Having set up the solar powered electrical system for my house and seen how successful it was, the next step for less dependence on outside services controlled by big business, was to have my water heated by solar.

Learning about water heaters.
Late 1996, when my 15 gallon Zip water heater tank had corroded to the point of brazing up the pinholes being a futile exercise, I was forced to replace it. This is when I learned that most water heaters for sale now are essentially junk. My original tank was made of cusilman bronze and it had lasted 30 odd years with full mains pressure applied to it. I'd also lived with copper and copper lined tanks which were also reliable. Much to my surprise neither cusilman bronze, or my other choice, copper lined steel (eg. Rheem Coppermatic) heaters were still being made. Something about water quality the manufacturers told me...but they could offer me any number of models with glass lined tanks. No way would I spend money on something which was actually destined to fail from the first day it held water. For those unfamiliar with the concept, glass lined tanks are a mild steel (just saying that makes one think of rust) with vitreous enamel (fancy name for molten glass) sprayed onto the inside surface, supposedly protecting it from corrosion. Which it would, except that the steel cylinder expands and contracts as it heats and cools. As glass is not flexible, minute cracks appear and the tank is now doomed. To slow down the rate of corrosion, a sacrificial anode is inserted into the tank. It's made of magnesium and is meant to corrode instead of the steel. However, most owners of glass lined tanks never check the anodes, or replace them once corroded away. Also, the corroding magnesium ends up in the tank and hence the hot water that issues forth. Such tanks often have words like "Vitreous", "Enamel", "Glass" or that abominable misspelling by Rheem, "Glas", in their model names. The inside of them, if you haven't ever seen a cutaway model, is just like an oven baking dish. If you're unfortunate enough to be stuck with this kind of heater, run it at the lowest temperature and water supply pressure you can, and check the sacrificial anode. Edwards and Beasley seem to be the predominate manufacturers of stainless steel water heaters, so in the meantime it's worth familiarising yourself with their models, ready for when the time comes.
Most water heaters in Australia are of the mains pressure type, hence the requirement for a strong tank. To allow for pressure build up, a TPR (Temperature and Pressure Relief) valve is fitted. This opens thermostatically at 99 degrees (in case the heating element doesn't shut off) so that steam doesn't build up causing an explosion. In addition, should the thermal part of the valve be not functioning, the valve will open with extreme pressure alone (chosen to be above the pressure of the water supply).
Once popular until the 1960's when the Rheem Coppermatics came into vogue, were gravity fed copper tanks, usually installed inside the roof space. These have an exceptionally good life, and fed from an open cistern, do not require all the pressure release paraphernalia of mains pressure heaters.
As I didn't have the roof space to accommodate such a heater, I stayed with the mains pressure option.

The Edwards Stainless Steel Water Heater.
Further into my enquiries, I soon discovered the stainless steel water heater. Problem solved. It does surprise me that so few water heaters sold are of the stainless steel variety. Perhaps the extra cost puts people off and they think five or ten years life is satisfactory.
The model I chose was an Edwards Energy Systems DES125 which is a 125 litre mains pressure model. It is made from 316 stainless steel and had a ten year warranty. I installed the lowest pressure reducing valve I could get (350kpa) to further increase the life. Interestingly, the Edwards water heaters come with a cold water expansion valve. This is a reflection of their Western Australian origins. In that State and also South Australia, this is apparently a compulsory requirement. In those areas the water supplies are prone to cause scale build up on the TPR valves. This can cause them not to reseat properly after being activated, and thus the familiar failure with constantly dripping hot water. The ingenious way out of this is to have an expansion valve on the cold water inlet of the tank. Because the cold water does not cause scale build up, the valve is reliable. The cold water expansion valve therefore has a lower pressure limit than the TPR valve, so in effect the TPR valve is only activated if the water approaches boiling point. The additional advantage of this scheme is that only cold water is lost during the heating of the tank, not hot water as is the case when there's only a TPR valve. Thus the heating energy is lessened. Unfortunately, in recent years, Edwards was taken over by Rheem. This has lead the existence of "Edwards" labels on some glass lined tanks.

Here resides the Edwards DES125 stainless steel water heater.

