Difference in readings between emonpi & iotawatt

teledatalance · 2018-08-10T09:03:13.476Z

I have just purchased an emonpi and an iotawatt for testing - and have connected them to the same power source with the same supplied current clamps on the same cable - and am seeing the emonpi consistently giving approx 7% lower values. I dont see anywhere on the emonpi to set the type of current clamp connected - whereas it is set for each channel in the iotawatt interface.
Am I missing something ?

emonpi1354×725 58.8 KB

Robert.Wall · 2018-08-11T14:14:14.721Z

If you’re unhappy with the default calibration of the emonPi, which is set up for the EU a.c. adapter and YHDC 100 A current transformer, then you can adjust the "scales = ..." coefficients in emonhub.conf for the emonPi itself (node 5 - shows as [[5]]) to take out the 7% difference. In the absence of suitable calibrated instrumentation, your best option is to record energy use over a period of time and adjust the power calibration to match your supplier’s meter.

There’s very limited knowledge of Iotawatt here, The IoTaWatt forum is a better place to ask a question that is specifically about the operation and use of IoTaWatt. That is managed by Bob Lemaire, its creator, and he and the other users there are more likely to have the specific knowledge to help you.

danbates · 2018-08-11T18:36:18.429Z

EmonPi and IoTaWatt are developed by different individuals, and thus can’t be expected to work in exactly the same way. There are different ways of calibrating each.
There are plenty of reasons why calibration is a good idea if you really want accuracy, never mind the unit’s default settings or manufacturing process.
If you consider that both need calibration, and the Pi hosts emonCMS…

They are different beasts, as one might say.

overeasy · 2018-08-11T20:51:41.111Z

danbates:

If you consider that both need calibration, and the Pi hosts emonCMS…

Id like to correct the record here as well. While it’s true that there is a simple browser based voltage calibration procedure available in IoTaWatt. That’s the extent of calibration available, and to say it “needs” calibration is not true. IoTaWatt has calibration factors for most of the available VTs and applies it automatically when you simply select the particular VT from a list during configuration. Sure, you can fine tune that, but most people get great results without.

The original question was about using different CTs, and it is definitely not true that these two devices are equivalent when it comes to using different CTs. IoTaWatt requires NO calibration for any of a couple of dozen different commonly available CTs, and support for new CTs can be added in the included device tables.

IoTaWatt has an internal 0.2% voltage reference that it uses to self-calibrate. It is also not necessary to match a burden resistor to the CT and load to get good results.

It’s true that the Emonpi hosts Emoncms. IoTaWatt has it’s own datalog where all of the raw data is stored at 5 second resolution for one year, and 1 minute resolution for up to 15 years. It has an API that supports an adaptation of the Emoncms graph application. The datalog can produce Volts, Watts, Amps, Hz, VA, kWh and power-factor both current and historically with no need to create feeds or any other type of configuration. Without a doubt Emoncms has superior analytics and dashboard type apps, but as far as a straight datastore, I’d argue IoTaWatt is easily a match for Emoncms. IoTaWatt can also upload to influxDB for industrial strength data collection and analysis.

So I’d be happy to participate in any comprehensive comparison of the capabilities of the two products…

danbates · 2018-08-12T10:49:57.931Z

I found your post relating to accuracy. It might be important for there to be an independent test of this…

Your device is a good one, running fine software, reflecting of course your decades of experience, anyone would deserve recognition for that.

Personnally, I would buy the hardware made by these guys. This is quite personal to me, I wouldn’t care greatly if the emonPi was off by 10% at time of install, and as Robert says, it’s best to check relative to the meter. We individually exercise responsibility with commissioning and accuracy claims. What they have done for energy monitoring I want to be a part of, and simply entering into consciousness the value of monitoring our energy consumption is more valuable to me than a calibration constant.

But if we are talking about devices and accuracy, I imagine one day the OEM team and collaborators will produce a device better than IoTaWatt and emonPi put together, no rush. They might even use some of your work, who knows.

In my own experience. The clamp connectors can vary from batch to batch, and I wouldn’t rely on a single device or batch of devices, as it seems you might have done here, to claim 1% accuracy, even if it would be the case most of the time.

danbates · 2018-08-12T11:14:14.106Z

I think the original question was “Am I missing something?”
The answer is no, everything’s working exactly as it should. They’re different devices with different development history. The emonPi doesn’t have a CT selection menu at present. The resources to change the settings for different CTs are here in the forum and elsewhere in the site, I think the ‘Learn’ and ‘Resources’ sections.

TrystanLea · 2018-08-12T21:02:06.995Z

This thread was going off topic in an unhelpful way in earlier posts, in the interest of keeping the discussion constructive I have edited one post and removed another, IotaWatt and EmonPi are complementary and @glyn.hudson and I (designers of the EmonPi) support them both.

@teledatalance we apply a default calibration to the EmonPi which could be in error by 7% if component tolerances align to this extent, it is typically less than this. It would be interesting to hear how the cumulative kWh recorded compares to your billing meter.

TrystanLea · 2018-08-12T21:12:22.625Z

@teledatalance this page written by @Robert.Wall gives a good overview of the worst case errors possible if all component tolerances align to increase the error (using the standard CT’s and ACAC adapter that we use for the EmonTx/EmonPi) without calibration: Learn | OpenEnergyMonitor

Robert.Wall · 2018-08-12T23:37:10.847Z

It’s also worth noting that for the emonTx V3.4.2 & emonPi, a Monte Carlo analysis on the components that affect calibration shows that 82% of production samples are likely to fall within 1% of nominal, and 45% of production samples are likely to be within ½%. (The figures for the emonTx V3.4.4 are better, at 98% and 72%.)

It should also be noted that, as I mentioned earlier, the default calibration for the emonPi is for the EU a.c. adapter, and not the UK a.c. adapter. That will introduce a 3.33% error if the emonPi is used with the UK adapter. (The EU adapter is 230 V : 11.5 V whereas the UK adapter is 240 V : 11.6 V)