Energy monitoring - going beyond smart meters...

So there’s a belt and braces approach to limiting that voltage. Good.

You agree, though, do you, that there isn’t a principle that says, as that voltage rises, the current goes down?

Because that’s not how induction works.

As you said, Modbus (classic) is based on the RS-485 physical layer, which was both well documented and quite common. Annoyingly, it is just a bit different from a RS-232 serial connection, but that is to be expected with a 2 wire design. All these field buses are driven by cost and simplicity, so a lot of them use just 2 wires. 

But it is very much a legacy standard at this point, and prices for the interfaces have gone up a bit. You can get a number of hats, and you can get USB adapters with corresponding software. I would do that just get a work endpoint, unless you already have one. 

You should also be able to buy a used PLC for little money. Or just an endpoint? It is a bit of a different world, but it has its own kind of logic. 

I realise that current doesn’t fall proportionally to the rise in voltage in a standard power transformer. But I believe a current transformer has a characteristic in which this suggestion isn’t too far removed from the truth!

The SCT-013 is designed to connect across a fixed resistor {*}

The cross-section of the ferrite ‘ring’ is massively greater than the cross-sectional area of the wire in the secondary coil. The flux is very low.

As the voltage induced across the secondary increases, you rapidly reach the point where the secondary is saturated.

If you replace the resistor in the emonPi with a higher resistance equivalent to human skin, the current remains fixed in proportion to the primary current, so the voltage rises across the secondary. It simply saturates quicker.

At saturation point no further energy can be passed to the secondary.

{*} The emonPi design uses two resistors with the input being applied to their mid-point. This allows current to be measured in either direction.

Thanks very much for this explanation, @Transparent - you’ve pitched it at just the right level for me to have a chance of understanding it (and being convinced).
 

I just need to remind myself a bit more about  how B and H arrange themselves in response to electric currents in the presence of materials like ferrite, i.e. how ferrite materials do their job.

At this point, the way I see it, there is a certain amount of flux around the primary. Magnetic flux is determined by the current and, as you put it so concisely, there’s only so much flux to go round (pun intended, I’m sure). The ferrite works, I expect, essentially by concentrating that flux into the volume it occupies (and it really only achieves that just as the ring is closing - until that moment, the gap in the nearly closed ring means that a significant amount of flux does not link the secondary winding). (At this point, I’m vaguely tempted to see if there is an easily accessible point in the Emonpi to which I can hook up my oscilloscope in the hope of observing that induced current appear just as the clamp is being closed. If not, I’ll have to rig up something separately.)

[But … hang on … if the protection is in the input stage, inside the Emonpi, how do things work (and how does that help) when the current transformer is clamped onto the supply cable if it’s not connected to the Emonpi? How do the zener diode and back-to-back diodes help if they’re elsewhere?

Edit - the zener diode in the Emonpi can’t help, but the back-to-back (zener!) diodes are in the clamp itself, which is where something is needed!

(I need to read more carefully the first time...)

It was the possibility of significant voltages being present on the connector, dangling there, unconnected to the Emonpi that I was originally asking about. I seem to be back where I started. If I do rig up something separately, I suspect I do actually need to be careful.

Resuming ...]

So, once everything is in place: clamp is closed, the ferrite is essentially continuous, the secondary winding is short circuited by a zero resistance conductor, I can try to answer my own question. Which is, what happens as the resistance of that short-circuiting conductor is increased from zero?

My (admittedly simple-minded) reasoning was that, other things being equal, the current remains the same and, as that resistance is increased, the voltage grows with it. What I will try to work out for myself next is, how best to understand the relevant ways in which “other things being equal” fails to be a useful starting point.

Thanks again :-)

I’ve just checked the specification for the SCT-013 current-clamp on the YHDC datasheet. We have no need to be concerned. It includes a pair of back-to-back diodes to limit the voltage rise! 

 

I assume you intended to link to this, @Transparent? That’s a good account (meaning, one I can understand ;-))

Just to wrap up this little exchange, I offer the link to OEM’s very own independent test report on the current clamp (by the ever-helpful Robert Wall) which says

The main text refers to tests on an earlier model, the text in italics refers to the item currently available.

In light of that, and to continue from my previous comment/question:

So, once everything is in place: clamp is closed, the ferrite is essentially continuous, the secondary winding is short circuited by a zero resistance conductor, I can try to answer my own question. Which is, what happens as the resistance of that short-circuiting conductor is increased from zero?

My (admittedly simple-minded) reasoning was that, other things being equal, the current remains the same and, as that resistance is increased, the voltage grows with it. What I will try to work out for myself next is, how best to understand the relevant ways in which “other things being equal” fails to be a useful starting point.

The answer must be that, for example, if the mains supply current in the primary winding is at the limit, 100A, then the current transformer output rises from zero to 50mA as the clamp is closed. If we then start to increase the resistance through which that current is flowing, the current remains unchanged, and the voltage across that resistance grows, until it gets to around ± 7.5 V, which is when the back-to-back zener diodes do their thing. At this point, the variable resistance will have got up to ~150 ohms, I suppose? As the resistance is increased beyond ~150 ohms, the voltage across it remains at ± 7.5 V so the current flowing through it must fall in accordance with Ohm’s law. I can cope with electronics at the level of what would once have been A-level physics :-)

As you will see, I have to explain things to myself in very elementary terms, but hope that I might still have something to contribute to collaboration…

PS - why did OEM opt for the current output version of the clamp? I assume this was because they can gain better control over the accuracy of the current measurement. By connecting the output to their own load, consisting of possibly higher precision/stability resistors than are hidden inside the clamp, they can estimate (and improve) the uncertainty on the Emonpi’s measurement of supplied current (and ultimately supplied power).

@Simon1D wrote

why did OEM opt for the current output version of the clamp?

I believe that sensors based on current are favoured because they are more tolerant of long cables, poor connections and hostile environmental conditions:

The SCT-013 current clamp is feeding a 470kΩ resistor in the emonPi monitor.

So if I use a kinked connecting wire running from my garden shed the entire length of the garden, through puddles, and with breaks mended by twisting the internal wires together, I might add 200Ω resistance to the path. I’ll still be better than 99% accurate on the current measurement

Remarkable. If I understand correctly what you mean by “feeding”, did I make a mistake in the comments about 150 ohms? Wouldn’t a 470k load put the zener diodes well into their limiting region? why would the signal from the current clamp relate to the mains current at all?

I understood the point of current mode to be that the load should be well below that point [150 ohms], so that the current is indeed hardly influenced by lead resistance (although, obviously, 200 ohms would not be low enough for that to be true).

Perhaps that 470k is in parallel with something much smaller, so that the combined resistance is a lot less than 150 ohms. (Still important that both be precision resistances tho...)

Agh - mea culpa!

I woz reading the schematic incorrectly.

I looked at the potential-divider setting the mid-rail voltage of 1.15v when I should’ve been paying attention to the pair of 22R resistors in parallel.

That was a waste of time creating my earlier pretty graphic.

I’ll go back to building my oak porch for the rest of the day

I suspect you’ll have as much fun with that as I’m having implementing a demonstration of how to solve that old “missing data” problem...

Take care, BW

(I did get as far as looking at that diagram after my comment, but really really am not proficient enough at deciphering such things to have spotted that those resistors were connected in the way I was thinking of ...)