emonTX v4 : incorrect CT reading

A couple of weeks back, I received a couple of emonTX v4s to add to my emonBase platform which I was setting up. One of them is in the garage connected to the main incoming feed, the other is in the room with my heat pump kit.

The CT readings from the emonTX connected to the main incoming feed seem more or less spot on, within about 20W of the consumption on the smart meter, and over the course of the day, the CT measurements are within 500W of that from the smart meter.

However, the CT readings from the emonTX connected to the heat pump seem wildly off. I have a 50A CT clamp on a port configured for 50A, which when the heat pump is idle, seems to be reading 270W when Iā€™d be expecting considerably less. On the heat pump distribution board, there is an SDM120 (not being monitored by anything) and over the course of a day, the readings on the SDM120 are about 3kWh lower than those measured by the CT.

I have tried a different CT, and plugged into a different port, and the readings are still wildly off. Iā€™ve tried the CT on different wires (the main feed from the CU to the DB, the live wire from the breaker to the SDM120, and the live wire from the SDM120 to the heat pump) and all have this 270W reading. Iā€™ve noted that when the heat pump is freshly switched on at the consumer unit and it reads 270W, the additional total household power consumption at the main feed only increases by 80W. I havenā€™t yet tried connecting a CT to the live wire to the main consumer unit for the heat pump circuit as I donā€™t want to be messing around in there.

The heat pump distribution board has only two outgoing connections - one to the heat pump and one to the immersion heater (which also powers a plug socket for the Homely system). The CT clamp on this does seem to read correctly (currently reading about 20W).

What might be causing this?

ā€“ Edit ā€“

Probably worth noting that I have a voltage sensor on the TXv4 at the consumer unit but not on the one monitoring the heat pump, but given that this ā€˜overreadingā€™ is only on the distribution board circuit to the heat pump and not to the immersion heater, and the size of the overreading, I donā€™t think this is the issue.

ā€“ Edit 2 ā€“

I only get a reading from the CT when around the live cable, but no reading around the neutral cable?

I noticed my AHSP seemed to be using a similar amount of power when idle but in fact it isnā€™t - you need to take the power factor into account.

Iā€™m just now reading the values directly over modbus from the SDM120 - the _P value on the input is what you need to log to a feed, not the _VA.

Which of course @IanHarris isnā€™t doing because heā€™s not using an a.c. adapter, so the emonTx is reading Apparent Power. not real/active power. And Iā€™d suggest that will largely explain the discrepancy, due to the reactive load of the compressor motor in the heat pump.

Hello @IanHarris and welcome to the forum!

Yes this is very likely to be the reason. Testing on my heat pump here I can replicate exactly what you see @IanHarris. A voltage sensor is needed for accurate real power readings (also often called active power).

Hereā€™s an example period with the emonVs voltage sensor connected:

  • heatpump_elec is an SDM120, reading mean value of 818.8W
  • emonTx4hp P2 - an emonTx4 with a 20A CT sensor (816.4W, -0.29% difference)
  • emonTx4hp P3 - a second 20A CT sensor next to the firstā€¦ (820.9W +0.26% difference)

Im very happy with that, 0.3% is very close!

Hereā€™s a test removing the emonVs voltage sensor on my 5kW Mitsubushi Ecodan:

  • When the heat pump is running at 862W real power, removing the voltage sensor and therefore changing our measurement to apparent power results in the power value increasing to 944W. The power factor at that point is therefore 92%
  • When the heat pump compressor is off and we are now just powering the controller board the apparent power only drops to 242W, the real power at this point is 17.5W. The power factor is 7.2%. This is a result of the power supply type used for the ecodan controller board, it has a very low power factor.

As you can see, the voltage sensor is critical for accurate real/active power measurement on the Ecodan heat pump and probably on most others as well.

2 Likes

:+1:
A most ingenious and clear illustration.

3 Likes

A most accurate solution, thank you @TrystanLea - the emonVS was moved from the consumer unit to the Heat Pump Distribution Unit and hey presto, accurate looking readingsā€¦ An emonVS-mini has been ordered, but might there be a way for an emonVS connected to one emonTX to share itā€™s output with the other emonTXs in the property (or is that simply not how it could possibly work?)

Ian

Thanks @IanHarris glad that fixed it!

Yes, a single emonVs could be used as a voltage sensor for multiple emonTx4ā€™s , a 6p6c RJ11/12 splitter would be required, it can only be done via a hard wired connection and a voltage calibration adjustment is required. Id like to get such a splitter in the shop as an option, I think it could work well.

1 Like

Nice!

The US standard for Revenue Grade metering is 0.5% or better. thumbsup highfive

Lots in here for the docs :rofl: #pleaseaddtothedocs

Same here; MID class-C meters are 0.5% (class-A and B are 1% and 2% respectively).

This UK government report, written in 2015, states the following:

Electricity meters are considered ā€œaccurateā€ if the permitted margins of error do not exceed
+2.5% to -3.5% throughout the entire load range at which the meter is designed to operate.