5A ACS712 "Locking up"

[rstor22]

I purchased a 5A ACS712 module similar to this one here: http://a.co/bs6iwAB

to monitor current of a 12V linear actuator with built in start and stop limit swtiches. Using the above module I wanted to know when the actuator has either extended or retracted fully (no current). I noticed that frequently during operation of the actuator the module will "get stuck" and report a consistent amount of current (around 0.18A), even when the actuator is completely disconnected. Resetting the Arduino still results in the same amount of current being reported (with the actuator disconnected from it) I would need to remove the power to the ACS712 module for it to operate correctly again however the same behavior reoccurs when measuring the current draw from the actuator (no load/force on the actuator). I read how magnetic fields can disturb the performance of the sensor so I re-positioned it several times to no avail.

One thought was that perhaps the actuator might be drawing too much current (>5A) however my power supply is rated at 3A and doesn't go into current limiting mode when started. When there is no load the actuator draws around 0.17A. It is rated at 0.65A.

I did try another motor which draws draws around 1A when running and over 3A during start which causes my power supply to go into current limiting mode however that motor works fine with the ACS712 module.

Here is what the waveform looks like when the actuator is running :

https://ibb.co/eJ0Fsy

I tried adding a 470uF capacitor in paralel with the motor and this is how it now looks:

https://ibb.co/eWfyCy

After adding the capacitor the problem is solved and the ACS712 does not "lock up" as described above. Why does the motor noise effect the ACS712 I thought it is supposed to provide isolation? What is happening here?

[IDEngineer]

...during operation of the actuator the module will "get stuck" and report a consistent amount of current (around 0.18A)...

I have a LOT of experience with the ACS series and I've never seen one "lock up". I presume you are aware that their outputs have a minimum output voltage? They don't generally go all the way down to zero even when no current is flowing, they have a bottom threshold of around 0.5V. Are you taking this into account?

I read how magnetic fields can disturb the performance of the sensor so I re-positioned it several times to no avail.

FYI there are differential versions of these parts that eliminate their sensitivity to external fields.

One thought was that perhaps the actuator might be drawing too much current (>5A)

Not likely to be the issue. One of my designs with these parts had a peak inrush exceeding 70A (!!!) and the parts continued to function perfectly. BTW the current path through these parts is rated to at least 5X the current measurement spec... the output will saturate but the device will live through it.

The current path through the package IS isolated, as you say, and the sensing side is tolerant of saturation. Something else is going on.

[rstor22]

They don't generally go all the way down to zero even when no current is flowing, they have a bottom threshold of around 0.5V. Are you taking this into account?

Yes I have taken this into account. When no current is flowing I get a specific value. When the linear actuator is active the value increases. It then "locks up" and gets stuck at a specific value. If I disconnect  the actuator the reported value remains (i.e. it doesn't go back down to the value indicated when no current is flowing). If I were to connect another current source to it while it is in this "locked" state the reported value doesn't change. I need to power down the module and power it back up for it to work correctly again. The only other solution I've found so far is to add the capacitor in parallel with the actuator and this behavior no longer happens. I have no idea why...

Perhaps the chip is bad or counterfeit?

[IDEngineer]

Perhaps the chip is bad or counterfeit?

I doubt it's counterfeit, this isn't the type of device that most counterfeiters are set up to copy and it probably doesn't move enough volume to be attractive.

To reiterate: I can't imagine anything you could do on the sense terminals that would have that effect on the active portion of the die. How's your power and ground going into the chip? Clean? Do you get horrible transients when you "spike" the actuator? Set up a scope to capture the power and ground near the chip and trigger off the actuator input power.

Bad chip, maybe. Certainly my next step would be to swap it out and see. I've never seen what you're describing and I've treated a lot of those ACS chips very harshly.    It would be interesting to debug but I suspect you'd rather be making progress on your actual project. So yes, swap it out.

[rstor22]

To reiterate: I can't imagine anything you could do on the sense terminals that would have that effect on the active portion of the die. How's your power and ground going into the chip? Clean? Do you get horrible transients when you "spike" the actuator? Set up a scope to capture the power and ground near the chip and trigger off the actuator input power.

I am powering the ACS712 module via the Arduino (which is powered through a desktop computer USB). The signal looks clean however when the motor is powered there is noise (see attached). Excluding the sensor completely and powering the actuator off a supply and measuring the +5V output on the Arduino still shows noise. Connecting the power supply (that is powering the actuator) to an isolation transformer still shows noise on the Arduino +5v line. Powering the actuator off a 12V SLA battery still shows some noise on the Arduino +5V line (that is still being powered via USB through a computer).  When I installed the capacitor (which solved the original issue) the noise on the +5V line on the Arduino is still present when the motor is running. 

I was able to source a 30A ACS712 (only the chip itself) from a local supplier that I feel is reputable. I removed the 5A chip and put in the 30A one and the issue is solved. Also the accuracy seems to be much better which is odd because the 5A module is typically 185mv/A where the 30A module is typically  66 mv/A. If I recall correctly with the 5A module 1A was reading around 750mA where on the 30A module it is reading 950mA. Also the noise from the motor appears to effect the display of my Instek GPS-3303 power supply. When it is drawing around 0.18A the 12V starts dropping to 11.8/11.9. If I touch my hands on the terminals of the supply while measuring with a multimeter it remains fixed at 12V... Anyhow all is good now. Thank you!
 

[IDEngineer]

My first reaction is a failed chip, based primarily on the "improved sensitivity" you're seeing from the 30A version. As you noted, that is the opposite of what you'd expect. Add in the latchup behavior and that chip is very suspect.

Be sure you have a bypass cap very near the ACS power pins.

Are you powering the electronics and the actuator from the same supply rails? I'd look to introduce some isolation there if possible. I'm familiar with the motor noise you're showing but your electronics won't like it, particularly if the voltage drops below the reset threshold on some chip. Might have even been the source of the latchup (I can't remember the fabrication technology used in the ACS off the top of my head).

Glad it's "solved", but I'd love to see what a new, trusted 5A part would do now.

[rstor22]

Glad it's "solved", but I'd love to see what a new, trusted 5A part would do now.

I ordered an ACS712 5A module/board from eBay which I would consider untrusted. I believe I replicated the similar setup that I had before. The previous behavior could not be replicated and it looked like everything behaved as expected.

[Seekonk]

I had one that was driving me crazy, bad solder connection on the board.