Yes, if you can compare two identical sets of panels, one into an MPPT and the other into a fixed resistance that should give a good illustration.

Since my previous post I've been wondering what I have as a resistor which I could put on my little toy amorphous panels to do a similar experiment.

Excellent, thanks Bxman.

I was going to connect my little “60 W” [¹] Maplin panels to a 2 kW heater this morning. Wiring the four panels in series/parallel (2S2P [²]) and switching the heater between the 750 W element, the 1250 W element and both together would allow not-terrible matches over a reasonable range of brightnesses. I had a junction box set up for 2S2P to give nominal 24 V to feed my MPPT controller with a 12 V battery but I can't find it now - got all the other bits but it must be in some mystery box in my spare bedroom. Your confirmation saves me bodging it.

Posted By: bxmanBut still clearly very inefficient except when in full sun […] But do they make 60 or 110 v elements that I could use ?

Are you sure that's what you want? If they're approximately matched in full sun then for lower sun you'd want more resistance (same voltage, less current) but 60 or 110 V elements of the same power would have higher current and therefore lower resistance.

Presumably your heater presently has the elements in parallel. Maybe arranging them in series would be more appropriate.

[¹] Actually about 45 W, I think.

[²] Or 2P2S if you follow the opposite convention. Where's the headslap emoticon when you need it?

I find the following a great help to work out DC variables:

Yes, you're right. But he has three 70 Ω ohm elements in parallel so can absorb 3×822 = 2466 W at most.

n anyone advise or help with tips on preserving format thanks


Keith can you help I seem to have lost what I had half an hour ago

thanks Patrick

@johnmuready
you clearly saw it before it disappeared I hope Keith can find it again If not I will have to have another go this evening
cheers Patrick

Yep, shame it's not easy to do tables in this software AFAIK. Trying HTML table elements doesn't seem to work, either (cjard will, no doubt, take that as a challenge ).

Still, I'm very confused by your “earlier on” readings:

1 element: 1 A at 215 V
2 elements: 1 A at 42 V
3 elements: 1 A at 24 V

Are you sure that's the right way round? I'd expect the voltage to be higher with more elements than with fewer. Or is that the voltage across a single element? In which case, I think it's inconsistent with the full sun readings.

Thanks, so:

number of elements, bright-sun power, dim-young-sun power:

3P | 1007W | -
1S |  840W | 24W
2S |  503W | 42W
3S |  399W | 215W

(PS, ha! using HTML <tt> and </tt> tags (“teletype”) gives a fixed-pitch font which at least makes tables a bit more reproducible. Also <pre> and </pre> tags preserve whitespace (rather than packing runs of two or more spaces down to one) which also helps. Though it does add some vertical whitespace underneath the table for some reason - I seem to remember trying this before and the vertical whitespace causing some sort of mayhem.)

Posted By: bxmanI am almost sure the higher resistance gave the best yield in the lower light levels.

It did: 215 watts vs 24 watts for the low resistance.

That's what you'd expect. To get maximum power you want to operate at the maximum-power-point voltage. For your panels the Vmp for each panel is a tad over 30 volts in full sun, a tad under in dimmer light. So for the whole lot you want to be operating at around 240 volts for best power. In low sunlight you'll only get a low current out so need a high resistance to make use of all that voltage.

(I don't understand how it could be 24 volts and 1 amp simultaneously into a single 70 Ω resistor. My assumption is that it wasn't actually simultaneous - that you have had to swap the meter round or something - and the sun changed a bit in the mean time.)

No, <tt> doesn't act as a tab - it changes the font. The start tag (<tt>) goes before the table and the end tag (</tt>) goes after. As input the whole table looked like this:

number of elements, bright-sun power, dim-young-sun power:
<tt><pre>3P | 1007W | -
1S |  840W | 24W
2S |  503W | 42W
3S |  399W | 215W</pre></tt>

If there's any reasonable amount of sun the open-circuit voltage will be pretty much the same. The short-circuit current is a better indication of the light level.

Your power outputs this week weren't a lot different from each other (165, 190, 175.5 watts) but still the 2S case seemed best compared with the 1S and 3S in dim and bright sun last week so presumably an intermediate amount of sun.

bxman - the notes below follow my thoughts and parts used. I should be live testing in the next two weeks:

The configuration of my 4 panels to produce the voltage / current combination was worked back from the immersion. It was very expensive to have a bespoke DC multi coil immersion built. I went for a 3 coil at 36 volt / 400 w per coil, approx £30 from Amazon. My panels are in parallel, 30 volts @ 32 amps. This also fits the cheaper Solid State Relays, DC @ 40 amp.

Max power transfer will be achieved by using the 3 coils in 6 configurations to provide the best path to max power transfer using a switching matrix of 5 solid stare relays and two diode to switch between the coil configurations. The Solid State Relays are connected to the raspberry Pi through an 8 channel optical isolated relay board.

Using the PV voltage to establish the max voltage point per coil, switch configurations and tank temperatures is achieved by using an analogue / digital converter on the Pi supplied by A B Electronics. As the Pi has no real time clock a clock board was supplied by A B Electronics and spare memory on that board to store kWh.

Tank over heat protection is provided by a mechanical sensor on the tank to cut off low voltage power to switching solid state relays.

To overcome tank stratification an external low voltage circulating pump to be fitted, required as my DC immersion is located in the top 1/3 of the tank.

System screen display ( using Nextion screen connected to Pi) showing voltage, amp, matrix configuration and total kWh to date diverted, date and time etc.

The Pi, Screen, Analogue to Digital and Real Time Clock programmed using Python.

Any chance of a a wiring diagram and the code you have written. May be helpful to others that want to try the same thing.

Looking at the data from the running logs I now think it would lead to very little power loss in my case to reduce the number of coil configuration from the current 6. Perhaps another exercise to crunch the data one day and find out.

Less coil configurations could play apart if the current Solid Sate Relays and mini optical isolation relays were replaced with mosfet type switching from a direct connection to the Pi GPIO

Wondered if anyone had come up with an idiot proof solution for this?

Started a similar thread on BuildHub listing a few devices that supposedly do this job - https://forum.buildhub.org.uk/topic/14863-pv-to-thermal-store/