Solahart.
I did some research into solar water heaters around 1989 when I'd wanted to install one at our weekend house. At the time I was keen on the 180 litre Solahart. They had developed this "black chrome miracle" collector panel which supposedly transferred more of the sun's heat to the water. They also had developed a novel heat exchange system for use in frost prone areas. In such locations, the panels can be damaged overnight in sub zero temperatures. The water in the panels freezes, expands, and thus ruptures the pipes in the collector. Ways to solve this problem are to use frost valves which open in the cold, allowing hot water from the tank to flow through the panel. Apparently such valves are unreliable as well as wasting hot water. Another method is to circulate a small amount of water through the panels by means of a pump. While there is no cold water loss, some of the heater water stored in the tank is wasted warming the panels. The Solahart approach was to heat the tank by means of antifreeze fluid which was in turn heated by the solar panels. This neatly solves the problem with no heat or water wastage. However, the heating efficiency is less due to losses in the heat exchange process, and given their tanks are glass lined, I would think the glass would function as an insulating layer.
When it came to replacing my water heater I could have used the Edwards version with stainless steel tank, but at that time cost really was an issue so I took the easy way out and stayed electric only. Electricty charges in 1996 were also less than half at the present time (2013), so running cost was not an issue.

Evacuated Tube Collectors.
Towards the end of 2007 I'd become curious again about solar water heating. I'd now had the solar electric supply operating for two years, as well as the wind generator. A large proportion of the house water supply was coming from a 1000 gallon rain water tank, so solar water heating was next on the agenda.
The people I'd spoken to at Edwards 11 years ago told me my tank could be converted to solar and this had always stuck in my mind...so onto the internet to see what I could find. My initial idea was just to get a conventional flat collector panel and connect to my existing tank. I quickly discovered that there was a new kind of collector with many advantages. It was called the Evacuated Tube Collector. Not only was it very light in weight, but very easy to install and frost protection was not an issue. In my area, frost protection would be necessary with a conventional system. The thought of antifreeze fluid leaking and running down the roof into my tank was also another reason to avoid the well known commercially made systems.
The collector consists of an array of glass tubes. Inside the glass tubes are copper tubes filled with liquid. These copper tubes are heated by the sun (they have black fins attached), and the hot liquid rises to the top of the tube which is in contact with a heat exchanger. Through this heat exchanger flows the water to be heated. The key to this working so well is that the glass tubes are constructed like a thermos flask. There is a vacuum between the inner and outer tube which means that heat trapped inside cannot easily get out. Also being round, it means the angle of the sun is not critical and the tubes receive radiation over a much greater period of the day than would a flat panel. As there is no water flowing through the actual collectors, they are immune to frost damage. The whole array weighs only about 60 kg, with the weight being only that of the glass tubes, heat exchanger manifold, and stand.

Retrofitting to an existing tank.
A typical vertical water heater tank has a cold water inlet right at the bottom and the hot water outlet right at the top. A third connection at the top is for the TPR valve, but some heaters combine this with the hot outlet. The heating element is located in the lower quarter of the tank and thus heats the water above it. A tank made for solar use has an extra connection about halfway up. This is where the warm water from the solar panel is fed into the tank. The cold water for the panel comes from the bottom of the tank. Thus when the panel is heated, thermosiphon occurs. How to use an existing tank? The cold supply for the panel is easy as there's already a connection at the bottom of the tank; i.e. the cold water supply inlet. The hot output from the panel is the tricky part. If it was merely fed into the top connection of the tank, there could be conflict with the hot water return from the solar collector; eg. if the water from the collector is cooler it will mix with the hot water from the tank and thus reduce the outgoing water temperature. I eventually discovered the trick is to use something called a "five way valve", or "solar conversion valve".

This diagram came from the eBay seller of my kit and illustrates the five way valve.

