Renewable Energy Questions/Discussion > Solar (heating or electric)

UK PV water heater DIY MPPT booster

(1/3) > >>

fourtytwo:
Hi All this is my 2nd year of PV production having dipped my toe in the water last year with a 520W system to trial the performance in my location, it has been so successful (e.g. no need to run an oil boiler in the summer) I have added two more panels to increase the power to 1040W. I must confess to not hand building the panels though I did investigate thoroughly but got a good deal on some panels at about 40pence/watt. Part of the reason for using PV is the startup cost to me as an electronics engineer is far lower than an equivalent solar/water heating system and I accept the consequential loss of efficiency.

The interesting part is converting the maximum solar energy into heating the water, as everybody knows to extract the maximum output from the panels they must be run at there maximum power point. I had already decided to use the existing electric water heater often fitted in our house water cylinders here in the UK as a backup in case of boiler failure (a long story but I have what some may say is an old fashioned indirect water heating system) this being a 240Vrms 3Kw element with a cold resistance of appx 19 Ohms.

After quite a bit of spreadsheeting with various panel and converter schemes the power loss of a simple boost converter outweighed the complexity of a full monty sepix converter (allowing both boost & buck), basically the boost converter only operates when it is possible to extract more power from the panels than the direct connection of the heater alone, this occurs at above about 50% insolation for these 60 cell panels with a 19R load.

A serious problem with boost converters is input ripple current, if this is to high a percentage of the nominal current the PV efficiency suffers accordingly, there are several solutions, lots of expensive capacitors, multiple converter phases or high frequency operation, I chose the latter but constrained by breadboard construction, switching losses and the chosen micro-controller (PIC) pwm frequency range and is in fact 125Khz.

The inductor is constructed using hand made Litz wire on an ETD44 core, the mosfet and schottky diode are both good for 200V output (2.1kW), the only capacitors required at this frequency being film rather than electrolytic. One word of warning about film capacitors they can have a surprisingly large ESR and it is often unspecified!

Cooling was originally by conduction to the steel enclosure but at continuous high powers the rather thin aluminum U channel used to conduct the heat was inadequate and caused thermal shutdowns. In redesigning the system for 1kW this has been replaced with a heavy extrusion force air cooled (when required).

Another issue with PV is generating the operating voltage for the control system in an efficient manner with a low start up voltage, originally this was done using a discrete linear regulator (not many chips for 80V) but with the advent of the fan this became a semi-discreet switching regulator that handles inputs as high as 150V (1kW system).

Apart from it's other duties the PIC processor sends out the power generated periodically and this is used by an indoor display/logger unit that shows present production and averages for the last week, month and year the data base can also be dumped to a PC.

I should mention also EMI was a nuisance when trying to use a scope so the mosfet was slowed down and more power dissipated in the snubber to combat it. Also this is not the end of the project and a complimentary GTI is presently under construction to use the surplus power once the water is hot.

Ohh I almost forgot to mention the thermostat! without this the tank would boil but switching 1kW at high voltage DC will destroy mechanical contacts so a mosfet is used controlled by a tank sensor, this raises the cut-in insolation slightly as the mosfet is in turn protected from low gate voltages. Note the converter keeps running to power the forthcoming GTI after the tank reaches it's setpoint.

I have attached some pictures often with weird file names to illustrate the parts of the system.
I would like to thank this website for its inspiration and those who helped me procure some of the more esoteric parts :)

Pete:
Hello 42, looking at your photos it appears that the tank temperature is not getting very high. Do you boost the tank with mains power at night to bring it up above 60 degrees. The reason I ask is that if the water is stored below 60 degrees for too long it can breed Legionaires bacteria.
Looks good though, pretty comprehensive data logging setup.
Cheerio
Pete

fourtytwo:
Hi Pete many thanks for the tip! I guess that pic is quite old and when it was just a 2 panel system, these days I have the opposite problem, to much heat! Seriously though if I have a bad day and the temps are a bit low I set the oil boiler to come on the following morning and boost it, this very rarely happens now.
What a wonderful place you live :)
Roger

MadScientist267:
Fourtytwo - Welcome to the forum :)

Looks like a fun project... A couple of things came to mind reading the original post... With the frequency you're running at, and schottky diodes, I'm imagining that's where a fair bit of loss is coming in...?

Wondering if you've considered sync rectification for a further mod to this version or possibly as part of a down the road revision? A little more involved to bring together but would generate a lot less heat at those speeds.

Either way, looks pretty good. Looking forward to hearing more on it as it develops.

Steve

fourtytwo:
Hi Steve, yup fun is always 1st priority :)

Well being a boost converter the diode is high side and operating at nearly 200V above ground so a mosfet would be a pain to drive up there! As for efficiency the simulated loss of the diode is around 5W at over 1Kw output (0.5%) so IMOP not worth the cost/complexity.
I am loosing more in the mosfet/snubber deliberately to reduce EMI, I know a much better way, construct it on a pcb to reduce the switching node loop length but once again cost comes in to play as a pcb of that size would be ~£50/$65USD.

Like all designs it's a compromise :) fortunately I had a big block of aluminum kicking around in the junk-box and that provides so much thermal lag that with typical British weather (sunny periods) the fan doesn't even come on, but it may do in July/August (hope) :)

The GTI portion is much harder, apart from having problems compensating the output LCL filter I think I have blown yet another unfolder mosfet and when integrated with the booster exposed a latent problem in the boosters aux psu that it shares causing the boosters thermostat mosfet to blow! All in a day's work and everything is at least 10 times more complicated than you first thought hahaha

Many thanks for your comments and I hope to keep posting progress all be it slow :)
Roger

Navigation

[0] Message Index

[#] Next page