Archived Forum

I've just been experimenting again, and now I see a power reading of between 0-10w even when no cable is going through the CT!

Is this normal (or can be explained) or have I got a faulty CT ?

No, it is a good way to calibrate them. Which sketch are you using? If it is the discrete sampling one, remember that they're not measuring the current at exactly the same instant so as your system voltage bounces around (due to other consumers on the network switching things on and off) your current will change, and that will account for the inconsistent part of the difference.

It's nothing to do with distance along the cable, sorry, but when you do the numbers that idea is utterly ridiculous. One wavelength at 50 Hz is 3 x 10⁸ / 50 m = 6000 km. To get one minute of arc phase difference, the two c.t's would have to be 277.7 m apart.

No, you haven't got a faulty c.t. You are most likely seeing noise picked up at the input of the ADC. I've yet to do any comprehensive tests on the V3, it incorporates design features that should make it better in this respect than the V2 in which we saw generally up to 20 W of "power" due to noise. The mechanism and why it varies randomly from one input to another is explained here.

It is the discrete sampling sketch I am using. It's a bit annoying that my solar PV is showing generation overnight! Other people have complained about consumption overnight, which I could understand, but f they can detect that, why can't I even get mine to show zero? Would the noise average out?

my V3 is in the standard box sold in the shop, and everything apart from the Pi is in my fuse cupboard, powered by a transformer from the shop. 

Could it be the phase calibration that I need to check?

Mentioning phase calibration leads me to think you have a voltage input via the ac adapter. The noise input should be a lot less with the voltage input because the noise does not get rectified by the conversion to rms current. Phase calibration should be checked anyway, but it's not likely to be so far out that it turns the sign of the current around - unless of course the standby current of your inverter is almost entirely reactive, and that's a distinct possibility.

Are all four inputs the same? If it is noise, then I'd expect each input to be different because where the d.c. bias sits in relation to the ADC steps is essentially random (and no doubt varies a little with time, temperature or whatever).

Also bear in mind that the maximum power is 24 kW, and it's a 10-bit ADC, and you're seeing 0.04% of maximum reading.

I'm hoping to be able to get some measurements done on a production V3 before too long, I'll be able to be a lot more specific then.

I have an ac adapter both powering the emonTx V3 and also providing the reference voltage.

The inverter display shows very strange readings for generation, in that it's completely different to that shown by either the generation meter or the emonTx. It was suggested that the inverter is showing apparent power, so I changed the sketch to output that as well, but the inverter reading does not agree with that.

The inverter is an Afore HNS3600TL

I will try the other inputs this evening and see if I get the same 'noise'. I tested CT1 and CT2, both of them seem to give a value between 0w and 10w when there is no current.

Here's what I get from an untouched 'out of the box' V3. The first few outputs are wildly out as the software filter is still settling. None of the c.t's are near a mans cable.

RFM12B Initiated:
Node: 10 Freq: 868Mhz Network: 210
15300 14704 15326 2681 322.28
294 146 295 26 240.77
5 6 7 1 238.97
0 4 1 0 237.75
1 3 2 1 238.98
2 3 2 1 238.68
0 3 0 1 239.30
0 2 2 1 239.35
0 4 1 1 239.53
2 5 1 1 239.52
1 6 1 1 239.47
2 6 2 1 238.93
0 5 0 1 239.53
0 4 1 1 238.86
1 6 1 1 238.67
0 5 2 1 239.17
1 4 0 1 239.06

And the equivalent from an emonTx  V2, using the same ac adapter and the same c.t's:

P1= 239.38 P= 239.89 P= 251.49 V= 239.55
P1= 11.20 P= 8.39 P= 12.10 V= 238.47
P1= 6.08 P= 4.89 P= 3.92 V= 238.60
P1= 6.56 P= 9.66 P= 6.57 V= 238.76
P1= 9.29 P= 5.51 P= 7.45 V= 238.88
P1= 8.51 P= 6.03 P= 4.47 V= 238.80
P1= 9.13 P= 5.97 P= 6.11 V= 238.76
P1= 7.04 P= 7.42 P= 4.81 V= 238.87
P1= 8.53 P= 4.73 P= 5.80 V= 238.92
P1= 7.33 P= 5.31 P= 5.33 V= 238.73
P1= 9.62 P= 5.88 P= 5.28 V= 239.05
P1= 6.00 P= 5.90 P= 5.56 V= 238.74
P1= 5.18 P= 8.83 P= 5.83 V= 238.62
P1= 5.55 P= 3.41 P= 4.43 V= 238.67
P1= 9.68 P= 8.40 P= 5.89 V= 238.45
P1= 11.83 P= 8.23 P= 4.02 V= 238.61

As you can clearly see, the V3 shows significantly lower noise on 3 of the inputs, and using averages the worst from the V3 is still better than the best from the V2.

Which sketch is running on the V3 when it comes out of the box?

Quote from the Shop:

Includes

    emonTx V3 PCB fully assembled
    3 x AA Batttery Holder
    RFu328 + 433Mhz RFM12B - assembled
    Aluminium enclosure & custom fascia - assembled
    Firmware uploaded and tested

That link is to the DiscreteSampling sketch that you're running.

I know it's been a while, but at last I tested each of the 4 inputs on my V3. Each is slightly different, but CT4 is by far the best, giving a value of 1 or 0 all last night (CT1 gives values ranging from 6 to 11, centred on 8).

So is it okay to use by shop purchased CT in CT4? I notice it has a maximum of 4.5kW, but since my panels have a theoretical maximum of 3.5kW then I assume I'm okay ?

Yes, that's perfectly OK. A current transformer behaves as a current source, meaning that it will generate whatever voltage it needs to (within its VA rating of course) in order to drive the current into the load. In the case of Input 4, the load - the burden as it's called - has a higher resistance than the others, so it develops the same voltage across the resistor for a smaller current. It's that voltage that the analogue input senses.

Thanks for the explanation Robert. 

I did find one problem though. Having switched to using CT4 instead of CT1, I found I was getting no voltage readings from my sketch. Looking at the code in more detail I see that it takes the voltage as

emontx.Vrms = ct1.Vrms * 100

That of course was now zero. So I have changed my sketch to take it from ct2, which is my grid monitor. It would be nicer if the sketch checked to see which were being used.

I've already pointed that bug feature out to Glyn. All it should need is a string of "if .... else" statements to pick up the voltage from the first active c.t.