Sorting out an Australian three-phase installation

bluetardis · August 19, 2018, 9:40pm

Agreed - the Sauna shows that we get some balance across the 3 phases assigned and they are close. I suspect the AC has multiple single phase units combined - eg compressor/fans/etc Ultimately its not as issue as long as we can get a handle on the total draw.

I see that your system has updated to 02_03_13, and there may be a recurrence of problems with zero voltage

I am guessing this was because I am on ALPHA and did a power cycle to remove the laptop.

We have now changed the 3 x CTs on the mains with new CTs from a new batch.
We are still reading low and I would like to get that addressed somehow.

derived reference may not be appropriate for your setup

Not quite sure what you mean here and very happy to take direction.
I assume this means we need to get 3 x VT across the phases. We will have a second iotawatt soon but not sure how/where to confiture the 3rd VT.

I will do the CT test tomorrow to see if we have any outliers - will also list circuits and suspected phases for going forward.

bluetardis · August 20, 2018, 10:55pm

I think my next steps are:

Do the CT test/audit on Mains1 (PhaseA) circuit with as many CTs as I can practically fit there and then post results. - audit all of them so I know if we get any outliers. (I can also get a TrueRMS clamp if needed to confirm whats really happening.)
Post a table of circuits (planning purposes - eg connecting CTs and VTs in preferred fashion)
Why (I want to get input as to how you want me to configure them as I we will be using at least 2 and possibly 3 iotawatt units).

Background Stuff
a) There is another IotaWatt coming with 2 x new VT. Will test to see if we get anything different on the Voltage reads (it also means I have 2 of the same units and can compare against my current though I understand it shouldn’t make any difference once calibrated.)

b) Also awaiting instruction/thoughts from @overeasy re derived reference / other approach to monitoring.

Once we have a plan for configuration that seems to work, I will organise an electrician to come in and clean up the installation and run me another external power point setup (so we can nicely mount the VTs/iotawatts/cables/etc)
I was hoping to defer this until I had some readings as I think I want to significantly upgrade my solar but wanted to get some stats first. Should it be needed we can also trace phases.

Table for Planning:

Circuit	Phase	amps	InputName	Notes
Mains1	TBC	nn	Mains1	Mains input/service fused
Mains2	TBC	nn	Mains2	Mains input/service fused
Mains3	TBC	nn	Mains3	Mains input/service fused
Solar	TBC	20	Solar	Volts read high at Solar Inverter (240 vs 230 at box)
HotWater	TBC	20	HotWater	Single Phase
Aircon1	TBC	32	Aircon1	3 Phase input / single breaker
Aircon2	TBC	32	Aircon2	the aircon doesnt seem evenly balanced across
Aircon3	TBC	32	Aircon3	all the phases
Garage1	TBC	32	Garage1	3 Phase input / single breaker
Garage2	TBC	32	Garage2	Workshop and Sauna
Garage3	TBC	32	Garage3	Sauna
Sunroom	TBC	20	Sunroom	Sunroom/part of recent renovation
Shed1	TBC	32	Shed1	3 Phase input / single breaker
Shed2	TBC	32	Shed2	Fridge room
Shed3	TBC	32	Shed3	Pool pump/setup (need to check out)
Lights1	TBC	16	Lights1	House lights - these circuits are close together and
Lights2	TBC	16	Lights2	if they are on the same phase we can use a single CT
House1	TBC	20	House1	will map/figure out and give a better name
House2	TBC	20	House2	will map/figure out and give a better name
House3	TBC	20	House3	will map/figure out and give a better name
House4	TBC	20	House4	will map/figure out and give a better name
Oven	TBC	20	Oven	Basic 20A Single Oven on Circuit

bluetardis · August 21, 2018, 1:55am

Testing CTs

IotaWatt Configuration:
Cleared the inputs/outputs and suspended logging to emoncms.
Numbered all the CTs from 01 (I) upwards to 23 (XXIII)
Named the inputs to match the CT.
Am using house power and boiling a kettle to generate ~2000W as the high load.

Testing CTs - Batch01
This is the second batch I received and are CTs that we have been using most recently to monitor the Mains1/2/3 and Solar inputs for discussion/testing.

SCT-013-000 Non-invasive AC Current Sensor Split Core Transformer
10 x 100A:50mA

To my eyes, they are close… CT04 and CT06 are a little low but probably within tolerances?

