11, 2019 at 21:51
  • \$\begingroup\$ That last sentence doesn't make much sense. AC coupling is often part of the output circuitry of the inamp, not the input. Also, you probably don't mean the galvanic skin response so much as simple junction potentials at the skin interface \$\endgroup\$ Nov 12, 2019 at 2:17
  • \$\begingroup\$ 1) I made 4* points to answer His question which was did I miss anything? 2) AC coupling with negative feedback becomes a low pass filter integrator so it makes perfect sense to be where it is 3) junction potentiometer of the skin modulated my pressure sensitive motion or vibration is technically called “galvanic skin response.” from my Biomedical education \$\endgroup\$ Nov 12, 2019 at 4:12
  • \$\begingroup\$ GSR is a change in the impedance of the skin, not a change of the skin/electrode junction potential due to motion artifact. One is a physiological measurement, one is an artifact. \$\endgroup\$ Nov 12, 2019 at 12:05
  • \$\begingroup\$ @ScottSeidman Thanks for the correction. the. Net effect of both is DC voltage interference that is not part of the signals in ECG,EEG thus the reason for DC blocking as done here with NFB integrator \$\endgroup\$ Nov 12, 2019 at 15:36
0
\$\begingroup\$

The integrator in this purpose is used similar as PI compensator. It will eliminate the low frequency signal. Something like moving average filter used in signal processing.

Are there any design challenges with integrator if you are running off a single rail supply with a virtual ground?

Yes they are. You will find a comprehensive explanation when looking at diamond chart. In few words, the signal is evaluated as a difference between + and - input, then amplified and then, only then, subtracted or added to ref signal. With high gain, you hit the rail voltage fast. Vout= G(Vin+ + Vin-) + Vref (more precisely: \$ G\cdot V_{in+} - G\cdot V_{in-} + V_{ref}\$), a saturation may occur very fast.

Further, you won't be able to subtract the moving average of integrator's output if you have a single supply, since the output of the integrator has to be negative. Or alternatively if you swap the inputs, the amplified signal has to be mean negative and integrator signal positive.

\$\endgroup\$
2
  • \$\begingroup\$ I am trying to understand the second paragraph. If I have a 5V supply and provide a virtual ground at 2.5V (mid-rail) then it seems to me the moving average problem is not present. Do I understand correct? \$\endgroup\$
    – EasyOhm
    Nov 11, 2019 at 23:28
  • \$\begingroup\$ @EasyOhm It would be better if you open another post with a real circuit problem. What you call it virtual ground (mid-rail) won't solve anything, because at last you end the same as single supply 5V and Vdd/2 = 2.5V. \$\endgroup\$ Nov 12, 2019 at 7:48

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.