What happens is that the cold water connection at the bottom of the tank performs two functions. The cold water is fed into here as well as to the solar panel. However, the hot water is also fed via this connection to a length of tubing inserted into the tank, the end of which is away from the cold inlet. So, cold water is drawn from the bottom of the tank, pumped through the solar collector, heated, and then injected into the bottom of the tank (away from the cold area) and rises to the top. Thus, no modifications to the tank are required. Obviously, the tube injecting the hot water must be of smaller diameter than the cold water inlet. Typically, the cold water inlet is 3/4" which allows a 1/2" injection tube. The remaining 1/4" gap is quite sufficient for cold water flow.

Five way valve made from ordinary plumbing fittings. The length of copper tube protruding into the tank is somewhat longer in my installation.

This type of conversion applies to a ground standing tank with solar panels on the roof. Unfortunately, for thermosiphon to occur, the tank needs to be above the collectors as hot water naturally rises. With this type of installation a circulating pump is therefore required, so unless this is powered from a photo voltaic panel the hot water is not entirely free. However, the pump is only for circulation instead of having to push a head of water so power requirements aren't that huge. My Wilo pump supposedly draws 46W. Somewhat cheaper to run than a 3.6KW element! The controller means the pump only runs when the water in the solar collector is warmer than that at the bottom of the tank, so it isn't running all the time. In practice, the pump cycles every few minutes on a sunny day; the stronger the sun, the longer the pump stays on. The Wilo pump was designed for central heating systems so it's overkill for this application. It has three speeds, which are actually selected by introducing different amounts of capacitance in series with the motor.  As it is, the pump does not remain on continuously during hot sunny days, which indicates the speed is not too slow.

The Wilo circulating pump. It has three speeds, the lowest of which is used for this application. It is important to make sure the proper unions are supplied when purchasing one of these. Do not use the cast iron body version (intended for circulating anti freeze solution) as it will rust, causing impeded flow, as well causing rust particles to interfere with the one way valve.
Use the bronze version instead.

What I did find very useful was to look at installation instructions for commercially made water heaters, in particular, the Rheem Loline which is a solar retrofit. 
Evacuated tube collector installed on my roof. Note the photovoltaic panels in the foreground.
 

The Installation.
First thing was to assemble the stand and mount it on the roof. The stand is stainless steel and the roof is zincalume. To those outside Australia, this is a zinc/aluminium material with similar appearance to galvanised steel and is used for the same applications. Given the corrosion possibilities with dissimilar metals in contact with each other, I mounted the stand on neoprene pads before screwing it to the roof. I simply obtained a sheet of neoprene from Clark Rubber and cut it to suit.
It was rather convenient that the roof area where the solar collector was to go, was right above the laundry where the hot water tank is. Thus there would be a short vertical run of pipe. There is approximately 10m of pipe in the circuit. To start with, I used green lagged copper tubing for the pipe run. All the fittings were designed to take 1/2" tubing. The short length of copper pipe protruding from the heat exchanger manifold is a non standard size in Australia. Compression joiners to connect this to 1/2" tubing were supplied.
Getting the pipe through the ceiling space was the most tedious part as there was a piece of framework that had to be cut away above the ceiling cornice. Once this was done, the pipes were installed and the roof sealed with a Rooftite ( a neoprene gland attached to the roof with stainless steel screws and sealed with silicone).
I was a bit concerned about mounting the pump due to its weight and the fact that it was meant to be supported only by the connecting pipes. The pump was not supplied with the correct unions either. Instead were just simple 1/2" to 1" reducers with home made silicon washers to seal the whole thing. I had to purchase the correct unions elsewhere. I decided to mount the pump on a board with pipe clamps.
The plumbing was straightforward until I attempted to make the five way valve. The way the Edwards tank is constructed does not allow such a thing to work; not on the cold inlet anyway.

Pump installation and cold water plumbing. This was the original cast iron pump.

Instead of the cold inlet simply being a port in the side of the tank, it's actually a connection to an internal pipe that runs underneath the tank to the centre. How did I find this out?
I tried making the five way valve and inserting the hot water copper tube but found it didn't go very far.
The design of this tank has the hot water outlet right in the centre of the top of the tank; not at the side. So, I could actually shine a torch down and see inside the tank to learn of its internal construction. Nice to see all that gleaming stainless steel inside with not a hint of corrosion. How many glass lined tanks would be like that after 11 years?
It was this top connection that solved my problem, for this was the logical place to put the five way valve. There's about 1.2m of copper tube going straight down into the centre of the tank for the hot water to be injected. The hot water is thus fed into the tank at the same location as if I'd put the five way valve at the cold inlet.