CSV Data over a slightly longer period:
Test01-Log.txt (28.1 KB)

Testing CTs - Batch02
These were the original CTs used. (13 in total)
SCT-013 15/20/30/50/100A Non-invasive AC Current Sensor Split Core Transformer

01 x 100A:50mA
03 x 30A/1V
06 x 20A/V
03 x 15A/`V

I have left CT01 from origin test on input01 and will perform the same test. The inputs are marked so you can see the type of CT used.

Big difference and I need advice. I am guessing that I should probably be only using the 100A:50mA for all circuits regardless of amp rating.

CSV Log for the graph (it didn’t fit the other way)
Test02-Log.txt (24.8 KB)

bluetardis · August 21, 2018, 3:20am

I am placing the 4 x 100A back on my MainsA/B/C and Solar so I can get some load indication vs voltage and see how things are going.

I wil stop here for now and await advice.

I have an electrician coming out that will label the Circuits/Phases for when we are ready to reconnect.

overeasy · August 21, 2018, 11:50am

That’s the problem. Those are “voltage” type CTs, they are not SCT013-000. They have burden resistors, which when combined with the burden in the IoTaWatt, present a lower overall burden and thus a lower voltage and so indicate a lower power. Moreover, those particular CTs have fewer turns than the SCT013-000 at 1860 vs 2000. Your problem has been the IoTaWatt reading lower than your meter. This explains it completely.

You should not be using those CTs. Install the new ones on your mains, solar, and hot water and see if you get better agreement with your meter.

bluetardis · August 21, 2018, 10:58pm

There were SCT013-000 originally on the mains but they were from the “old batch”.
The new ones are from a different supplier and have a slightly shorter, thicker cable and all read the same.

I had been told to purchase the other units as it was better to match the circuit with the AMPs from the CT as it should have been more accurate".

Having run my experiment and also found this thread, that logic proves to be incorrect when used with IotaWatt hardware.

I did read the doco prior to purchase but this caught me out and it might be worth an additional note on both sites that not all SCT013 are the same or usable.

I am happy to pm you or post the source I found for the CTs as it might be useful for other aussies.
Once we have our Inputs//Meters lined up I can order the others I need and place with more confidence.

Going Forward
I have installed the “new CTs” on Mains A/B/C and Solar as this is what I can reference meters against. It will take another 24 hours to be able to line up with the meters as I have been logging my meters at 10:00am in the spreadsheet. I will report back.

I have validated that with load on known circuits for A/B/C that Iotawatt reads 1.00pf (at ~2000W) eg moving the kettle.
I can see about 680W on Solar with pf1.0 which is closer to what the inverter tells me its making. (Still about 75-80% though).

The pf for PhaseA does fluctuate a little between 200-400 watts (washing machine on pre wash).
This may be normal - its when the load decreases over time vs load stopping (eg motor reducing RPM over 30 seconds). I don’t know if its an issue but mentioned it for completeness.
I cant comment on PhaseB/C as anything that fires up starts around 500W and it locks in pretty quickly.

Basic setup as follows:
Inputs

EmonCMS Web Server Inputs (logging to a australian azure host for now).

Power Factor on Mains A examples

Zoomed in Graph from Last night - pause in logging?
We didn’t get any voltage reads as 0 but there a few times it looks like we flatline.

overeasy · August 21, 2018, 11:29pm

V[quote=“bluetardis, post:70, topic:248”]
There were SCT013-000 originally on the mains but they were from the “old batch”.
[/quote]

Not to put too fine a point on it, I believe the correct models for those voltage type CTs is SCT013-050, SCT013-030 etc. Educating the world about the confusion isn’t anything I’m in a hurry to do, although I’ll be more aware of the possibility next time I get a low reading inquiry.

While the SCT013-000 works fine, I don’t recommend them. I’ve tested all of the CTs you see listed in the config Dropdown with a special rig that measures linearity and produces a phase shift profile. The Echuns, in my opinion, are superior in just about every way, and they have UL and CE certifications. I’m not aware of any certifications for YHDC CTs.

I’ve seen quite a bit of advice handed out that encourages careful CT and burden matching to get optimum results. While the case can be made academically, I’ve yet to see where anyone has actually shown any imperical evidence that it matters. The proof of the pudding… Maybe I didn’t look hard enough. I’ve run 800, 1000, 2000, 4000 turns on the same wire and had equal results.

In the test that you did on one wire, there were some outliers At 2% of the average, but in a true test you would need to space them an inch or so to eliminate interference.

If your solar is still low with the meter after a couple of days, try changing the CT.

