Archived Forum

I detect a small degree of confusion here (unsurprisingly!) so I'll try to clear it up. I'm no Arduino expert, but I believe you can only have one shield per Arduino. Each has 4 current and one voltage input.

The problem is not that the accuracy of the CT's falls off at low current (which it does) but that the wanted voltage signal generated by the burden resistor approaches the noise injected by the digital parts of the processor. So it's really a matter of signal/noise ratio.

I believe that the -000 (100 A) and the -030 (30 A) ct's are the same beast, the difference being that the 30 A one has an internal burden resistor. If you are allowed to disconnect the wires to thread through a ring core ct, then a small one that is adequately sized to give you the required 1.75 V or so output for the emonTx shield at the maximum current you want to measure would probably be better. The snag is, most are rated for 0.33 V out, so if you get one of those, you need a "current output" type (i.e. no internal burden resistor) and you need one rated at 5.3 times the maximum current you want to measure (that's to get the correct VA rating!). Then you will almost certainly need to change the burden resistors on your shields. You can happily feed all the shields with the same voltage reference signal as long as they all share a common ground and 5 V supply.

If you want to measure the incomers, you'll probably need a couple of Magnelab split core SCT-2000-000.  My crib notes say the -1250-000 is good for 200 A @ 1 V output + 120% overload, with a burden resistor of 37.5 Ohms, so if you increase the burden resistor to give 1.75 V, they're only good to 137 A, after which they aren't guaranteed to be linear any more. [They're OK to 200 A for the emonTx, which has a lower input voltage than the Shield.] We'd need to ask Magnelab for more details to calculate the correct burden for the -2000- for you.

If you use the standard GitHub sketch, you don't need to change anything for the different frequency. That change is only necessary if you use the phase locked loop sketches.

You could use your RPi with radio or a direct serial connection, but as you see from the forums, emonCMS on the RPi is a work in progress. Both the Nanode and the 'rock solid gateway' for the RPi are stable, but they require a server running emoncms somewhere - which can be emoncms.org or your own.

Hello Matt,

Greetings from Oklahoma.

Regarding your choice of CT -- For each individual circuit you want to monitor, you'll want a CT with a max current rating as near to the breaker rating value as possible. From your description of your load center, and the circuits you want to monitor, your best choice would be 20 amp units. A 30 amp CT will work, but with a loss of some resolution. For your whole-house application, the 100 amp units would of course be the CTs to use, you'll need two of them, one on each leg, installed facing in opposite directions.

I run version 2 of the EmonTX, which uses a different sketch than the EmonTX Shield, but if memory serves, I didn't have to make any changes regarding line frequency.

Everything that *I've* seen points to the Raspberry Pi as the unit of choice for the base station. I use a Model B, and it has run, without issue, since July of last year.

You may want to take a look at this: http://boredomprojects.net/index.php/projects/home-energy-monitor

It's a 12 circuit monitor that uses an Arduino Due and Raspberry Pi.

Regards,

Bill

Robert,

Would a CT that has a 2 VRMS output at the rating Matt needs be suitable for his application? You mentioned 1.75V and I immediately thought of the 2V unit.

"you'll want a CT with a max current rating as near to the breaker rating value as possible"

Er, not quite! You need one (with its burden) greater than but as close as possible to the maximum load without a fault being present, but you also need to recognise that the ct will saturate and not read correctly under overload conditions, and that the breaker will typically carry 1.5 times rated current indefinitely, and 2 times for 3 minutes before tripping.

Circuit breakers used in the US (Matt mentioned he's in the US) are rated to trip at approximately 80% of rated load. They will trip immediately at 1.5 times rated load and definitely won't go 3 minutes without tripping at twice rated load. (Info taken directly from a Journeyman Electrician)

We get more for our money than you do, then! One that tripped at 80% of rated would be condemned as faulty here. I was looking at IEC standard curves: http://electrical-engineering-portal.com/purpose-of-miniature-circuit-br... (OK, I rounded the numbers and looked at the "must trip" edge of the band.)  IEC stands for International Electrotechnical Commission.

Here's some additional info that explains the 80% rating...

Circuit breakers have an ampere rating (typically marked on the end of the operating handle). This is the maximum continuous current that the breaker can carry without exceeding its rating. As a general rule the circuit breaker’s ampere rating should be the same as the conductor’s ampacity. In other words we would not want to put a 60 amp breaker on a 10 amp wire. Breakers are tested in open air, with a temperature of some 40 or 50 degrees C.

