Electronics help please

[Davo]

As a newbie to PV and the Navitron blog the clever folks on here have inspired me to have a go at building a very basic load controller (without it costing me an arm and leg). I have 2 spare CT coils (from cheap hme power monitors) that I thought I could use - 1 for the PV generation and the other for the house load. Simply rectify their output and feed them into a voltage comparator. The snag is the voltage is very low from the CT coils for the useful PV output at 3A to 10A circa 0.2vac to 0.8vac. Sadly my schoolboy electronics has let me down at the first hurdle (it was 50 years ago).
can anyone supply a simple diagram to amplify the signal to something useful - say 0 to 9vdc. I have a circuit for the comparator (12v) and the bits to build it.



In anticipation - Please keep it simple
Cheers

[johnrae]

How are you measuring the output of the PV panels (DC) using a CT (AC only).

Starting such a project from the beginning I'd have suggested using hall sensors, which operate for both AC and DC and give a linear output of around 5 volts DC.

[Davo]

Having a couple of CT coils I thought I would have a play to see if I could use them. I didn't want to spend lots on other kit  - I thought (probably unwisely) it would be a simple job of soldering a few cheap components and a relay onto a bit of veroboard. I suspect with more knowledge I would have considered someother method.

[Davo]

Sorry - didn't read your reply fully.
I am using the CT coils on the AC side. One measures the PV output (AC) and the other the house load (AC). I thought I could rectify their output and feed the 2 values into a Voltage comparator circuit. But the ac voltage range from the CTs is very low at the values I want to compare (circ 0.2vac to 0.8vac).

[Davo]

I have knocked up this circuit diagram (from bits of diags I have borrowed from elsewhere). I don't have an electronics background, so please be gentle.

The idea is: If the PV output is greater than the house load by a predetermined amount (set by the Offset/bias pot), then the relay operates to control a load (say immersion heater or heater). Once the relay has operated, the house load increases - R Hysteresis value is chosen to match the increase to stop hunting. All the resistor values are theoretical, so would need a bit of experimenting with the values.
The 2 Comparator inputs would have an ac element in them, but they would be in phase, so in theory the ripple wouldn't matter.
A couple of capacitors maybe be needed across the inputs to the Comparator though.

Would it work anyone ?

[Twenty4Seven]

Where is Ken B these days?

[pb]

The idea is: If the PV output is greater than the house load by a predetermined amount (set by the Offset/bias pot), then the relay operates to control a load (say immersion heater or heater). Once the relay has operated, the house load increases - R Hysteresis value is chosen to match the increase to stop hunting. All the resistor values are theoretical, so would need a bit of experimenting with the values.
The 2 Comparator inputs would have an ac element in them, but they would be in phase, so in theory the ripple wouldn't matter.
A couple of capacitors maybe be needed across the inputs to the Comparator though.

A couple of things to think about:

I'm not sure it's safe to assume that the waveforms seen by the two CTs are going to be in phase.  If the house load is anything other than resistive then the current may end up being a non-sinewave, and/or leading/lagging the output from the PV.  You probably need to average out the current to some approximation of a DC level.

It looks like there might be enough leakage to ground through Rhysteresis that the transistor will never switch off completely.  That might or might not be a problem in practice but it would be worth checking.

[Davo]

Hi PB
Thanks for your reply and comments - very useful and appreciated.
1) I was clutching at straws hoping the 2 inputs would be in phase - it would save me having to complicate things. I had considered they might not be and did think capacitors across their inputs to the comparator might sort that.
2) RHysteresis - I knew there was going to be an issue with hunting once the load is switched. The problem was how to solve it. It's a bit of a stab in the dark with RHysteresis resistor. As I say, all the resistor values will need fine tuning.
I have a couple more bits to get before I can get testing - so watch this space.
If it goes very quiet, you can assume it's been assigned to the recycle bin (like many of my projects)
Thanks again for your interest.