Don’t get hung up on power factor. Unity (1.0) is useful for sorting out the phases, but the problems you Were having with voltage type CTs should not effect PF.

bluetardis · August 21, 2018, 11:33pm

Thanks for confirming this.

This is a shot with kettle/lights and the HWS kicking in. Its still winter here so not too much sun yet.

I will need some more CTs so lets get the right option.

While I do have some space in the back of the box its not massive. I want to get a lot of CTs in there to determine “what” is draining power. AC is a culprit but there are some unexpected heavier usage that needs to be traced/ongoing monitored.

Biggest cables are 6mm - I think I saw a post where you had recommended some smaller Echun? Can you please confirm model and I will see if I can source a batch of them?

overeasy · August 21, 2018, 11:45pm

ECS10-50 50ma output. I think you need to also ask for TVS diodes, and 3.5mm jack.

bluetardis · August 22, 2018, 12:50am

They are a bit hard to source in Australia but this seems to be the unit.

Can you confirm you are happy with these?

I am inclined to place a order for 33 (30 and a few spares) which is overkill but I know some other local installations that might benefit.

overeasy · August 22, 2018, 12:59am

ECS1050 L59E is the correct model. Looks like what I have. Am I happy with evaluating from halfway around the world? Not really. It’s up to you.

bluetardis · August 22, 2018, 1:00am

I get it - just can’t find any local supplier for Echun whereas I can get the SCT13 locally.
Price is similar so happy to get either - just wanted to make sure I had right unit.

I can purchase from you but I think for 30 CTs thats likely a major hassle with double shipping etc. Let me know.

overeasy · August 22, 2018, 1:11am

Not shipping to Australia, sorry.

bluetardis · August 22, 2018, 1:15am

All good - I have asked for a sample. Will ask around as there are a few local installs here that might also want them. Will get a batch in.

Can you have a think about how you want me to setup the 2 x IotaWatt and the circuit list?

bluetardis · August 22, 2018, 4:55am

Update:

We have new batch of CTs monitoring MainsA/B/C and Solar and will report against meter over the next few days.
New IotaWatt and 2 x VTs have arrived.
Echun CTs are on their way from China.

Please advise if you want me to do anything other than gather stats for now.

bluetardis · August 23, 2018, 12:10am

I will post an update tomorrow.
My computer has crapped out and needs a service. Still logging though.

Good news is that we are closer.
Against the meter, Solar is about 90% mains is about 110% of metered reading.
The solar might be ok, just that we have 2 decimal places on the IoTaWatt and 0 on the meter.

Have moved VT input calibration from 23.03 to 23.09 and tracked it against multimeter over 10 minutes. It was accurate when we initially installed but with updates, etc I figured it was worth repeating the process.

overeasy · August 23, 2018, 12:42am

Not too clear on that 10:00 check in time, but over a few days, should not matter. Post the data after you get back on the air.

Yea, that’s 0.2%, which I’m sure is better than the multimeter spec and definitely more accurate than the derived voltages being used for two your three phases. So don’t obsess about that.

bluetardis · August 24, 2018, 12:48am

Not too clear on that 10:00 check in time, but over a few days, should not matter

EmonCMS calculates dailies and kwhd at GMT which for us is +10 or 10am so by measuring the meter then then I can get dailies that should line up. Once we get agreement between meter and iota watt then this can go away.

Looking at the graph/delta, it appears iota watt does the same and as all my readings were taken at 10:00. This gives me a consistent time (that appears to line up with the software/reporting).

Post the data after you get back on the air.

Here is the data for the last 3 days - this now give us 2 prior days of clean measurement.

I can go further back if needed but didnt think it was useful as we did a stack of testing and changed CTs so its not going to be particularly accurate.

Date	Mains	Meter	IotaWatt	Solar	Meter	IotaWatt
21/08/2018	28,454	43	49.79	10,052	4	2.21
22/08/2018	28,497	40	44.77	10,056	3	2.87
23/08/2018	28,537	32	68.97	10,059	2	1.99

Notes:
a) The meter readings are a bit lower and closer to what I would expect. That said the AC off at the meter for 20 hours a day and Sauna not used so this is basic house+oven+hws etc.

b) CTs are all from the second batch of SCT013-000 100A units that we tested and positioned as discussed.

c) Reading the Solar Watts from the IotaWatt Web Status vs the Digital meter is very close and this now lines up in the meter readings.

d) When measured with load, each phase appears to be correct - our kettle load test when checked on Phase A/B/C shows very similar watts and a pf of 1.0

e) The total comes from the Outputs created “Mains” of Mains1+Mains2+Mains 3 and then this graph.
The graph uses energy, with a delta and daily to get the totals.
Please advise if this isn’t the best way

