0
\$\begingroup\$

I am quite novice in electronics, but I have a great interest.

I am analyzing an analog signal coming as input to a speaker.

The purpose is to introduce that signal (after its corresponding leveling) as input to an ADC pin of an ESP32 to simply detect when it carries a signal and when it does not carry a signal.

When the signal is at rest (off), when measuring with a multimeter, it gives me a value of 1.75v and 50Hz, as you can see:

enter image description here

However, when I analyze the same signal at rest with an oscilloscope, it draws me a sinusoidal waveform in the range from -800mv to 17v (!!!):

enter image description here

When the signal carries audio, it can fluctuate between -20v and + 20v.

My doubts:

  1. How is it possible so much difference from one measurement tool to another?
  2. What is the most correct for the signal at rest, 1.75v or 17v?
  3. If I connect that same signal (assuming the values ​​of the signal at rest) directly to the ADC pin of ESP32, would it burn it?

Note:

  • The probe terminal of the oscilloscope is set to 10x, as the oscilloscope config.
\$\endgroup\$
7
  • 2
    \$\begingroup\$ is the probe a x100 by any chance? The scope should have a squarewave test point. Connect the probe to it and confirm the magnitude. \$\endgroup\$
    – user16222
    Nov 30, 2021 at 22:26
  • 1
    \$\begingroup\$ There seems to be a factor of 10 difference. So which measurement is correct? I would seek out a similar signal of known voltage to confirm your measurement methods. Eg a mains transformer. That should answer #1 and #2. Would it ‘burn’ the esp32? Depends on your input circuit. I wouldn’t apply an external signal to the esp32 without suitable input protection. \$\endgroup\$
    – Kartman
    Nov 30, 2021 at 22:29
  • 1
    \$\begingroup\$ Be sure to measure it with the speaker connected. Otherwise, it seems you're picking up a lot of AC noise! Probably have a long wire. If you wont connect the speaker, atleast put a decent resistance (100 ohm) at the point of measurement. If you'll have a 20V signal, that resistor will be dissipating 2W, mind you. Even a 1k ohm resistor will reduce noise some. At 20V, you'll only dissipate 0.4W with that. \$\endgroup\$ Nov 30, 2021 at 22:50
  • 1
    \$\begingroup\$ One way this is possible is if the AC signal has a high source impedance, e.g. through the inter-winding capacitance of a transformer (which is a few hundred pF) The scope has 10M input impedance, but is the DMM 1M or 10M? If 1M it will attenuate such a signal much more than the scope. \$\endgroup\$
    – user16324
    Nov 30, 2021 at 23:47
  • \$\begingroup\$ @JonRB Hi, the prove is at x10 (is the first thing I checked) and calibrated. \$\endgroup\$
    – beer73
    Dec 1, 2021 at 7:28

5 Answers 5

1
\$\begingroup\$

How is it possible so much difference from one measurement tool to another?

The meter is trying to measure RMS (or something close to it), it could also have some issues with clipping the signal (since your range of the DMM is not shown). Another problem that could be throwing the meter off is the DC offset of the signal.

If you take the oscilloscope peak to peak values of 21.2 and 16.6 and subtract them, you get 4.6.

\$ V_{pkpk} = 2.828 V_{RMS}\$

so if you take the value of 4.6 and divide it by 2.828, you get 1.626Vrms which is close to what the meter is getting, which means the meter is probably measuring AC coupled version of the signal. If you turn on AC coupling on your oscilloscope you could see this.

It looks like Either way, I'd trust the oscilloscope over the DMM.

What is the most correct for the signal at rest, 1.75v or 17v?

The signal is never 'at rest' because of it's AC content, it also has a DC offset.

If I connect that same signal (assuming the values ​​of the signal at rest) directly to the ADC pin of ESP32, would it burn it?

Yes it would, it exceeds the voltage range of the ADC pin several times, don't do this, use a voltage divider.

\$\endgroup\$
4
\$\begingroup\$

You are reading mostly 50Hz mains hum, which will be different when measured with a battery powered multimeter vs a grounded oscilloscope. If you connect both at the same time they will be more congruent most likely (keeping in mind that peak-to-peak is almost 3x RMS (what the meter reads in).

Whatever is going on seems like you have an incorrect reference point for the measurement. If there was even a 1.7V 50Hz voltage directly across the speaker the hum would be intolerable.

\$\endgroup\$
2
\$\begingroup\$

Change the scope to AC coupling and it should read the same as the meter.

Change the scale on the meter to DC and it should read the same as the mean value displayed on the scope.

The meter is reading just the AC component. The scope is reading the combined AC + DC. Neither is wrong but both are useful.

\$\endgroup\$
1
\$\begingroup\$

Generally you would not connect a speaker level signal to MCU. Speakers may be driven in bridge configuration so neither speaker terminal is ground. But it is unclear if you mean a speaker level signal or line level signal as used with RCA or 3.5mm connectors.

  1. The multimeter is handheld/isolated/battery powered device, and the oscilloscope is a mains-powered mains-earth referenced equipment. Or at least hopefully the scope is earthed. They may also have wildly different input impedance. Most likely cause is you have a bad connection or just measuring things incorrectly.

  2. Most likely neither. A signal at rest sent to speaker should have 0V and 0 Hz. Which makes even more likely that your measurement setup is not correct. Likely a bad or missing ground connection, and you most likely live in a 50 Hz mains country so you see 50 Hz mains common mode hum with the earthed scope. If the scope is not earthed, it can even cause the hum.

  3. Definitely that signal would burn a MCU input pin, without any other knowledge about the signal.

\$\endgroup\$
1
\$\begingroup\$

What is the most correct for the signal at rest, 1.75v or 17v?

None of the above, but the scope is telling you something about a single-ended output, while the meter is in AC mode so it blocks DC. Yet the scope shows a DC offset perhaps from a single supply but clipped asymmetrically so the input may have DC offset floating.

Perhaps it does not agree because it was connected at a different times or the scope ground affects the output.

Also at "REST" it should be 0Vdc or mean=0V and 0Vac pp

The scope is more accurate, but you need to define the audio and probe impedance where ground=0V is defined for any measurement to trust it's results.

Trust but verify by using the DSO calibration pin.

You can also use two probes to measure a differential or bridge output using Math or A-B mode or A+B inverted.

\$\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.