13
\$\begingroup\$

I'm building a hobby oscilloscope on an ATmega16 microcontroller. The main problem is that I receive a large amount of noise while measuring the signal. I used LF353 amplifiers to shift voltage and I suspect that they might be causing the noise.

This is the schematic with an op-amp. The signal goes to 'Input' and the 'Output' goes directly to the microcontroller's ADC pin.

Schematic

And this is the signal that I'm getting when the leads are not connected:

Signal

\$\endgroup\$
  • 2
    \$\begingroup\$ Any chance you are picking up the mains frequency (usually 50 or 60Hz)? \$\endgroup\$ – jippie Jun 24 '13 at 6:47
  • 1
    \$\begingroup\$ What does the ATmega16 brochure say about impedances driving the ADC pin - usually MCUs require quite a low impedance at this point to keep noise down. What rate are you sampling at? Are you sampling any other input at the same time. Have you tried taking out the op-amp and grounding where the op-amp output would have been? What supply de-couplers on the op-amp have you used? \$\endgroup\$ – Andy aka Jun 24 '13 at 7:18
  • 2
    \$\begingroup\$ Any voltage rail that you apply directly to the analog signal path through resistances such as your R6 will have to be dead quiet to prevent noise problems. It also is obvious that using the 5V supply to bias the signal path the way that you are will result in the signal input to the A/D having a dependency on the absolute value of this supply voltage. I would think that you will want to rethink your design some so that the only thing directly biasing the signal path is the input voltage and the op-amp outputs. This way you remove the effect of variations of the (continued) \$\endgroup\$ – Michael Karas Jun 24 '13 at 9:02
  • 1
    \$\begingroup\$ (continued from above) supply voltage by a factor of the PSSR (power supply rejection ratio) of the op-amps used. Finally I think for best bandwidth support, as you refine your technique, you will want to drive your A/D input from as low of impedance source as possible. Your current source impedance is approximately 33K || 82K. This seems rather high and may need to be significantly lower if you plan to ever be trying to multiplex several channels in sequence. The last comment I'll make is that you should try to leverage separate AVCC and AGND pins on the MCU (continued) \$\endgroup\$ – Michael Karas Jun 24 '13 at 9:28
  • 1
    \$\begingroup\$ (continued from above) such that you use a separate filtered 5V and GND for the analog circuitry and connect them to the MCU GND right at the MCU. \$\endgroup\$ – Michael Karas Jun 24 '13 at 9:30
4
\$\begingroup\$

Any voltage rail that you apply directly to the analog signal path through resistances such as your R6 will have to be dead quiet to prevent noise problems. It also is obvious that using the 5V supply to bias the signal path the way that you are will result in the signal input to the A/D having a dependency on the absolute value of this supply voltage.

I would think that you will want to rethink your design some so that the only thing directly biasing the signal path is the input voltage and the op-amp outputs. This way you remove the effect of variations of the supply voltage by a factor of the PSSR (power supply rejection ratio) of the op-amps used.

Finally I think for best bandwidth support, as you refine your technique, you will want to drive your A/D input from as low of impedance source as possible. Your current source impedance is approximately 33K || 82K. This seems rather high and may need to be significantly lower if you plan to ever be trying to multiplex several channels in sequence.

The last comment I'll make is that you should try to leverage separate AVCC and AGND pins on the MCU such that you use a separate filtered 5V and GND for the analog circuitry and connect them to the MCU GND right at the MCU.

|improve this answer|||||
\$\endgroup\$
5
\$\begingroup\$

Perhaps adding at the DC offset at the input would give more controllable results. For example.

enter image description here

|improve this answer|||||
\$\endgroup\$
  • \$\begingroup\$ +1, After Michael's suggestion I've also thought about passing the offset voltage through the amp, rather than applying it to the ADC pin directly. I'm also feeding negative voltage (-5V) to the negative power supply of the op-amp through MAX1044 voltage converter instead of grounding the negative power supply. \$\endgroup\$ – Ashton H. Jun 25 '13 at 16:55
  • 1
    \$\begingroup\$ What is the purpose of R5? \$\endgroup\$ – johnfound Oct 26 '13 at 14:42

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.