When a breaker is placed within an enclosure, cooling airflow is restricted; this reduces the ability of the breaker to carry a current to 80% of its ampere rating. When they are installed in an electrical enclosure, breakers will trip when a current in the amount of their rating is placed upon them continuously. Breakers are designed to be able to safely carry a current in excess of their rating for very very short periods of time to allow some types of electrical equipment (called inductive loads) such as motors to start up.

While not as common, some breakers are rated for 100% continuous loads. These are typically called supplementary protectors (SP) and not circuit breakers.

taken from: www.mikeholt.com/mojonewsarchive/EES-HTML/HTML/ElectricalCircuitBreakers~20030621.htm

Adding to that, the general principle of protection is: You determine the maximum rating of the circuit - normally the maximum continuous current of the connected loads. You then choose a protective device that will operate just above that, or as closely above that as is practical, taking (as Bill points out) ambient temperature and starting conditions into account. You then choose the cable size so that its current carrying capacity is greater than the rating of the protective device, taking into account the ambient temperature and the cable's ability to lose heat to its surroundings, and the voltage drop.

Of course, in the real world you might have to work the sums backwards to determine the maximum load given the installed cable and protective device, for example.

The last thing you want is to use the circuit conductor as a fuse, because in 'blowing', it might well set fire to your house. The cable's ability to lose heat explains why cables buried in the ground are comparatively small diameter (the ground conducts the heat away) while cables buried in insulation might be rated at only half of the capacity of the same cable clipped on the surface of a wall.

The last thing you want is to use the circuit conductor as a fuse, because in 'blowing', it might well set fire to your house.

Amen to THAT!

Since Matt's max breaker fault current is 20A (except for the 240V circuit that he says he doesn't particularly care about) would a CT that has a 2 VRMS output at the rating Matt needs be suitable for his application? You mentioned 1.75V and I immediately thought of the 2V unit. http://www.byramlabs.com/product_info.php/products_id/20486/product/WattcoreWC1%20SERIES%20WC1-025-RV002%20Split-Core%20Current%20Transformer%20%2825:2000mV%29

That, presumably, has an internal burden so at the maximum input voltage of his shield, it will read 22 A, which seems to fit his needs. But it's split-core, which will be substantially larger than an equivalent ring-core, and it looks as if space in his switchboard is limited. That's why I suggested ring core types, if he can disconnect each feed to thread the ct onto it.

[You don't mean "Fault current". Maximum Fault Current is maximum current that it can safely break without damage, which will probably be in the hundreds if not thousands of amps region.]

I'm curious as to how you Americans would measure a 220V circuit if you needed to.  Would you just clamp each leg, measure each respective to its corresponding V (~110V), and add them together in software?  That is, can you treat them as two unrelated 110V circuits and just add the two power readings together?

If you do that, are you effectively wasting a CT?  That is, will each leg be carrying the exact same current, or is it permissible for a 220V appliance to draw more on one leg than the other?  For example, can an appliance connect its heavy heating element across the 2 legs (for 220V) and then run its electronics say between one leg and Neutral (for 110V), or must everything be powered from across the 2 legs to keep them balanced?

Quite correct, the max fault current for a large proportion of the OCPDs made for use here is 10kAmps. I definitely got THAT term wrong! The WC-1 is quite small, (I use 2 of the 1V units to drive my EmonTX (v2)). From the picture in his post, he's got plenty of room to fit the WC-1s. The steel enclosure is ~100mm deep from front to rear. Disconnecting each circuit is easy - loosening a single screw on  each breaker is all that's needed. But, the solid core units would definitely take up less space. Probably comes down to how comfortable he is working inside his load center. Most folks here don't don't like to, or flat out won't touch anything inside a beaker panel. Not to mention that, strictly speaking, the only people that are supposed to do that kind of work are licensed electricians.

Hi dBc,

As Robert has said, the 120-0-120 split phase system we have is effectively three sources. One across both hot legs, and two across each hot leg and neutral. The load balance can vary quite a bit. It depends on how the loads are distributed between the two hot legs and neutral in the load center (circuit breaker panel) and the number of them that are active at any given time. In my case, I have an all electric house, so I have 8 loads across both hot legs, and the remaining circuits connected to each hot leg and neutral. (roughly split evenly between the hot legs) So, depending on the number of 120V loads I have on, and which leg they are connected to, the current in each leg can vary considerably. Usually, the 240V appliances we use will have an internal step-down transformer if the electronics or other parts of it (e.g. a fan motor) need 120V. Otherwise, the entire unit is designed to run on 240V.

