ACS712-30A current sensor output problem

kumareshan · Joined: 02 Nov 2016 Posts: 1

Hi all

I am using FIVE ACS712-30A IC in a single PCB to measure current reading for AC loads.

1. VCC to the IC's are 5.0v.

2. Bypass capacitance = 1uF and 0.1uF

3. Filter capacitance = 0.47uF

4. Output of each IC's are 2.5v check in multimeter.

5. When i convert using ADC (MCP3204-12bit ADC IC), each value is oscillating and unstable at no load. What could possibly be the problem? How to make all outputs stable to zero without load?

THANKS

KUMARESHAN

Ttelmah · Joined: 11 Mar 2010 Posts: 19518

What are you using as the Vref for the ADC?.

A DVM, is not fast enough to see noise from a chip like this. Most take something like 0.2 seconds to actually take a reading.

The chip produces a ratio-metric output from it's supply. It's vital that the sensor supply and the ADC's Vref come from exactly the same point.

As has been said on many threads in the past. When dealing with things involving current (particularly if this is being 'switched', you need careful analog design round the PIC. I'd guess you actually have significant HF noise on the Vdd rail. The DVM is not quick enough to see it, but the ADC you are using is....
You probably need to isolate the Vdd fed to the current sensor from the main supply, and then feed this rail to the MCP3204's Vref, so both chips are measuring from the same rail. Even then the count will change by a small amount. The output will be 151.5 counts/A, with your high accuracy ADC.

temtronic · Posted: Thu Nov 03, 2016 4:42 am

I'll back up Mr. T and add that Vref should come from a precision V reference chip ! Anything over 8 bit ADC needs an ULTRA stable Vref !!
I also agree you need a lot of properly sized bypass' caps NEXT to every device as well as thick PCB traces. The unit should be in an aluminum or metal enclosure to reduce possible EMI affecting it. Also every I/O point from the PCB needs some form of 'bypass' be it inductors,caps,chokes,ferrite beads,small Rs, etc.
Sucessful analog design is half high tech reading and half black magic. I'm wondering if any level of school actually teaches anything analog aside from very basic opamp stuff ?

Jay

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

basic trouble shooting: do you have an oscilloscope - ideally a DSO - to analyze what is going on in analog terms?
or are you flying blind ?
Question

newguy · Joined: 24 Jun 2004 Posts: 1908

Ttelmah · Joined: 11 Mar 2010 Posts: 19518

drolleman · Joined: 03 Feb 2011 Posts: 116

When did you go to school? When I was in school, in the early 80's they taught all of that. I know even that in grade school, now they don't teach how to do hand writing.

temtronic · Posted: Thu Nov 03, 2016 10:35 am

hmm..re vref... I was referring to the ADC parts (MCP3204-12bit ADC) requiring a real stable Vref not the actual sensors.

Jay

newguy · Joined: 24 Jun 2004 Posts: 1908

1988 - 92.

Ttelmah · Joined: 11 Mar 2010 Posts: 19518

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

I've used the STmicro LF50CV with a tantalum cap on the output pin
for your sort of problem. It's a very clean LDO regulator
but you need at least 5.6V of inlet voltage to be sure its won't
dropout and defeat the purpose.

Ttelmah · Joined: 11 Mar 2010 Posts: 19518

Yes, I've used a similar IC.
It's worth adding, that the key here is getting the supply smooth. The units introduce a time delay on the signal output, relative to changes in their incoming supply voltage. If there is any significant noise on the supply the ADC's reference changes as well, but the 'now' output from the sensor, is based on the supply voltage a few moments ago. So you see the change in the value. In fact the filter capacitor on the chip will make this lag worse. So if the supply is not smooth you actually get better results by reducing this....
Keeping this supply 'separate' from anything likely to introduce noise, really is vital.