Off-Grid system with grid fall back.

[paulca]

Longer term a larger 600-1000kW system primarily to heat water.

I am sorry but I had to laugh at the idea of a megawatt water heater, how many Litres a second is that raised by 40C OMG communal showers for the entire village 

Ooops!

On a more serious note, learn to wind your own inductors in time, ETD cores and bobbins are very simple to use but avoid the horrible BLOCK make with square poles as wire much prefers to be bent around a circular pole. Do you know Farnel, they have a good selection available, also can you use Spice ?

I did use spice last night to run some numbers on a 30V 30A buck converter, seemed to work at 18kHz with a 100uF cap and 500uH inductor.  "worked" in that it had a saw tooth current wave on the inductor, fairly stable low ripple output and the inductor was in continious operation, even at low PWM values.  I cheated and just used a voltage source for the panel having googled how difficult it is to simulate a panel.  The only thing that freaked me out a bit was the voltage source current showed as negative, which I simply couldn't explain, so I parked it.

[GeorgeOfTheJungle]

I didn't expect to see those kind of numbers, but thinking about it, 1000W can heat tap water from cold to 60*C in 7 hours. 

To heat 250 litres = 250e3 grams, at 4.18 Joules per gram per °K, from 20 to 60°C you'd need:

250e3*4.18*(60-20)= 41.8 MJ = 11.6 kWh, which in 7 hours is 11.6/7= 1658 W.

That 7*C loss in 12 hours, is that for 62*C down to 57*C?  That would be the worse case for heat loss and works out about 60W.

250e3*4.18*7/3600/12 = 169 watts

No?

[paulca]

I didn't expect to see those kind of numbers, but thinking about it, 1000W can heat tap water from cold to 60*C in 7 hours. 

To heat 250 litres = 250e3 grams, at 4.18 Joules per gram per °K, from 20 to 60°C you'd need:

250e3*4.18*(60-20)= 41.8 MJ = 11.6 kWh, which in 7 hours is 11.6/7= 1658 W.

That 7*C loss in 12 hours, is that for 62*C down to 57*C?  That would be the worse case for heat loss and works out about 60W.

250e3*4.18*7/3600/12 = 169 watts

No?

Correct.  I did realise I missed the part where I was calculating for my tank which is a stubby one which I believe is 125 litres.  I should have mentioned this when I remembered or edited the post.

Though I think we disagree on the heat loss, so I probably went wrong somewhere.  I think I was using an online calculator as I'm lazy.

[Bratster]

You may be interested in the digital MPPT thermal controller that this guy made and sells.

https://electrodacus.com/

Here's a forum member here, and there is an extensive thread of that product on here somewhere.

Edit: found it:
https://www.eevblog.com/forum/crowd-funded-projects/digital-maximum-power-point-tracking-no-dc-dc-conversion/

Sent from my Pixel 2 XL using Tapatalk

[paulca]

You may be interested in the digital MPPT thermal controller that this guy made and sells.

https://electrodacus.com/

I think I spotted this before, the downside is, it makes significant requirements upon the installation.  This would be ideal in a new build or renovation/rebuild type situation, but I'm not in that situation.

It does no DC/DC conversion but instead using binary digit loads, such as 1, 2, 4, 8 Ohms which the controller switches in and out to get the correct resistance to match the panel power point.

So in my case I would need a hot tank with a bunch of different elements.

The case study presented for the device actually uses multiple circuits of concrete floor heating elements for space heating.

Nice idea, way to expensive to install.

[fourtytwo42]

Are you adverse to making something yourself ? I see you are familiar with Atmega and that has a pwm and adc capability. If you start small you can learn as you go, you already have a ~12V 50W panel just make your own pwm controller and start with that.

I could recommend an IRFB4227 mosfet as being virtually indestructible driven by a TC4427, also an MBR20200CT schottky is likewise almost indestructible those components can be re-used again and again and are good for up to 200V. In your situation you also need a decent film capacitor to absorb the ripple current, just invest in some Panasonic brown jobs (they have a proper published specification unlike so many others) a few 2.2uF/250V will do you. Just use your electric fire as a load and away you go for a simple PWM! As mentioned before it will radiate like mad without an inductor but perfectly ok for experimentation.