Because of the imbalance, I use a CT on each leg, (facing opposite directions, because of the 180 degree difference) and add the two values. The Voltages on each leg, referenced to neutral, tend to stay relatively equal, usually varying from ~.3 to .5 volts from each other.

Taking an overall view of reported use from the USA, all bets are off regarding current balance. It appears that the voltage balance is reasonably good, but - especially where the incomers are concerned - two ct's are definitely required. Only if there is no neutral connection can you guarantee that only one ct will suffice. When you have two ct's you do indeed add the powers together to get the total. It's exactly the same principle as wattmeters - do the maths and you can prove that you need one less than the number of wires. It doesn't matter which wires you put them on, though. So equally valid are ct's on two lines and measure two voltages lines to neutral, or ct's on one line and neutral and measure voltages line-line and line-neutral.

Thanks Bill.  Can you explain why/how the voltages stay so close?  If you and all your neighbours were to foolishly load up one leg and not the other, would that gap widen, or is there something about the nature of the transformer that would keep them balanced even in that worst-case scenario?

When you say:

Because of the imbalance, I use a CT on each leg,

I assume you're talking about whole-house measurement?  (since the rest of your response implied there would be no imbalance on a single 220V circuit).

How would you measure the power consumed by a single 220V circuit?  I can think of a few possibilities, would they all work?...

1. measure V (~220V) and I for the circuit in question

2. measure V on one leg (~110V), double it, and measure I for the circuit

3. measure V and I on one leg, V and I on the other leg, and add the resultant powers together

#1 has the disadvantage that you need a whole 'nother V measurement

#2 relies on the legs staying at the same voltage

#3 wastes a CT, since each CT will be measuring the same current

I guess there is the obvious variation of #3, which is to only use one CT, and add the two V leg measurements together to get the actual V feeding the appliance.

Right, whole-house measurement. Robert's post just above yours answers your question about measuring a single 240V circuit. With only two wires, you only need one meter.

The power companies limit the number of customers they connect to each transformer, typically 4, but sometimes as many as 6, depending on the size of the buildings.

As most folks don't have any way of knowing who among them on the same transformer has what loads on which leg, and the load center circuit breakers alternate their connections to each leg, (i.e. even numbered breakers on one leg, odd numbered breakers on the other leg)  the chances of that happening are going to be mighty slim.

Robert, I have to defer to you for the answer about why the voltages remain fairly close to each other...

I think what Robert is saying is that provided the appliance is fed L1, L2 and Neutral, there's no way of knowing how it's going to draw it's power, and so 2 CTs are required?

The driving EMF for the two halves of the secondary winding will always be the same (to within the accuracy of the winding process and the flux distribution in the core), so there remains only one explanation as to why the two voltages are closely balanced - the source impedance, including the resistance of the feeder cable on the secondary side, is relatively low, and/or the two currents are relatively closely matched; hence the voltage variation due to the difference between the two IR drops is equally small.

I think what Robert is saying is that provided the appliance is fed L1, L2 and Neutral, there's no way of knowing how it's going to draw it's power, and so 2 CTs are required?

Right. Quoting Robert from another post: (http://openenergymonitor.org/emon/node/3265)

Classical theory says you need one fewer wattmeter than the number of wires, and a voltage and current input pair to the emonTx is exactly equivalent to a wattmeter, so you must measure two line currents and two line-neutral voltages.

The driving EMF for the two halves of the secondary winding will always be the same (to within the accuracy of the winding process and the flux distribution in the core), so there remains only one explanation as to why the two voltages are closely balanced - the source impedance, including the resistance of the feeder cable on the secondary side, is relatively low, and/or the two currents are relatively closely matched; hence the voltage variation due to the difference between the two IR drops is equally small.

Got it. That ever-present energy robber "copper loss." (among others) Or in this instance, more or less equal copper losses between the two legs. I should've gotten that one. Guess I've been a system level tech too long. It's been a long 40 years since learning the basics!

