Pulse sensor using 3-phase sketch on TX for cumulative kWh


I’m hoping someone may be able to advise/critique on my process list.

I’ve got the pulse sensor feeding into a Wh Accumulator before x 0.001 and then into a kWh Accumulator as per below:

My questions are:

  1. is Wh Accumulator correct in this instance? I see that the notes state

Use with emontx, emonth or emonpi pulsecount or an emontx running firmware emonTxV3_4_continuous_kwhtotals sending cumulative watt hours

and I am using a TX with the latest firmware from github (openenergymonitor/emontx-3phase)

  1. if Wh Accumulator is correct, is it implicit that kWh Accumulator is correct?

  2. Wh Accumulator states the maximum power is 25kW, and with A/C and EV charging, its possible I could go over this. What do I do?

  3. Using the above process list, I am seeing an approx. 5% discrepancy between the pluse kWh reading and the sum of the three CT readings (using Power to kWh), over a couple of hour period. I have not calibrated the CTs as I’m still getting my head around everything so I’m wondering whether its simply a lack of calibration (and inherent inaccuracy of CTs) or whether there’s something fundamentally wrong in the way I am pulse counting.

Thank you for any assistance you may have :slight_smile:


I can’t comment on how to use the Wh & kWh accumulators. Many users seem to have trouble with them, whenever I’ve used one when I’ve been testing a modified sketch or emonLibCM for example, which ever one I’ve used (and would probably be power to Wh) has done as I expected, and its output has agreed closely enough (about 1% or better with voltage and current inputs calibrated using a multimeter) with the meter pulse count. (But I don’t think I’ve done that with the 3-phase sketch in total - not having access to a 3-phase supply, all I can do is check that it gives the expected (wrong) values on a single phase.)

The point I can answer: A 5% discrepancy would be attributable to a lack of calibration. I’d have expected better on the law of averages, but 5% is certainly well within the worst possible error you could have.

Hi Robert,

Thank you as always for your help and apologies for the long delay on my part.

I have no particular reason to think that the pulse to Wh/kWh isn’t working properly, I just wasn’t sure as I’m not specifically using the stated firmware.

Do you have any idea why there would be a maximum of power of 25KW noted for the Wh accumulator? Surely it’s just counting pulses.

Regarding number 5 (discrepancy between pulse kWh reading and the sum of the three CT readings), I was wrong when I said 5%. I’ve downloaded the figures into excel and done it properly – on 12/8/2020 there was a 2.29% difference which actually fluctuated between 2.29-1.8% until around 16/8/2020. From then it’s steadily gone up to around a 2.88% difference between the pulse and power conversion readings.

Do you think this would suggest that that there’s one phase more out than the others?


As I wrote above, I can’t help you with that because I don’t know the inner workings of it.

With the YHDC c.t’s, you must expect the error to vary according to the load and the type of load, i.e. the value of the current and the power factor, because the c.t’s phase error in particular (also the amplitude error to a lesser extent) is markedly dependent on the amplitude of the current. What that means in practice is you can adjust the calibration to be accurate at any one load, and it will be wrong to a varying degree at any other load current, or you can adjust it so that over time, the accumulated energy diverges away from and then converges towards your meter reading.

In particular, because the phase error affects the real power to an increasing extent as the power factor gets worse, if you have a small load with a poor power factor, it can have a much larger error proportionally compared to the error for a high current unity power factor load - depending of course on the current you used for calibration.

Possible causes for the increase could be either your average load has changed over time, or the calibration has drifted - either resistor values or the voltage reference - due to time or temperature, or a combination of both.

That’s anybody’s guess. Unless you can get the individual values for each phase over a period and compare them, nobody can know.

1 Like