0
\$\begingroup\$

I am trying to understand this op-amp circuit that came from one of our suppliers. The input on X9 connector is a speed sensor (magnetic pickup sensor.) Its voltage amplitude is proportionnal to a motor speed (up to 7500 Hz.)

From what I understand, the op-amp IC11 plays the role of a differential amplifier circuit. With the voltage dividers made of R282, R306, R283 and R305, there is an offset of 3V on both inputs (probably because it is a single supplied op amp) so the signal coming from X9 is shifted up. I have three questions:

  • What would be the input impedance of this circuit (seen from the sensor?)
  • What is the role of D515 and R281 + R318 + R308 ? I assume that D515 is here not to disturb the 5V supply but why are there resistors?)
  • Is Z1 acting as a voltage limiter? If so, are D516 and D517 not enough?

Circuit

Thanks!

\$\endgroup\$
5
  • 1
    \$\begingroup\$ Why not try simulating it? \$\endgroup\$
    – Andy aka
    Commented Oct 21, 2020 at 15:05
  • 3
    \$\begingroup\$ The LM2903 is a comparator. You should also note that the feedback is positive. \$\endgroup\$ Commented Oct 21, 2020 at 15:12
  • \$\begingroup\$ Its voltage amplitude is proportional to a motor speed (up to 7500 Hz). Given the circuit (and in particular the net name on the output node) I would expect that the input signal frequency is proportional to motor speed. \$\endgroup\$ Commented Oct 21, 2020 at 16:00
  • \$\begingroup\$ I can't simulate it today, but I will surely do it tomorrow. I didn't realise it was a comparator and a Schmitt trigger. What would be the threshold value? I am not sure about it. Why is the output bad ? @user287001 \$\endgroup\$
    – Ultra67
    Commented Oct 21, 2020 at 16:04
  • 1
    \$\begingroup\$ Here's an online calculator to figure that out: sim.okawa-denshi.jp/en/compkeisan.htm. As far as why the output is 'bad', maybe it isn't given that you thought it was a voltage out vs. a pulse out? \$\endgroup\$ Commented Oct 21, 2020 at 16:33

1 Answer 1

3
\$\begingroup\$

It’s a Schmitt trigger circuit. All that gobbledygook on the front end is there to protect the input, and to bias the op-amp (actually a comparator) inputs at a working midpoint.

The sensor signal is AC coupled to the (-) input through C786. It's biased at 3/5 VCC by R305/306, to a net impedance of 120K. R307 acts like a voltage divider with R305/306 and reduces the gain a bit, acting as a high-pass RC filter with C786. The actual input impedance is frequency-dependent and is about 195K at the right side of C786.

Going further up the input, R281/R318 and D515 form a biasing network. D515 is reverse polarity / leakage protection. Overall impedance will be a bit less than the divider you add in the downstream stuff: about 1K depending on frequency.

R308 and Z1 are input protection. Z1 limits the voltage swing to +5.1 / -0.6V. R308 limits the current to Z1 so it doesn't take the full brunt of an ESD spike.

D516/D517 further limit the input to VCC +0.3V and -0.3V. Being Schottky types they have faster response than the Zener Z1. Kind of belt-and-suspenders in my opinion.

On the comparator (+) side, there's the biasing network R282/283. Not only does it set the 3/5 VCC operating point, it works with R309 to make the hysteresis snap-action (Schmitt triggering) to square up the signal.

That all said, if the designer were that concerned about transients, they haven't taken care of the ground input. That's a major vulnerability. I would add a common-mode filter between X1 and rest of the signals.

\$\endgroup\$
8
  • \$\begingroup\$ Thank you very much for your complete answer. I have some comments: -Isn't R305 and R306 forming a bias at 3/5 of VCC ? 2/3 Would be with 100k and 200k resistors. -How do you find the 325kOhms and 2.5kOhms input impedance ? -How would you design this common-mode filter ? \$\endgroup\$
    – Ultra67
    Commented Oct 21, 2020 at 17:37
  • \$\begingroup\$ I updated my computations. The common-mode filter would use a type for a differential input; the sensor leads would feed through that. I’d further isolate the input by AC coupling both the ‘GND’ and signal side. \$\endgroup\$ Commented Oct 21, 2020 at 17:50
  • \$\begingroup\$ One more thing. If this is for a reluctor to sense position (like a crank angle sensor) the polarity matters as the signal isn’t symmetrical. \$\endgroup\$ Commented Oct 21, 2020 at 18:01
  • \$\begingroup\$ The application is a speed sensor for a motor. The sensor is magnetoresistive and the output voltage is proportional to the speed. I'm reading everything, trying to understand. Isn't there a mistake with R318 and R281 values that should be swapped? \$\endgroup\$
    – Ultra67
    Commented Oct 21, 2020 at 18:08
  • \$\begingroup\$ Ok, so the MR sensor is outputting pulses and requires a DC bias to operate. It’s being biased at 1.7V. (2/5 * 4.3V). Maybe not enough? Check the sensor spec. If you swapped the bias resistors then it would be biased at 2.6V and have a bit larger of a swing. \$\endgroup\$ Commented Oct 21, 2020 at 18:19

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.