"It's been a long 40 years since learning the basics!"  Greenhorn!  I finished my Elec.Eng. degree course in 1971. ;-)

This is fascinating stuff, a bit far afield of my original questions, but very apropo for how I will finish off my project after the individual circuits are monitored. I'm curious how different the sums of the individual circuits, the two legs, and the power company meter will be.

After some additional research Robert is correct, the 30 amp CT I mentioned is the same as the 100 but with a built in burden resistor. So scratch that. 

The space saving nature of solid core CTs appeals to me since space between the circuits is tight. Based on Bill and others comments above I located these, CR8410-1000 The 20 amp model puts out 2.1v at load, which looks promising.

Seems I have some time until the rfm12pi comes back in stock to work this out and whether to go with 5 unos and shields or the more homebrew route as in the 12 circuit monitor above.

-mjf

Ya got me there indeed, RW. <grin>

Mine was mere 6 month course during 1974, in basic electronics, avionics navigation systems and airborne search/weather radar.

dBC,

Here's a graph of the Leg1 and Leg2 currents my system monitors. As you can see, while there's not a large difference between the two, it varies constantly.

I think what Robert is saying is that provided the appliance is fed L1, L2 and Neutral, there's no way of knowing how it's going to draw it's power, and so 2 CTs are required?

Thanks Bill, very interesting.  I assume the units on the y-axis are Watts?  And I'm guessing that your fridge is on the red leg. It looks like all your significant loads are 220V loads, hence the synchronised spikes?

Right, y-axis = watts. I'm not sure which leg the fridge is on, but I'd say you're guess is more than likely correct. The big spikes are the water heaters - one is a 4.5kW unit, the other, 5.5kW. Also correct regarding the sync'd spikes. Loads more than ~2kw are typically 240 vice 120V.

Matt:

It looks as if that CT will do the job for you. As it stands, you'll clip the input at 16.76 A, if you put a 538 Ω external burden resistor as well as the internal one (106.26 Ω according to their formula), that will reduce the output voltage and allow you to go to 20 A maximum reading before the emonTx Shield's input clips. (It is probably better to go low and use a 510 Ω standard value, which will give you a bit of headroom - 1.738 V rms or 4.91 V p-p given a sine wave. The current calibration constant for that combination will be 11.508 )

[Note: you overload a CT with a higher value of burden resistor, so shunting the existing burden with another is actually lowering the load on the CT - think current, not voltage, when you're dealing with CTs.]

I have a follow on.

I am also in the US - and I need to monitor a few of the 220V circuits (heat pumps/water heater/etc).  How do I do this? I have a dual socket breaker with 2 wires into it.  Do I need to CT clamp both? or can I just clamp one and double?

These look like the top-left breaker from the original image in this thread.

Thanks

M

The general answer is you need one less CT than the number of wires feeding your load (discounting earths of course). So if any of your 220 V appliances have a neutral connection, you need 2 CTs, if none have (i.e. there is nothing connected to that breaker that needs 110 V), you need one (and you measure the 220 V of course).

Thanks Robert - i don't believe there is a neutral - the entire circuit will be 220V - so just a single CT then.  I'm guessing it doesn't matter which leg i put it on as they should both have the same current correct?

So I would need to set the single CT to 220V and make sure I only monitor other 220V circuits on that emonTX?

"So I would need to set the single CT to 220V and make sure I only monitor other 220V circuits on that emonTX?"

That doesn't make sense! A current transformer measures current - the voltage on the wire is irrelevant. You can monitor other circuits, but you need to be clear what you're measuring. Assuming the 220 V is reasonably accurately balanced about the neutral - and most reports indicate that it is - you will lose little in accuracy if you either:
(1) Measure Line-Line and halve the voltage for the 110 V circuits, or
(2) Measure Line-Neutral and double the voltage for the 220 V circuits.

If there's a 2 V difference in the two lines, so one is 111 V and the other 109 V, you've got roughly a 1% error in voltage/power/energy in the 110 V loads if you're measuring L-L, and 1% again in the 220 V but 2% in the other 110 V Line if you're measuring 110 V.  So if you're prepared to lose a bit of accuracy, then yes, you can measure both 110 V and 220 V circuits. If all your high power loads are on 220 V, it makes sense to measure that to keep the overall error to a minimum.

Of course, you'll need to edit the sketch to account for the different voltages.

(I'm assuming you are measuring the voltage - if you're using the declared voltage and measuring only current, then it makes no difference, but you'll still need to edit the sketch.)