Question about background noise of ~80W

Submitted by Guest on Thu, 23/05/2013 - 17:08

Hello everyone,

I have been browsing the forums for a while now, trying to find an answer to an issue I am having with a arduino shield, one emontx and one rpi module.

I have set up all three boards, and the emontx+rpi module are working great. values are correct, and it is showing 0W with no load.

However, today I soldered the emontx shield for an arduino uno, and all the CT-s are showing 80-90W of base power. Connecting some load works, but i am uncertain of why the shield shows 80-90W on an empty curcuit, but the emontx module shows the same as 0W.

I have found references to this "background noice" on the forums, but not how to correct for it. I could have accepted this as "normal", but since the emontx is not having the problems, I am sceptical.

I have now hardcoded a -88W adjustment in the code until I figure out what this is.

if (CT1) {
    emontx.power1 = ct1.calcIrms(1480) * 240.0 - 88;
    Serial.print("ct1: ");
    Serial.print(ct1.calcIrms(1480));
    Serial.print("A/");
    Serial.print(emontx.power1);
    Serial.println("W");
}

See screenshots:

http://i.imgur.com/czPFeMf.png

Output from my modified sketch.

http://i.imgur.com/B93Q54P.png

ct1-2-3 and bv are the emontx module, ct4-7 are CT1-4 on the arduino shield. as you can see the ct2 is at 0W, but ct4-7 are all ~80W

Thank you in advance,

Jon

Re: Question about background noise of ~80W

Submitted by Robert Wall on Thu, 23/05/2013 - 19:55.

80 W does sound rather high, but you don't say what your full-scale power is. The software is rectifying everything the input sees and presenting it as a current. If the input is biased so that the quiescent level is exactly in the middle of an ADC step, nothing will be registered until there is enough input to cross into an adjacent step. If however the input is biased so that the quiescent level is exactly on the edge of an ADC step, then the slightest whiff of noise will cause it to cross and register a current. Where the quiescent level sits depends on the tolerances of the bias chain resistors.

It has been suggested that a small capacitor from the processor input pin to ground will reduce the effect of noise.

I think you have been exceptionally lucky with the emonTx. It is normal to see up to about 20 W (less if the voltage input is used and real power is calculated) with a full-scale of 24 kW, so it's actually about 0.1% of the maximum reading, which considering the ADC is only 10 bits, is really quite good performance.

There a page in Building Blocks about this.

Re: Question about background noise of ~80W

Submitted by psax (not verified) on Fri, 24/05/2013 - 07:38.

Hi Robert,

Thank you for the reply. I tried moving the the shield+dino around in my house for a bit, and found that it fluctuates between 40-90W. I will try to insulate the shield a bit to see if i can reduce the readings.

I will try to make a breadboard where I 1) try the capacitor tip, and 2) see if the arduino actually is causing some of this noise. Since I will not be measuring more than 32A, I could try another SCT if I can get hold of one.

Thanks for the tip about the article in building blocks, I cannot believe I missed that one.

/Jon

Archived Forum