0
\$\begingroup\$

I need to monitor a current on a PCB trace. The current is in sine form and its peak to peak 25A. I need to monitor it very precisely (+/- 50 mA) because I have to issue an interrupt output when the wave is at its peaks (maxima and minima points). Also I need to detect the frequency and issue another interrupt on 50 degree phase. I want to be sure because these chips are very expensive.

I fear of this propagation delay and dynamic response issue:enter image description here

Sine wave frequency will be varying between 500 Hz and 1 KHz. Can this chip keep up with the trace current in those frequencie and can it tell me real time current level precisely?

\$\endgroup\$
  • 2
    \$\begingroup\$ Precision is a very vague word. What precision do you want? That also depends on the ADC. Is +-1A ok? Or you need +-1mA? If you want cheaper, you could use a resistor. ACS711ELCTR-12AB-T/ACS711KLCTR-12AB-T are optimized for +-12.5A which is what you need. But total output error is +-5% as stated in the datasheet. Bandwidth is 100kHz so that is ok. \$\endgroup\$ – Andrés Jun 13 '17 at 16:56
  • 1
    \$\begingroup\$ Detecting the 50° frequency is a different task altogether as you'll need to do a PLL if you want precision. If you want it to run from 500Hz to 1kHz you'll need a very good and fast pll. For 1° accuracy (roughly, back of the envelope calculation), you would need to run it 360times per sine signal, or up to 360kHz. Are you sure you can do that? Cheap way, detect zero crossing and use that to measure frequency and estimate when your phase is 50°, but it could be way off. \$\endgroup\$ – Andrés Jun 13 '17 at 16:58
  • \$\begingroup\$ @Andrés thanks for the suggestion. I editted the Q as you suggested. I can tolerate +/- 50 mA I guess. \$\endgroup\$ – Alper91 Jun 13 '17 at 17:23
  • 3
    \$\begingroup\$ The 620-1483-1-ND chips in your link are only USD 1.73 each. Why don't you explain the application. You've locked your mind onto one solution. You have an army of experienced engineers here who may suggest a simple reliable means of solving the problem. My concern would be that noise on the waveform will give you false max / min points. \$\endgroup\$ – Transistor Jun 13 '17 at 17:36
  • \$\begingroup\$ @Transistor hi. A better one is 5.66 $ with 200 u Ohm path resistor. And I need to buy a lot of them. The application is for a 3 phase motor driver. 25 A current is phase current. Thats why I need to know phase difference because I will excite the phases with 120 degree (There are three phases, I want to drive them seperately. Maybe you can better imagine the issue from this other question of mine: electronics.stackexchange.com/questions/300650/…). \$\endgroup\$ – Alper91 Jun 13 '17 at 17:52
1
\$\begingroup\$

The quick answer is no.

If you examine the data sheet, you will discover two facts:

1) At +/- 31 amps full scale, the output voltage scale is 45 mv/amp.

2) Output noise is rated to be 11 mV.

This means that the noise component of the output signal is equivalent to about 250 mA, or 5 times greater than your desired accuracy.

Granted, this is a simplistic analysis, and accuracy can be improved by various means, particularly heavy filtering of the signal to exclude as much of the noise as possible. This, of course, will add to the delay which worries you.

However, your concern about delay is completely unfounded, since the figure which you show is not what you think it is. The figure simply shows how long it takes for the output to settle after power is applied to the chip, and since I presume you would be applying power continuously. But since you did not realize this, and were unable to appreciate the figure's caption, I'm dubious that you can compensate for filter delay.

So the longer answer is no, also.

\$\endgroup\$

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.