Anybody moved forward with switching Immersions to accomate a variable PV source of power?

Tom, see the second post in this thread - it's very inefficient at low light levels. (And a lot of the rest of the first page, for that matter.)

Johnuready, are you the John U who emailed me about this at the end of December? I did reply but didn't hear anything further so wondered if my reply went to whoever's spam folder or something. For background, here's my reply to them:

No, sorry, I don't know of anybody who's actually done this,
yet. At the time I wrote that page I was under the impression
that the system described here:

https://lifeattheendoftheroad.wordpress.com/2012/08/18/work-in-progress/

had multiple immersions but it turns out that it just has
a single custom 24 V one.

That came from http://www.tpfay.co.uk/ who seem like a good
possible source for other more complicated designs. E.g.,
two element immersions for PV in a single standard boss:

http://www.tpfay.co.uk/our-products/immersion-heaters/domestic-immersion-heaters/

I'd be interested to hear more about what you have in mind.
For just direct PV to multiple immersions then an Arduino
ought to be very suitable. Would you be using direct FET
switching or via relays/contactors or what?

For my own house I plan to have quite a few small 60 volt
immersions of maybe a bit less than 1 kW each for a nominally
48 volt system (16 3.2 V LiFePOâ‚„ cells in series) which when
properly balanced and fully charged to 3.6 V/cell would have
a voltage of 57.6 V. Overall management will be done by
something a bit more power (ARM & Linux) but low-level control
of the switching will be via an Arduino.

This is not quite the setup considered in that blog page, of
course, as there will be batteries present. Still, the idea
would be to match the solar (and, perhaps, wind) production
when the batteries are full. Also, there'd be the option to
prioritize at least some water heating when the batteries
are partially full (hence the higher-level Linux-base control).

As I say, I'd be interested to hear more about what you're
thinking,

Assuming the system is sized for the DHW need and there is nothing else to use the output for in summer.

First thought is to design the system for when the sun is at 50% strength, and just except that you are not getting the highest possible output in summer, but the panel size would have been chosen to give enough hot water at other times of year anyway.

http://www.greenbuildingforum.co.uk/newforum/comments.php?DiscussionID=11149 is anther thread about the same problem. Other links I found http://www.ata.org.au/forums/topic/12294

http://www.google.co.uk/patents/US5293447 and http://www.google.com/patents/US20150104160 may be of some use.

I'm gearing up to start on a project, look at Navitron under Johnuready "PV water heaters". I was thinking of using a 2kw Aztec Trianco system boiler bits adding the mosfets and using the pump etc to push water around a coil in a tank. Now thinking of building the mosfet load device to test the theory of monitoring a variable voltage and current from a bank of PV panels.

I have designed a switching matrix in my head allowing to switch between 3 No. Immersions in one unit in Parallel and or Series. It could be easy to run an Arduino device to monitor the voltage and current to switch between immersion configurations, hence as close to max transfer you can get by resistance.

In my case it's got to be off grid as DNO said no to any more panels on the grid and with cost of PV panels coming down at self install is real.

Any ideas?

Thanks for that interesting Navitron reference Johnuready - I'd missed that thread. I like that Brian Drury design - particularly his first bipolar emitter-follower circuit.

I'm wondering about something like it with multiple transistors in series. It'd be slightly less appealing as you'd have to insulate the cases of the transistors but would allow the whole thing to run at a bit higher voltage. It's all a compromise but I'd think there's a lot to be said for getting the voltage from the PV as high as possible (to keep the amount of cable required down) while at the same time staying within the ELV voltage limits for safety.

Johnuready, do you want to say more about the switching matrix you have in mind?

ELV = extra-low voltage (where mains is considered low voltage). Less than 60 V AC or 120 V DC.

