4
\$\begingroup\$

I have an issue with a noisy VR-sensor signal that my MAX9926 IC isn’t managing to interpret correctly, and instead outputs rubbish. The wheel that’s being used has one long tooth and one short.

Here is the circuit. VR1 is the channel used in this case. enter image description here

Here is a trace made today with my oscilloscope. Yellow is the VR sensor. Green the output of the MAX9926. The red dots are the actual edges on the encoder wheel that’s supposed to be picked up. Blue lines are gaps between the teeth. enter image description here

As seen in this, more zoomed in scope, the noise is limited to about maximum 200mV. enter image description here

I have tried to reduce the R4 to 1Kohm, in an attempt to "load down" the VR-sensor and rid some of the noise. Didn't have any affect at all.

I reach out to you to help me with suggestions on the circuit to filter out the noise or configure the MAX9926 in a better way to handle the signal. Thanks in advance.

Update 2020-03-10: I have added a 400 ohm resistor in parallel with C8 in an attempt to pull the voltage so low, that the MAX9926 doesn't react on it. It kind of worked. But as you can see, the square wave is time shifted from the VR very randomly. And the actual teeth are not translated into square wave.

enter image description here

enter image description here

enter image description here

Here I have made an illustration in RED, as what the actual teeth on the wheel look like, in relation to the YELLOW VR-sensor signal: enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ It looks like the output of a differentiator, so have you thought about integrating the signal? It should (might) give you back the pulses, and filter the high noise a bit. \$\endgroup\$
    – a concerned citizen
    Commented Mar 9, 2020 at 19:02
  • \$\begingroup\$ Never thiught about it, never done it. =) Do you have any example circuit or similar? \$\endgroup\$
    – Tobias Rådenholt
    Commented Mar 9, 2020 at 23:40
  • \$\begingroup\$ Added some more troubleshooting above. \$\endgroup\$
    – Tobias Rådenholt
    Commented Mar 10, 2020 at 2:02
  • \$\begingroup\$ I think your rotational speed is too slow, and the adaptive threshold watchdog is kicking in before the next tooth pulse is received. The timeout for the watchdog is ~85ms, which seems to be when the unexpected output pulse occurs in your first plot. \$\endgroup\$
    – elchambro
    Commented Mar 10, 2020 at 4:59
  • \$\begingroup\$ I meant a simple integrator, with an opamp, which should try to reconstruct the red trace in the last picture. But it's a brute-force approach, it doesn't touch the problem, what and why is it generating the signal as it is. Besides, a fixed time constant might be troublesome for a variable speed (higher frequencies pulses will be mangled). \$\endgroup\$
    – a concerned citizen
    Commented Mar 10, 2020 at 16:26

1 Answer 1

Reset to default
2
\$\begingroup\$

Let me first thank you for your help and suggestions.

After a bit of trial and error, I found a solution. I added more resistance to the shunt resistor R4 and also added a resistor in parallel with the cap C8. This lowered the voltage overall, which made the noise so small, the MAX-chip doesn't pick it up at all. The proper signal is still of enough amplitude that it is picked up properly.

\$\endgroup\$

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.