I agree, the drop in current causing a corresponding drop in voltage results in very inefficient use of available power in less-than-fully bright conditions. A post on my blog on the subject:

http://edavies.me.uk/2012/11/pv-immersion-gotchas/

Sorry if it's grandmothers and eggs but do note the issue of using a thermostat designed for AC with the DC output of a PV panel. Could result in a nasty accident.

Posted By: SprocketWhilst there is an optimal peak point to do with internal losses, as long as the heating element resistance is selected such that the voltage across it at your current does not exceed the nominal cell voltage you should get pretty close to maximum power transfer.

Indeed. But if you pick the resistance to be good for one level of sunlight it will be a very bad choice for significantly different levels.

> But if you pick the resistance to be good for one level of sunlight
> it will be a very bad choice for significantly different levels

OK, maybe, if the internal resistance of the PV cells is high enough, but can you explain a bit more what you mean? What are these losses for typical useful irradiance levels?

I can see how if you choose for something like 400W and then generate 500W the extra power is lost because the panel reaches maximum voltage and the resistance of the heater will then limit the current to below the maximum.
But I can't see how if you choose for maximum output you will lose much efficiency when the irradiance is lower. The mismatch will mean more a bit more heat lost in the panel's internal resistance but not all that much surely? Sure, if light levels are really low, but do we really care about that? I would quite like to see numbers for just how many watts in what circumstances.

Charging and discharging batteries will be horribly inefficent. Even ignoring charger losses, if you put 100Wh into any battery, you certainly don't get 100Wh out. Not to mention the cost and deterioration of the batteries with use.

If you really want to match the panels to the heater for that extra bit of efficiency at low light levels surely it's just a matter of a few relays and a simple controller to adjust number of elements in series or parallel. There are no significant losses in the relays so that would be close to peak possible performance. This is way simpler electronics than a typical PV inverter.

The resistance of a 3kW immersion is about 20 Ohms, so at say 120V it will draw 6A which the 1kW (4x 250W 30Vdc panels in series) can just about do from 70-100% sunlight reasonably efficiently.
When the light drops, the current will drop so say at 3A (38% sun, assume max current approx 8A) the immersion will run the panels at only 60VDC (pulling only 180W from the PV array losing a massive 200W! (38% of 1kW = 380W)...lower levels are even worse.

Yes I agree switching the panels from 4x1 in series to 2x2 and 1x4 is possible, but would requires quite sophisticated electronics to control and would happen frequently, so would prefer to use transistor rather than relay switching, requiring some serious design, heatsinking, weatherproofing, mounting etc..
I accept that batteries wont be that efficient but its got to be better than doing nothing! When the light level rises the batteries wont be required so I guess a bypass could easily be designed for this.

Not sure what you mean by 'elements', do you mean the strings of cells within the panels themselves? these are not normally accessable, or do you mean the individual cells?

The capacitor idea was my orginal thought but unfortunately the equation CV=It, so C=It/V, means a capacitor of thousands of Farads which i dont think exist just yet!

Yup.. got it re Heating Elements, but again quite sophisticated switching electronics and only 2 different elemnts in one boss doesnt give much room for power matching. Probably better matching could be achieved switching the PV panels

Posted By: welshjim99...and only 2 different elemnts in one boss doesnt give much room for power matching

Yes, but you don't need much more. For your 1 kW example suppose you had a 600 W element and 2 x 200 W then you could have 200, 400, 600, 800 or 1000 W easily. That should match pretty well. Three relays driven via transistors from an Arduino or something which can measure the panel voltage. So long as the software doesn't keep clicking the relays all the time that should be fine, I'd think.

[Edit: better 150, 300 and 600 W elements allowing 150, 300, 450, 600, 750, 900 or 1050 W. ]

Posted By: Ed DaviesSo long as the software doesn't keep clicking the relays all the time that should be fine, I'd think.

Just put a cycling delay on it, say 5 minutes between samples regardless of what is happening.

Maybe these people:

http://tpfay.co.uk/

I think I got that reference from Paul Camilli (http://lifeattheendoftheroad.wordpress.com/) who I also think has had “funny” elements made up by them.

If you had three elements designed to be 500, 1000 and 2000 W at 240 V you could connect them up fine (in parallel) for 240 V operation. At 120 V DC they'd be 125, 250 and 500 W giving the options of 125, 250, 375, 500, 625, 750 or 875 W with spare power capacity to run the PV at a bit more than Vmpp in bright sunshine.

[Edit: should really only use the 1000 and 2000 W elements at 240 V.]

Thanks Ed & Sprocket, had some sun today at last and read a massive 6.37A through the immersion, rathar than the 0-1A whislt cloudy confirming my suspicions.
I am endeavouring to log all data; current, voltage and sunlevels from a hand made reference cell and of course water temperature, and will happily present the findings, with of course an automated immersion switching circuit, probably using immersion voltage as the reference source.
I will contact tpfay and see what they can make for me...its a shame I've got to go to work tomorrow!
Thanks again...

Welshjim99 hasn't logged on to the forum since this conversation. He might still be watching but not logging on but I rather doubt it. Pity, as it would be nice to know the ending.

Had a brief exchange of email the other week with somebody (British, I assume from his name) in Spain wanting to do something similar and having read my blog post referenced above. His problem was that the immersion in question had a very small boss (1.25", I think) so the option of multiple immersions actually in the tank wasn't there. I did suggest immersions in a side-arm heat exchanger. I'll hope the hear what he comes up with.

Sounds more like Smith and Weston

Posted By: Ed DaviesThe bit I'm not clear about is that if you plumb in to the feed and outlets of the tank how you make sure that there's enough resistance in the side-arm path to prevent a “short-circuit” when water is drawn off but keep the resistance low enough to ensure the thermosiphon works well enough.

Good question. I suppose that (a) with a PHE, the heat exchanger itself has a significant resistance and (b) the side arm is probably plumbed with a smaller diameter pipe than the main feed and is certainly much longer than the feed length into the top of the tank. So there's much less resistance for water coming from the top of the tank and the vast majority will come from there. Hand wave, hand wave and no numbers, sorry!

Posted By: alistairmthere is a system called EMMA which transfers power to emmersion heater when there is no demand from power outlkets. Not sure if it would work on off grid system but nice piece of kit all the same

Nice kit yes, but have you seen the price of the EMMA (then added in the price of an inverter to convert the DC from the panels to the AC that EMMA works with)?
Even if possible it's unlikely to be an economic solution for a simple case.

I have heard of people using much cheaper immersion diverters with off-grid systems that are AC coupled (e.g. Sunny Island), but that's a long way from "direct connection to an immersion"...

But the point here is just using the PV to directly heat the water as a substitute for solar thermal (but with less maintenance). Cheaply.