The diecast box at the bottom contains the relay for switching the electric element and socket for the pump. There is an additional switch that allows the element to be permanently turned off when required. Neon pilot lamps show pump and heating element operation.

Once I'd finished the plumbing, turned on the water, and got the leaks out of the connections, it was time to install the glass tubes. The instructions recommend talcum powder to lubricate the neoprene glands in the heat exchanger, into which the tubes are inserted. I didn't have this so I tried CRC2-26. This worked perfectly and the tubes slid into position very easily. Next, the hose clamps at the base of the stand were tightened to secure the tubes. Finally, the stainless steel reflectors were attached between the tubes.
At this point I hadn't installed the control unit so the pump wasn't running. By the time I'd finished inserting the 24 tubes I found the hot outlet on the heat exchanger too hot to touch.

Not my unit, but the Endless Solar version, this shows how the tubes are inserted into the heat exchanger.

Within a few seconds of turning on the pump I had to let go of the hot water pipe from the exchanger going into the tank, as there was a burst of very hot water flowing through it.
It was at this point that I finally believed these tubes might actually work.
One annoying thing during the installation was leaking screwed fittings. I used compression fittings or soldered Yorkshire fittings where possible. Even with copious amounts of teflon tape, the screwed connections always leak. I've done them up to the point where the threads are about to be stripped, and yet I can almost count on seeing a drip when I turn the water on. So annoying did this get I just  ran solder into the threads. Problem fixed! I only did this where serviceability isn't goint to be affected of course. The real problem is that many fittings don't have a tapered thread and this is what's really required to prevent the problem.
The final part of the installation was to install the controller. To this are connected three thermistors. One (designated T1) is inserted into the heat exchanger near the hot outlet and the other is located at the bottom of the tank (designated T2). The third thermistor (T3) is mounted at the hot outlet of the tank. Its purpose is only to activate the electric element if the water has not come up to the required temperature by solar alone.

T2 thermistor held against tank body by a piece of tinplate secured under the terminal block.

The default setting of the controller is such that the pump turns on when the collector temperature is eight degrees warmer than the cold supply to the collector at the bottom of the tank. The pump then runs until the temperature difference is four degrees.
Two other functions which are not really necessary in my installation are frost and over temperature protection. Should the T1 sensor detect imminent frost (default setting is two degrees), the pump will circulate warm water until four degrees is reached. This prevents the pipes freezing and bursting. This would be more applicable to flat plate collectors, however. The other feature is to turn the pump off should T2 reach 70 degrees. This stops the hot water flow under extreme temperature conditions.

Diagram supplied for my solar heating setup. In addition I have made use of the T3 temperature sensor input and the H1 heating control for automatically switching the electric element. Unfortunately there's an important omission: There should be a non return valve either on the pump outlet or solar collector return to tank. Without this, reverse thermosyphon occurs at night. The pump wiring via tank thermostat should be ignored.

The controller has a a feature where it can switch on the element up to three times a day for a specified time and until a preset temperature is reached. This temperature is sensed by a third thermistor at the top of the tank (T3), where the water is at its hottest. I simply pushed the thermistor in between the foam insulation and the brass T fitting at the hot water outlet. While it isn't in direct contact with the water, it works well enough as the brass is a good heat conductor. In practice, I found the thermistor shows about three degrees less than the water temperature. So, to have the element switch off at 50 degrees, I set the controller for 47 degrees. I have set the controller so that the heating element is switched on (if it needs to be) at 17:30 for an hour, which is just long enough to get the water temperature up to 50 degrees, from about 36 degrees, which is typical for a cloudy day. This works in with my hot water requirements being mainly at night, which is the best way to use a solar hot water heater.
The controller supposedly can switch 16A for the heating element, but I find this hard to believe given it's all on a printed circuit board and the relay doesn't look very substantial.
I decided to make a second box with a contactor inside, as well as some neon pilot lamps to indicate when the electric heating was activated, and when the pump was running.
The heating relay inside the controller simply supplied the low current of the contactor coil which then switched the 15A heating element. Unfortunately, the contactor having an AC coil buzzes loudly at 50 cycles. As the whole thing is mounted on a hollow fibro wall the sound is amplified to a large degree and is very unpleasant. I fixed that by using a 12VDC relay with 250V 30A contacts. The relay coil is fed from a small transformer and rectifier. Being DC it is totally silent in operation. Because I don't always need the water heated automatically, I put an extra switch in series with the transformer primary.