Initially the control loop could simply measure panel voltage and increment/decrement duty cycle to try and keep it at the specified (for the panel) MPPV. 25Khz is a good pwm frequency to go for at the beginning, not to much in the way of switching losses (or other parasitics) but high enough to keep component sizes small.

Get yourself a copy of LTspice, it's a really useful tool as is an oscilloscope if you don't have one already.
PS please dont use your SLA as a power source it has no current limit and is unsuitable, stick to PV

[paulca]

Not adverse to making something myself.  I have a fairly useful set of tools including bench power supplies, USB oscilloscope, DMM etc.  I have actually built a buck converter and a dummy load so not brand new to mosfets et.al. either.

One wrinkle is that I just packed all my electronics lab into 2 large plastic storage chests to make a fair dent into my house move work load when the time comes.

The thing that makes me nervous however is running 1kW of power at something like 30-90V through something I made.  However I think both agree running 50W though something I built at 12V should give me some experience.

The new house does have a "token" electric fire in the living room, but it has fancy lighting flame effects which might get upset of I vary the voltage considerably. 

However I do have options, a handful of 50W power resistors which I don't mind sacrificing to water damage or a 3kW water boiler for homebrew which has already been tested to not mind a variable power supply as I use a 40A dimmer circuit on it when boiling beer to prevent pumping 3kW into evaporating beer.  The thermostat actually tops out above 100*C so I can prevent the need of using it's thermostat contacts at all with DC.  Not that a 50W panel will come close to boiling more than a few 100ml of water in it given heat loses.

Funny story... until I bought and used the 40A dimmer on the boiler it literally rained beer in my kitchen.  Literally there were droplets forming on the ceiling and dripping onto my head.  I made a post on here the day the 40A went wrong and started smoking but... getting the beer made was most important so I let it burn and need a new one as it's terminal blocks melted caused by a slightly loosened live wire.

[fourtytwo42]

Another homebrewer! I guess I make about 25 gallons of beer a year and another 30+ gallons of wine/champagne, I use an old variac for simmering AND still lug it around as I still have not made an electronic version!

Sounds like now is the perfect time for you to get into virtual circuits with LTspice, to get you started here is a simple PWM both as a file you can use and a schematic to see. It just steps the pwm through various on times, if that were a real PV rather than a 12V supply with a series ballast you would find MPP
I have models of cells but more complicated.......

[paulca]

Another homebrewer! I guess I make about 25 gallons of beer a year and another 30+ gallons of wine/champagne, I use an old variac for simmering AND still lug it around as I still have not made an electronic version!

EBay!  Mine was sold as something like "250V Mains AC 40A Dimmer PWM Motor Control", I don't believe it was any of those things, but it was recommended from a large popular How To thread on homebrewtalk.  Even they thought it was PWM, but... does PWM even work on AC?  Inside it's just a pot, 2 caps and a TO220 package on a heatsink.  It worked though... until I killed it and I suspect it will still work with a new connector block soldered in.

The volt/amp/power meter also recommended in that thread was reported to come with two different sets of instructions with contradictory wiring.  It's was basically a coin flip, not knowing how to reverse engineer the circuit, I choose wrong, followed the instructions I received with the meter and have the burn holes in my carpet matching the lovely air vent holes I had drilled in the case to prove it.  it did make a nice bang and gave the inside of the box a nice copper/electrolytic finish too.

[fourtytwo42]

I cheated and just used a voltage source for the panel having googled how difficult it is to simulate a panel. 

I am sorry very remiss of me to miss this! Here is a 6x6" cell model, just connect as many as you want in series for a panel, vary the input voltage 0-1V for insolation 0-100%. Change the current multiplier for other cells and you may have to adjust the diode parameters to get a good match around the knee and slope but as given here is a good match to my Axitec 260W panels.

I have also included models for a 260W panel, I hope you can get your head around LTspice hierarchy!
Ooops it doesn't like .asy files so changed extension to .asy.txt

Please ask for anything else that might be useful to you