[artful_bodger]

Hi Davo,
This has peaked my interest, as I'm doing something similar at the mo. I'm firmly in the research and play part of the design as I am currently off grid, waiting for the grid connection (since july 2010)  Should have it by March.....
I can get my PV power direct off the grid inverter, in amps RMS.  I've yet to look too closely, but I notice all the GTI manufacturers quote the injected power at a power factor (PF) of 1.  Hence your coil on the PV side (I assume GTI output) should give a true power.  I'm lazy, so I'll extract the current from the GTI using a microcontroller.

I think you have more electronics experience than you're letting on.  The circuit is very good for a first stab.  It's the hard part to start and the easy part to criticise.  I'd suggest some small caps after the diodes to get a more stable DC.. The comparitor wont thank you for 2 wobbly waveforms. Start with 0.1u.  Too large and you'll lose response time.  Consider replacing the virtual earth with a zener or cheap voltage regulator as this will harden it up and give a smooth(er) output - delay that change until you've got something working.  Try some caps on the resistors for now - say 0.1u.

To get a larger CT output, increase the burden resistors a factor.  Be careful tho' as the voltage will increase and can get dangerous. There no difference between a CT and ordinary transformer and there's a lot of turns on the secondary.  If you want a simple amplifier try a 741 in classic configuration http://www.electronics-tutorials.ws/opamp/opamp_2.html.  On the + terminal take this to your current 10k/100k point.  Hmm this might give more problems.  I'll need to think some. 

I don't think phase shift will be a real problem, as you're looking at average output, not rms, so that's a compromise already. 

The only down side of this that I can think of is that whilst you will know when you're solar positive, you wont know by how much.  So if the sun comes out weakly, you may then switch in a large load, which will drop out quick quickly once detected.  This on/off might cost you more in grid charges!  Consider a minute timer?

[Davo]

Thanks Artful Bodger
1) Yes the capacitors are a must, thinking about it - I will start with a 0.1uF as you suggest. I did wonder about going larger though - adding a bit of CR time would slow response down, which could be a good thing. It would act to smooth out the continuous fluctuations etc. It's not not life or death to switch on or off the split second it happens.
2) Sadly to say, you lost me about the 'virtual earth' and zener - I told you I'm an electronics beginner.
3) I had a careful play with the CT coils yesterday. They didn't exactly behave how I expected them to. In the end I got the acV output against a load current (as shown in the graph) - but was expecting higher open cct voltage ! As I previously mentioned, they are borrowed from commercial wireless load monitors (Owl type). I did wonder if there were extra components in them to stop the general public killing themselves when they plugged the CT coil into the transmitter.
4) Will try desperately to steer clear of op-amps etc if possible
5) Not too fussed knowing about knowing by how much I am solar positive. The off/sett/bias resistor would be sized so that I need to be solar positive by at least the load I'm going to switch (more than likely my immersion heater). RHysteryses would also be sized to ensure once the load is switched in it offsets the new loadvalue by altering the bias to prevent hunting.
It all seems very easy in theory ! Once it's warm enough to get in the garage and get the soldering out we shall see.

Thanks again for your interest

[artful_bodger]

Hi Davo,
My cunning plan to catch you out failed - virtual earth.  In your drawing, you don't 'technically' have a virtual earth - I used the expression incorrectly  - not to trip you up - just 'cos I wasnt thinking.  Sorry - hope I havent confused you.

When you're using an comparitor on a single power supply, you need to create a middle-ish point, as it doesnt like comparing volatages near the supply rails.  In your case the mid point of the 100k & 10k.  Stick a cap from there to ground.  This needs to be stable otherwise the input to the comparitor will jump around even more.  If you find it is unstable, we can talk about zeners.

Had a look at the owl info.  The burden resistor must be built in.  Is it a 2 or 3 wire connection?  BTW As there is no voltage connection (you can do it from some CTs) there will be no phase info, so the power figures on the display will be inaccurate when the power factor is not 1 (most of the time..generally speaking)

Dont be afraid of opamps.  They are really useful.  A comparitor is a kind of refined op amp.