Electric heating.
It needs to be pointed out that a large proportion of Australian electric water heaters normally run from an off peak supply. That is, they are metered separately at a lower rate, but the supply is only present overnight. This does not suit a solar heater.
Take the scenario where it's been a cloudy day and the water hasn't heated up for evening use. Firstly, the off peak supply does not become available until late, and then it takes an hour or so for the water to actually warm up. Having then heated the water up to full temperature by morning, imagine what happens if it's a sunny day.
Because the water is already hot, the solar contribution will be mostly wasted. And the cycle continues. There will be little saving.
Ideally, the water needs to be heated, if necessary, prior to the evening/morning use, and only just enough. This way, the electrically heated hot water is used in the evening and morning only, and cooled down by mid morning when it's ready to absorb solar heat.
For these conditions, I've found switching on the electric heating, if needed, is best done late afternoon, once solar contribution stops. 5.30pm is my chosen time. From experience, I find that only one hour of heating is needed to cover the night and morning usage. So, the electric heat switches off at 6.30pm.
For this reason, the ideal pattern of hot water usage is that most of it occurs in the evening, when the water is hottest, leaving cooler water for the morning, ready for solar heating.
Furthermore, most people run their hot water tanks at absurdly high temperatures. The default setting is often 70 degrees! 60 degrees is more than sufficient.

I am very sceptical of claims that certain solar powered devices will work in cloudy conditions, and evacuated tube collectors are no exception. The answer is simply that they will provide some heat, but the truth is not to a usable degree. The electric element is usually required after two days of cloudy weather. Having said that, the use of solar collectors on a cloudy day will still save electric power because the element is only heating the tank from say 35 to 50 degrees, rather than starting off at say 18 degrees. It takes less energy to obtain a 15 degree rise in temperature than 32 degrees would require.

What the kit supplier didn't tell me.
Once into Autumn, something started to become very evident. My suspicions of reverse thermosyphon at night was confirmed as the pipes up to the collector remained warm even though the pump was not operating.  With cool night time temperatures outside, most of the heat accumulated during the day was simply dissipating into the night sky. I was beginning to think a non return valve was needed, and as I looked at the installation manuals for the commercially made systems, it was now quite obvious; they all used one. To put it simply, I was rather annoyed that not only my kit supplier did not provide or mention the need for such a valve, none of the other eBay sellers mention it. I eventually decided on the RMC solar non return valve type SNR502. An ordinary non return valve should not be used as it may not be intended for the high temperatures in a solar hot water system. Luckily I'd seen the RMC price list for this valve before buying one. It was around $57 retail. So, when I rang up Reece to order one and was quoted $147, I looked elsewhere. Tradelink in Penrith had one ordered for me in a couple of days, for the correct price.
Having installed the non return valve, the change in performance is like the difference between night and day. The water is still hot in the morning! The pipes running up the collector are cold at night as they should be, and now I can look at the T1 temperature reading to see what the outside night time temperature is.
Quite obviously, eBay sellers of evacuated tube retrofit kits do not understand the finer points of solar hot water. They are merely importers from China. You are on your own if you buy one of these kits; do not rely on what they tell you. Check that you have been provided with adaptors for connecting Australian pipe sizes to the heat exchanger manifold. If they're not supplied don't buy it unless you have good lathe skills. You're unlikely to be able to get the non standard fittings here. Having said that, I've learnt alot from this exercise, and hopefully my notes here will enable your system to be successful too.
A further 'non mention' was that thermal paste should be applied to the ends of the tubes where they insert into the manifold. This not only seemed logical, but again the commercial instructions specified it. I used ordinary heatsink cream as used with electronics. Again, an improvement was evident after this was done.