2018-08-16 10:00:00, 219.0, 25.10, 3.31
2018-08-17 10:00:00, 227.9, 37.66, 3.15
2018-08-18 10:00:00, 204.4, 62.98, 3.19
2018-08-19 10:00:00, 230.5, 79.43, 3.35
2018-08-20 10:00:00, 232.0, 49.97, 3.23
2018-08-21 10:00:00, 233.1, 49.79, 2.21
2018-08-22 10:00:00, 233.1, 44.77, 2.87
2018-08-23 10:00:00, 232.2, 68.97, 1.99
2018-08-24 10:00:00, 234.4, null, null

As you can see its reading way high compared to meter.

Here is the graph with the table showing the phases if it helps for the same time period. The basic calculation adds up.

2018-08-16 10:00:00, 25.10, 7.34, 7.01, 10.76
2018-08-17 10:00:00, 37.66, 9.99, 8.08, 19.59
2018-08-18 10:00:00, 62.98, 17.81, 17.47, 27.70
2018-08-19 10:00:00, 79.43, 23.99, 22.23, 33.21
2018-08-20 10:00:00, 49.97, 18.02, 9.73, 22.22
2018-08-21 10:00:00, 49.79, 9.98, 21.88, 17.93
2018-08-22 10:00:00, 44.77, 12.19, 11.41, 21.17
2018-08-23 10:00:00, 68.97, 8.67, 16.49, 43.81
2018-08-24 10:00:00, null, null, null, null

Please advise next step.

overeasy · August 24, 2018, 1:57am

I’m not too clear on how all that works. When talking accuracy, I prefer to deal with the data directly from IoTaWatt, from which you can get the current kWh reading at any time.

You have an output defined called Mains, which is the sum of the three main inputs. If you run the graph app, and select Mains under the “Energy” tab, you will get the running kWh total for the last 24 hours (or any period you select at the top). Because I am split-phase 120/240 in North America, I have two mains. So I have an output “meter” defined, which is the sum of the two mains:

So right now, my virtual meter reads 10820.5 (You can see by Diff that we used 15 kWh in the last 24 hours). My point is that you can read your IoTaWatt meter anytime you want to. The way I track accuracy, is that I read my physical meter and the IoTaWatt kWh total at the same time. Then a day, or two days, or anytime I think of it, I read the two again. The difference between each should be the same. There is no need to read your meter at a particular time to do this accuracy check. It’s more convenient and there is no doubt about the figures. You can do the same with your solar feed, recording that meter and the Max kWh for your solar under the Energy tab.

I prefer to work with these numbers, with date and time of each reading. You can use the Emoncms agregate function in your reporting to track daily use if you fing that convenient. You can also get daily use from IoTaWatt:

From the previous graph, I selected Week, Type “Daily”, In the Feed line I selected Type “Bars” and “Delta”. Just like Emoncms, you can save this graph to recall anytime. This graph is UTC days, I think there’s a way to make it do your local time zone. The data is certainly available.

But lets start using a log with the IoTaWatt numbers take simultaneously for comparisons.

bluetardis · August 24, 2018, 3:18am

The way I track accuracy, is that I read my physical meter and the IoTaWatt kWh total at the same time. Then a day, or two days, or anytime I think of it, I read the two again. The difference between each should be the same. There is no need to read your meter at a particular time to do this accuracy check. It’s more convenient and there is no doubt about the figures. You can do the same with your solar feed, recording that meter and the Max kWh for your solar under the Energy tab.

I get this - I have been reading daily at 10 as I am home and its daylight. just a habit to read at the same time so I could compare usage over a time period until we get things sorted…

My method to date was to take readings simultaneously (within a minute of both the meters and iota watt data). As my meters are manual, I had take a photo then gone to the computer, refreshed the graph from iota watt and then entered the data into the spreadsheet before posting here.

I think we are in violent agreement with the Graphs and have posted what you asked for below. The “Delta” method is what as I using as this gave me history. Please tell me if I misinterpreted

I will post more results for tomorrow - I did a quick compare now against my last read (10:00) and both close - e.g. within 10% for both but with only 4kwh on Mains and 1kwh on solar, e.g. IotaWatt vs Meter.

49%20pm

These is consistent with numbers shown for the last few days. Overall power usage a bit simpler/cleaner - vs what happens at night (e.g. last 24 hours here)