3
\$\begingroup\$

For my project, I need to measure AC current for a 220V load (a more detialed description of the initial approach and problems can be found here.

I'm using an ACS724 for +/-5A; the next step will be to feed its output into a RMS-to-DC converter, more specifically an Analog Devices LTC1966.

With a DSO I tried to get an idea of the signal that comes from the sensor, and I get this :

No load connected
(all wires entirely disconnected from the board)
With load connected
enter image description here enter image description here

Which is pretty much as expected, except for the noise. The measurement for the 'load connected' case seems to be the sum of the right sine wave for the load and the 'no load connected' noisy signal.

  1. Is this normal?
  2. Do these sensors need additional signal conditioning?
  3. If yes, how?
  4. Is it possible to digitally compensate for this noise once the signal is measured by an ADC?
\$\endgroup\$
2
  • 1
    \$\begingroup\$ 1. is it normal? The datasheet has a noise specification that quantifies what you can expect. \$\endgroup\$
    – Kartman
    Feb 17, 2021 at 8:57
  • \$\begingroup\$ I'm not sure which parameter tells me that.It's in the 'Common electrical characteristics' on page 5 ? \$\endgroup\$
    – stefanu
    Feb 17, 2021 at 10:57

2 Answers 2

3
\$\begingroup\$

I'm writing this as an answer because I want to add images, but the solution was provided by Lorenzo Marcantonio in a comment :

The ACS has a declared noise of 20mA (about 16mV) with 4.7nF of filter capacitor. This reduces the available BW to 18kHz. The polulu board seems to be filtered with only 1nF (and space for another cap) so noise would be probably be higher. It doesn't seem you need the full bandwidth so you could simply use that. – Lorenzo Marcantonio Feb 17 at 11:55

I have tried several capacitors : 4.7nF, 10nF and 100nF (in addition to the existing soldered capacitor). The results improved as the value went up :

10nF + 1nF 100nF + 1nF
enter image description here enter image description here
enter image description here enter image description here

I'll definitely go for the higher value (100nF).

Thank you all for your help.

\$\endgroup\$
2
  • \$\begingroup\$ If you are still using this forum you should come back and accept this answer. Nothing wrong with accepting your own answer. \$\endgroup\$
    – user57037
    Oct 24, 2022 at 4:47
  • 1
    \$\begingroup\$ @mkeith indeed, you are right; I completely forgot to do that. \$\endgroup\$
    – stefanu
    Oct 25, 2022 at 5:14
0
\$\begingroup\$

It seems you have 50 Hz noise entering the circuit. The 20ms period tell me that.

You should post a schematic of how is the sensor connected to your circuit. The Allegro has an output impedance of 1.8kOhm which is not very low so the wiring should be very short since it seems that your module has no buffer amplifier. Try with a piece of shielded wire and see if it gets better.

Alse the LTC converter has a sampling capacitor and an input impedance model that are quite susceptible to loading. Read carefully page 26 of the datasheet, even attaching the scope probe could load it.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ There is no circuit yet; just the sensor board powered at 3.3V (same from my previous question). This is the output of the Hall sensor. The 50Hz sine wave what I get when the AC load is connected to the sensor, but this is correct behavior, right ? \$\endgroup\$
    – stefanu
    Feb 17, 2021 at 10:55
  • \$\begingroup\$ If it's an AC load yes, it is \$\endgroup\$ Feb 17, 2021 at 11:42
  • 1
    \$\begingroup\$ Yes, it is; it's an old light bulb, so I can get a higher current. I'm worried about the noise. \$\endgroup\$
    – stefanu
    Feb 17, 2021 at 11:48
  • 1
    \$\begingroup\$ The ACS has a declared noise of 20mA (about 16mV) with 4.7nF of filter capacitor. This reduces the available BW to 18kHz. The polulu board seems to be filtered with only 1nF (and space for another cap) so noise would be probably be higher. It doesn't seem you need the full bandwidth so you could simply use that. \$\endgroup\$ Feb 17, 2021 at 11:55
  • \$\begingroup\$ First tests : I only had a 100nf capacitor available, but the result is already better. Will try the suggested value, too. \$\endgroup\$
    – stefanu
    Feb 19, 2021 at 7:08

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.