I was thinking of using an Aztec 2kw boiler by taking out the heating section and using the mosfets on a pipe type device and heating water that way. I use one of these for my only heating on towel rails and 2 radiators they work very well as designed. On the hacked version you could push the water round a tank coil. It comes complete with expansion tank and pump in the case. Look on eBay: 371279987503

You see on Ebay for <£100.

Now though the switching of Immersions is beginning to sound more exciting.

My thoughts

Immersions in Parallel

I1+I2+I3 = A+D+F
I1+I2 = A+D
I1+I3 = A+F

Immersions Series

I1+I2+I3 = A+C+E
i1+i3 = E
I1+I2=C+D

Immersion & Series

I1+I2 = C+D
I3 = F

Personally, I get confused by diagrams which don't have (conventional) current flowing downhill. Here's my attempt to redraw like that:

If it is DC, you really need DP switching don't you. You can't risk a gas build up if there is a fault.

How about?

Diode orientation is on the assumption that “Live” is positive. Obviously reverse if the other way round.

I've renamed the switches to avoid confusion with your naming.

Any immersion on its own by just closing the switches above and/or below it. Any combination in parallel by again switching on the required individual immersion's switches.

I1 and I2 in series by just closing N2.
I2 and I3 in series by just closing L2.
I1 and I2 in series with I3 in parallel by closing N2 and L3.
I2 and I3 in series with I1 in parallel by closing N1 and L2.

With all switches open it's I1, I2 and I3 in series which'll cause a trickle of heat into the tank. Maybe having a separate thermostat-driven all-off switch separate from the power-point tracking function would be a good idea anyway. Or there could be a switch in series with D1 or D2 to stop this.

I1 and I3 in series is not possible as drawn; it'd need another switch (from the bottom of I1 to the top of I3, maybe via one or both diodes or another diode).

Assuming the diodes keep dioding I don't think you can make a short circuit.

Thinking of running 4 panels - 120 volt @ 9 amp

For this kind of project, how do you go about: fuses, RCDs, earthing, and separating the high-power DC from the logic DC and the pipes?
Is it notifiable Part P etc?

Posted By: johnuready120volt DC is classified as low voltage

From a regulation point of view “low” voltage is up to about 1000 V or so.

ELV (extra-low voltage) has various definitions for different circumstances but is generally less than 50 V AC or 120 V DC. Lot to be said for keeping the Voc of your panels on a cold sunny day to less than that.

Edit: looked low voltage up; it's up to 1000 V AC or 1500 V DC between conductors, 600 V AC or 900 V DC from any conductor to earth.

I am considering this sort of thing. Batteries are a no no, doubles the cost of the installation. I am of the opinion that a copper tank filled with water would absorb heat very readily from an external heater (remove the foam first!). The nicest way would be to use stainless steel banding with a layer of mylar tape between it and the tank, then it could easily be tensioned with external springs. If the resistors are selected to have a binary value (R, 2R, 4R, 8R. . ) then by switching them in via suitable logic 15 of resistance between 8R and about .5R can be used.
To find the best value you need either a multiplier IC (V X I = W), so you measure the volts and the current, look at the output of the IC, remember it. Alter the resistors, see if the new W is greater then the first. repeat ad nauseum. Or just calibrate the resistor switching with a voltage sensing IC and a load of comparison circuits. So you go out read the voltage see what the current is, fiddle with the resistors until V X I is a max, then set one comparator to select this combination, repeat for different levels of sunlight. I would not use a microprocessor unless you like programming them and are prepared to keep the thing going for at least 10 years.
Frank

Posted By: chuckeyswitching them in via suitable logic

I suspect you might find gray codes useful in this application.

I would not use a microprocessor unless you like programming them and are prepared to keep the thing going for at least 10 years.

Indeed, best to choose a popular design that will be available for a long time. Pi would be my pick at the moment for that reason.

Posted By: chuckeyjust calibrate the resistor switching with a voltage sensing IC and a load of comparison circuits

But the same question of longevity applies to this arrangement, not to mention the logic circuits to do the encoding and all the possible dry solder joints.