Hadnt thought about sizing the comparitor output above the available load.  I guess this heater wont be 230V operated then.  Either that or you have a massive array!

Garage!  Get that soldering iron on the kitchen table.  Better still, buy a breadboard and get it working before committing to solder.

[Davo]

Cheers Artful for all the suggestions.
I knew there were some clever folks on this forum who could fine-tune a laymans effort.
1) Good point about the 100k and 10k. I was going to use 100k and 100k but I thought the 10k option may give me more room for manoeuvre. I suspect I won't need that room in hindsight.
2) The CTs are both 2 wire. I was confused when I first started messing with them. The graph on my first post shows what acV I read on the 2 wire output of the CT for different values of load current through the CT (just my DMM across the O/P). It's also pretty linear - so I suspect you are correct - the Burden resistor is already built in (when I destroy it experimenting I will take it apart and have a look  ). This means I can dispense with my 100ohm.
3) Still steering clear of op-amps. If I can keep it simple I will.
4) I only have a 3.8kw system and the immersion is fed via 240/110 transformer. The methodology (in my head) was for the output only to switch when there is excess PV available above house load. What I have on paper maynot achieve that ! Obviously if the house load goes above say 3kw I wouldn't want it switching
5) Not too much excess PV at the moment, so not too much of a rush. Good idea about the breadboard - I'd forgotten all about them. When I am happy that it may just work (without destroying all my components) it should only take an hour or so to knock-up ! mnnnnhh

[ericw]

Unfortunately because the diode voltage drop will be a significant proportion of the output voltage from the current transformer the simple circuit will introduce errors.

If you use a mains powered circuit (which you must do to get voltage * current) you can get the phase of the curent flowing in a single CT on the mains supply to work out which way the current is flowing.

Look here to find how to accurately measure power http://openenergymonitor.org/emon/node/59

[Davo]

Cheers EricW
I knew it wasn't going to be straightforward - so thanks for the pointers.
1) re the 2 CT circuits and diodes - Surely the biasing arrangement (100k and 10k) would take care of any PD across the diodes. If both circuits are the same, with the ac (1/2 wave rect sat on top of the DC) wouldn't this make the diodes OK to use ?
2) I sort of 1/2 understood the bit in the link - but most of it was beyond my understanding.

I don't think what I was proposing to do would have resulted in a very accurate method of measuring and switching loads. It would be very agricultural (like everything I do  ). I can see from other posts on this forum, that the only real way of doing it is like you say, measuring Amps, Volts and probably using PLC (or PC). My main criteria was to keep it simple and cheap.

[artful_bodger]

Food for thought.  Or in my case, food that kept me from sleeping last night.

A 3kW immersion on 110V will be about 750W.  To raise the temperature of a 150l cylinder from 10 to 60C needs about 9kW of energy.  If we ignore that it loses 1kW/day, with 750W, this is about 12 hours of heating.  As you have a large array, you probably will have a 750W excess, but  for how long?  I currently have 500w surplus on a 1kw array, with another 5 hours of (potential) sunlight.  I would have more, but someones got the router and a PC plugged in.  So I suspect you'll have  a greater surplus.  You could think about a large dimmer switch to control the power and hand tweak it.

I think your circuit is fine.  The comment about the diode is a good one.  The AC values you gave are most likely RMS so the lowest figure of 0.2 to 0.8RMS is 0.28 to 1.2V peak.  If you use a germanium diode, that's 0.2V gone, so a large error at 5A, which is roughly where a 1kw heater starts.  It might work in your favour, as the error will mean (I think) you are firmly into surplus energy. (you could look here  http://sound.westhost.com/appnotes/an001.htm at precision OPAMP rectifiers, but dont look too carefully)  Equally, the out of phase load I & V should work in your favour as the current vector (bear with me) will be longer and hence will make you think that you are using more power than you are.  Hence your system will be pessimistic.  Better that way.  Comments anyone?

My brain hurts.  Time for a cuppa.  Can you please just build this so we can all go back to reading the paper?