Incidentally, you won't be eligible for government rebates with this kind of DIY conversion. For one thing, the eBay kits are not goverment approved, and the conversion would have to be done by a licensed plumber.

Cheaper power bill.
The first power bill since installing the solar conversion was interesting. From $223  dropped to $161. This is a quarterly account and I should point out that the billing period in question commenced about a month before installation. So in actual fact the saving might be more. In any case, that is a very pleasing result. Even through winter, I've seldom used the electric element. Provided the day is sunny, the water heats. Come to think of it, I've never used it after a sunny day.
It is hard to believe that in Australia with its climate that there are so few solar water heaters. They clearly do work and save a substantial amount of electric power. I've had the electric element switched on only about five times in about two months. Yet, people continue to replace their failed (glass lined, of course) off peak or continuous hot water systems with the same.

Performance.
While this installation has worked well for me, I need to point out that it is supplying hot water needs for one or two users.  It is realised that the temperature is variable, and not to expect huge volumes of burning hot water after an overcast day. On the few days where the water is quite hot enough for a shower, but not enough to wash dishes properly, a jug is boiled for that purpose (porcelain type with exposed element, of course). The 1000W jug element run for a few minutes is cheaper than turning on the 3.6kW electric boost for an hour.
While the 125 litre tank works for one or two users who are careful with hot water use, a larger tank is required for more users if the electric boost usage is to be minimised.
The thing to remember is the tank capacity is what ultimately limits the amount of hot water one can consume, not so much the number of tubes. It is pointless to have 36 tubes to heat up a little tank by mid morning with the rest of the day's heat having nowhere to go. You will still only get that tank's capacity in hot water, and once it has gone you have to wait until the next day for more.  A larger tank will however take in that heat and provide a reserve for a couple of cloudy days.
I think 24 is the minimum number of tubes that is practical for my part of the world, and that's with what I consider the smallest practical tank. My rule of thumb would be to use 30 tubes for a 180-250 litre tank and 36 for a 315 litre tank.
It is possible to simply daisy chain the evacuated tube collectors if the heat output from one is inadequate (e.g. use two 18 tube sets to get 36 tubes).
However, care would have to be taken to see that excessive temperatures do not occur from using too many tubes. It is simple matter to cover some of the tubes in summer if this occurs.
It is important to realise that high temperatures are possible with a solar heated hot water supply and while one day the water feels lukewarm, after a hot sunny day it can burn you. Remember, the thermostat controlling the electric element has no control when the water is heated by other means. So, don't just blindly turn on the hot tap and put your hands under it expecting it to be the same temperature it was the day before. If a licensed plumber does the installation for you, a thermostatic mixing valve will be fitted to provide a 50 degree limit, as this is required for all new hot water work.
It is quite helpful to have the temperature display where it will be seen. Not only does it give an idea of what to expect when I turn on a hot tap, but also if I need to think about letting the electric boost operate before the water gets too cold.

Later problems.
After a bit over two years operation, the system was starting to work poorly. First suspicion was the non return valve, and sure enough this proved to be faulty. I dismantled it and found a lot of dirt inside keeping it open. Even after I cleaned it, it still did not seal perfectly. So, again all the daytime heat was being radiated into the night sky. At the same time, I noticed the pump was almost blocked with its own sediment. Fairly obviously the cast iron body was a poor choice. Another thing the supplier got wrong.
I contacted RMC about the valve and much to my surprise sent me a new one for free. Apparently there was a known fault with some of the valves around the time I bought the original.
It was great to get such service from an Australian company. With the new valve in and the pump cleaned out, all was good again.
However, reading operating instructions for commercial systems got me intrigued to experiment with the pump switching temperatures. The default for my controller is if the collector gets to 8 degrees more than the tank, the pump switches on. When the temperature difference drops to 4 degrees, it switches off. It seemed to me that if the pump "on" temperature was higher, then the water in the collector would be hotter before it was pumped into the tank. After some experimentation, I settled on 10 degrees. As for the "off" temperature, it seemed that it should stop pumping once all the heated water had entered the tank. It was quite obvious when this happened simply by feeling the copper tube at this point.
I set the temperature difference so that the pump switched off just when the water cooled again. This turned out to be 7 degrees.

