I’m still looking around for a replacement of an old EmonTX 3.4, and one of the contenders is an Elkor WattsOn device. They’re very accurate, and can be obtained at a fair price used, like this.
Reading the docs, I was wondering if I understand the resolution right. It seems that whatever clamp I add it to from 60A to 450A, the internal scale is spanning the 0…5000 range. That’s pretty weird, as going upwards with measured currents, the resolution is going downwards. For instance I can get a 300A Elkor CT, but we rarely go above 25A on a phase. The max. internal value should be 25/300*5000=417. One increment is 60 mA, or about 14.4W at 240 VAC which sounds rather coarse. I wonder if it gets better if I set the internal scaler so that it spans as much of the 5000 values as possible, but I guess not. It just changes the scale so the unit can report the proper measurements, If I understand it correctly.
Is my understanding right? Do some models offer a better internal resolution? And what are the recommended CTs to buy? A proper mA range or 333 mV? Let’s say, max. current would never exceed 50A (in reality, much less, but future expansion is not impossible).
I remember a few folks here know these devices well. Hints are appreciated.
The resolution doesn’t directly say anything about accuracy (unless it’s very coarse). The 5000 figure is in the manual, on page 15. I found various docs on the Elkor official site that seems to recite this figure. The device seems to be happy when the CT’s range is scaled into that 5000 bracket.
I remember talking to @Bill.Thomson who has a few of these devices, maybe he could also chime in sometime…
I think it’s saying it assumes a 5A max deflection on a CT input, and then gives you 1mA resolution, so yes, if you put a 50A CT on it you’ll need to scale everything by 10 giving you a 10mA resolution on the current readings. 500A CT, 100mA resolution etc etc.
The Power and Energy calculations/readings are unrelated to the Irms readings so while you’ll lose resolution on the Irms reading, you won’t lose a corresponding resolution on the Power readings (i.e. it will be much better than 14.4W in your example)
I don’t really understand this, as you also noted that for e. g. an 500A:5A CT the resolution will be 100 mA so there’s no way to discern smaller values or steps (be it RMS or instantaneous current sample), the voltage is fixed, and the instantaneous power reporting resolution is the product of the 2. I can be wrong, if so, can you explain where? Energy is a sum over time, so it can be more precise.
The RMS readings (V and I) use a completely different path through the IC than the Power (and hence Energy) readings. The only thing they have in common is the initial raw reading from the ADC and that’s typically 24-bit so has more than enough dynamic range to cope.
The meter designer gets to choose what resolution they offer you on each reading. For example, on my meter (which I think uses a very similar IC) I offer current readings to a resolution of 10^-5 Amps (i.e. 10 uA) and Power readings to 10^-1 Watts (i.e. 100mW). But the meter assumes I’ve installed a 20A->333mV CT. If I’ve installed a 200A->333mV CT then I have to manually multiply everything by 10. So my Irms resolution becomes 100uA and my Power resolution becomes 1W.
Usually it’s Power (and Energy) that most people care about. What resolution does your meter offer on those readings?
Looks nice, what are the accuracy specs? I decided to get a better accuracy device than my utility meter (which, I think, is 1%) if feasible, i. e. not overly expensive. Will this work in Europe on the 230V/240V grid?
EDIT: NM, found them, they even spec the accuracy for various load conditions as well. Sadly the 400V or 480V versions for EU grids are much more harder to get at good prices.
Due to the flexibility of the device to incorporate a wide vaiety of CTs, all current inputs are scaled to “5A” full scale.
However, if a mA or mV CT are used, the full scale output value (either 333/1000mV in the case of mV output CTs, or full scale current as defined by the configuration in the case of mA CTs)
is scaled to represent the number 5000.
This methodology allows for the use of the same scaling formula to perform calculations regardless of the actual type of CT used (5A/mV/mA).
Yeah, most of my confusion stems from the standard 5A output CTs, scaling specs, part numbers and the lack of easily available documentation on the CTs and part numbers. For instance, the MCTA 60A CT doesn’t exist. Weird. There’s the MCTA 300A (300A in, 120mA out) and that’s it. It’s more easily available but I was worried about the attainable resolution. Which, according to @dBC, is a non-issue. There’s also a way to use 333mV CTs with the appropriate models (using unambiguous model numbers), maybe I’ll look for those.
I’m not sure I said that. What I said was that the resolution of the Irms registers after you’ve scaled them to match your CT sensitivity is not relevant to the resolution of the Power (and Energy) register readings.
Presumably you will also need to scale the Power (and Energy) register readings to match your CT sensitivity. So I asked what the resolution of those registers are. But I found the answer myself:
That also confirms you need to apply the same scaling to these readings. So you will get 1W resolution if you use 5A CTs, and 10W resolution if you use 50A CTs etc.
Thanks. Although one doesn’t ship to Europe (nor to the US), the other is a bit expensive on the shipment, but otherwise good bet.
I’d use 2 phases, one utility and one solar. There are no installations any more like this - all sites utilizing solar must use a 3-phase system now. But I might be able to get better res on individual phases. They shoulda chosen bigger register sizes. Where is the excerpt from?
Yes, looks like per-phase power resolution is 1/10th of a Watt. Do you only have 2 of the 3 phases running to your property? If so, you might need to dig deeper to determine if the underlying IC used in that meter can cope with no connection on 1 phase. I vaguely recall some of the poly-phase ICs want all 3 phases wired - but don’t quote me on that, it’s a long time since I’ve looked at the poly-phase stuff. In the end, we went with 3x single phase ICs in the one box to gives us 3 phase capability.
I think you’re right. I haven’t had time to read the docs on your meter in depth, but if our assumptions are correct I think they’ve taken a very capable core metering system (with huge dynamic range and resolution) and then lost if all at the last minute with their UI soft register choices.
At the other extreme, I wanted to report energy in WattSeconds and wanted the result preserved for the life of the meter, so ended up going with 128-bit registers for that.
I think that was just the next page after your reference, so p16 I guess?
Our house is single phase, so to be more accurate, it’s not 2 phases, but one branch is current flowing from/to the utility and another branch from the solar inverter. The meter can deal with this. The voltage inputs simply need to be shorted.
They don’t need to be shorted, although one can choose to do so.
I used my WattsOn - without issue - for about a year, connected as shown in the diagram above. After installing a PV system, the third V/I inputs were used to measure PV energy harvest as per the following diagram:
I might need to re-read the docs, but IIRC it mentions that phases are sampled independently and depending on the used phases (and the electric system, like the split-phase system in NA), a set of calculations are made towards the totals. It would then follow that one would need to hook up the same AC voltage input to all measured phases of the meter in my case. But I’m not 100% sure on that.
