1
\$\begingroup\$

I have a 0-12V voltage range (Vin) that I need to transform to a 100mV-900mV voltage range (Vout). The 0-12V input is fixed from another system, and my analog circuitry needs the 100mV-900mV for proper operation of the IC to which I must supply this voltage. I have tried following this guide (Application Report SLOA097 from TI) called Designing Gain and Offset in Thirty Seconds. I have calculated the linear transformation to be Vout(Vin) = m * Vin + b => Vout(Vin) = 66.66mV * Vin + 100 mV. This works out fine. A few discrete values as examples: Vout(0) = 100mV, Vout(3) = 300mV, Vout(6) = 500mV, Vout(9) = 700mV & Vout(12) = 900mV, which maps the input range to the output range perfectly. My problem comes from the actual implementation of the circuitry. The TI app note specifies a biased voltage divider and non-inverting amplifier circuit combination for positive m and positive b. My transformation has m = 66.66mV and b = 100 mV, so these are both positive.

Circuitry

Using a seed value of R1 = 10kOhm and a Vref of 3.3V, I calculate R2 = 22kOhm. Using a seed value of Rf = 10kOhm gives a Rg value of -11kOhm. I haven't been able to purchase a -11kOhm resistor anywhere :) I guess the issue arises because my m value is between 0 and 1, i.e. non-inverting but attenuating. I am looking for circuit implementation suggestions to solve this problem, any recommendations are greatly appreciated.

\$\endgroup\$

1 Answer 1

3
\$\begingroup\$

You need another resistor (from the non-inverting input to ground) if attenuation is required. Call that Rs.

Let's pick R1||R2||Rs = 10K

Then we can immediately write:

(12V/R1)*10K = 0.8V so R1 = 150K

The output will be zero when the input is -(1/8)*12V or -1.5V.

So Vref/R2 = 1.5/150K by KCL. Say Vref is 5V then R2 = 500K (499K is closest E96 value)

And finally Rs is 1/(1/10K - 1/150K - 1/500K) = 10.95K.

(and Rf = 10K, Rg = open) -- Rf = 10K cancels out offset component from input bias currents if they are equal.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
6
  • 1
    \$\begingroup\$ This works perfectly! Thank you very much. \$\endgroup\$
    – Lunde
    Commented Aug 20, 2020 at 6:16
  • \$\begingroup\$ I have to expand the range from 100mV-900mV to 50mV-950mV. I have tried to redo the math, but I struggle with the step: "The output will be zero when the input is -(1/8)*12V or -1.5V." Where does 1/8 come from and what changes when the range is expanded? Thank you in advance! \$\endgroup\$
    – Lunde
    Commented Aug 24, 2020 at 10:09
  • \$\begingroup\$ It’s linear so calculate the intercept (theoretical). In this case it had to go down 100mV and the gain is 12V:800mV. This makes the calculation much easier because Rs disappears and it’s substitution rather than solving a system of two equations in two unknowns. \$\endgroup\$ Commented Aug 24, 2020 at 10:47
  • \$\begingroup\$ The linear transformation, for a 50mV-950mV range, becomes Vout(Vin)=75mV+50mV. The intercept must then be Vout(Vin)=0 => Vin = -0,667V. Redoing the math, as per your previous equations, give me R1 = 133kOhm, R2 = 1 MegOhm and Rs = 10K9 Ohm. I dont see how Rs disappears as I still need attentuation, as in the previous case? \$\endgroup\$
    – Lunde
    Commented Aug 24, 2020 at 13:23
  • \$\begingroup\$ I mean that Rs disappears for the calculation of R2. Then Rs drops out from the parallel combination resistance. Because the - input = + input = output = 0V, then Rs is not doing anything much at that particular special point only. \$\endgroup\$ Commented Aug 24, 2020 at 15:04

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.