Broken Tube.
Around February 2012, I noticed a crack at the bottom of one of the evacuated tubes. This would have been from a falling branch. It took several months for the air to leak in; this being evident from the tube changing to a whitish colour and becoming more transparent. At that time of year, the loss of performance was not noticeable, but with winter on the way I decided to obtain a replacement. By now it seemed that 58mm tubes were now the standard, as few suppliers had the 47mm tubes my system used. Luckily I found someone with a few old tubes left over and bought a couple of spares.

New Pump.
Another 3 years later, and again it was back to reverse thermosyphoning at night. This time I decided to do several things. As before, the pump was clogged by its own rust, so I decided to finally get a bronze bodied one. And again, the one way valve had failed. The  supplier where I got the new pump from happened to sell a different version of these, so decided to give it a try.
If it fails again, I will try a passive method of preventing reverse thermosyphon. It's a method I've seen with one commercial system. Briefly, the incoming hot water pipe from the collector is formed into a U shaped loop, the bottom of which is below the connection at the tank. The idea is that if this part of the pipe is left not insulated, it will be cooler than the water at the connection. And heat does not flow downwards. In theory this will prevent reverse flow. Another option I have considered is a solenoid valve connected to the pump supply. However, such a valve has to be able to work at the high temperature and low flow pressure in the system.
The third modification was to install a flow meter, as the rate at which water flows through the collector has an effect on performance. Also, being able to see the flow rate allows one to see if any blockages are forming.

New bronze bodied pump. One way valve is immediately above the outlet connection, followed by the flow meter.
Small screw at top of flow meter controls flow rate.

The flow meter consists of a perforated spring loaded disc. The water pushing against it opposes the spring force, and thus the disc moves. Alongside is a calibrated scale to show litres per minute. It appears the standard recommendation is a flow rate of about 100ml/min per tube. A screwdriver set ball valve sets the rate of flow.
Since installing the new pump, and now being able to adjust the flow rate, I have gone back to the default of 8 degrees on and 4 degrees off, and have set the flow rate to 1.5 litres/minute. This has made a huge improvement in performance.
It seems that my experiment of increasing the temperature before the pump operated was flawed. This is likely to be because there's a lot more time in between pump cycles which not only gives time for the water in the pipe between the collector and the tank to cool, but also while waiting for the pump to run, the solar energy isn't actually going into the tank.
So, to take in maximum energy, it appears best to have the pump run as much as possible while keeping the collector temperature just above the tank temperature. This is effectively what happens with a thermosyphon system. This means having the pump switch off temperature not much above the tank temperature, and setting the flow rate slow enough so the pump is almost running continuously on a sunny day at the peak of summer.

Conclusions.
There is no doubt that solar hot water works very well and I would now never use any other system. However, in view of what I have learned with my system, there are some important points to keep in mind:

•  Use a bronze bodied pump. Cast iron will rust and block up about ever two years.
• One way valve to prevent reverse thermosyphon is necessary.
• A flow meter is not essential, but it does allow fine control over flow rate. The speed control on the pump is not sufficient.
• 58mm tubes are to be preferred over 47mm types.
• I would recommend a minimum of 30 tubes; 36 would be better.
• Ensure connections to adapt to 1/2" copper tube are supplied with the pump and solar collector.
• For one person, a 180 litre tank would be the preferred minimum. Tank should be stainless steel, not glass lined. A cold water expansion valve should be fitted to eliminate TPR valve unreliability. A pressure reducing valve (if a mains pressure supply) and line strainer should also be incorporated in the incoming cold supply.
• Be prepared to experiment with pump switching temperatures. The defaults are not optimum for all installations.
• Use good insulation over the plumbing to and from the collector - considerable heat loss occurs otherwise.
• Use thermal paste on the ends of the evacuated tubes.
• Don't power the electric element from an off peak supply. Heating the water overnight electrically means the solar will contribute almost nothing the following day.