My read with this system is this is the sensor. It looks like it could be used with a compression fitting as the actual sensor looks to enclosed in a metal housing. They mention: "Uses Arizona Microchip 8700A sensor chip" which I had no luck chasing down?

Ron

 

I wasn't able to find any info on that Microchip either, Ron.

Thanks for the suggestions on waterproofing the sensor. After some research, it looks like using epoxy is a workable waterproofing solution (it just delays reaction time a bit).

I was able to track down another controller similar to the SolarFriend3, but which uses waterproof sensors. It's not quite as sexy, but it gets the job done. Here's the link. Mine will have a digital readout of both sensor temps, an adjustable differential setting which tells the pump when to turn on, and a relay for the pump. Cost is about $50USD + $6USD shipping.

This is turning out to be quite the fun little project.

 

Cost is about $50USD + $6USD shipping.

I don't think you'll regret that. I could do it for less because I everything on hand including some experience with the parts. But if I had to start at zero, design a board, find and order all the parts, test and build it, that $56 looks pretty competitive if you just want to get on with the larger project.

 

Looks to be a nice little unit for the buck. I never realized how popular those little controllers for solar were.

Ron

 

I have been reading this post with interest. I was having much the same query.
Then I built this....

Despite the several recommendations not to use an LM741, this circuit works.
Also attached is the PCB design I used as a PDF file.
I used ExpressPCB to design the schematic and design the PCB outline.
I have included the ExpressPCB schematic in this post.

You may want to read my previous posts from earlier here and also here

I am not suggesting to use an LM741 (and is probably not as good as a LM393 or LM358) but the circuit above appears to work well and I am happy with the result. I also stuck a manual override switch in to enable the pump to run when it is off.
Using the 1N4007 can handle a pump up to 1A at 12vDC
I have also used 10k NTC waterproof thermistors

 

this circuit works

Don't you find the relay chatters? There is no hysteresis around the 741 (unless the 220k was intended to feed back to the non-inverting input of the opamp).

 

Don't you find the relay chatters? There is no hysteresis around the 741 (unless the 220k was intended to feed back to the non-inverting input of the opamp).

Yes, you are correct there is no hysteresis.
I used the 220k resistor as a negative feedback resisitor attached to the inverting input not a positive feedback resisitor attached to the non-inverting input, there it appears that the relay switches on hard at a determined temperature and the relay does not chatter.
I have set the 10k preset so that the relay switches on when there is approximately a 20°C difference between the thermistors.
The only problem I have noticed is that the LED glows dim for a few seconds after the relay switches off. But apart from that it appears to work well.

 

Additional....
I have also built this...

When using an LM393 I am getting quite serious relay chatter and the LED flashes as the relay is switching on and off.
I have got a 220k resistor being connected to the output and the non-inverting input to add some hysteresis, but the relay still chatters.
I cannot understand why this is happening. Maybe I should increase the size of the resistor to about 1M.
And when using an LM358, the relay does not switch off at all, despite the LED being off
I also cannot understand why this is happening!!

 

I have got a 220k resistor being connected to the output and the non-inverting input but the relay still chatters.
I cannot understand why this is happening. Maybe I should increase the size of the resistor to about 1M.
And when using an LM358, the relay does not switch off at all, despite the LED being off

The relay chatters but also it does not turn off?
I suspect the relay not turning off is because the LM393 circuit is missing the base-emitter resistor that was present in the LM741 circuit.

 

I suspect the relay not turning off is because the LM393 circuit is missing the base-emitter resistor that was present in the LM741 circuit.

Interesting, I shall add a base-emitter resistor and see what happens.

 

I have made some calculations the voltage dividers.

If the two NTCs is at a difference of 1 degree centigrade, there will be a difference of approximately 0.1 volts from a nominal midpoint of 6 volts. 220K hysteresis resistance should also bring a change of 0.1 Volt on the non-inverting input, so there should be close to 1 degree Celsius hysteresis of the relay.

I have outlined some modifications in the schematic below:

• 5K1 resistance in potentiometer: to provide at least 2 degrees Celsius hysteresis wherever potentiometer is set. The resistance ensures also that SW1 can't short-circuit the supply through the potentiometer.

• Relay driver base resistance: Moved up on the anode of the LED so transistor can go totally off.

• 100uF and 100nF capacitors: Decoupling the supply, 100nF mounted close to the OP Amp.

 

I have just added a 2k2 resistor into the circuit between the base and emitter of the transistor.
With the LM393, there is a very slight delay and chatter upon the activation/deactivation of the relay and the LED flickers when switching.
I then tried the LM358 and this is working beautifully, there is no chatter whatsoever, the LED changes from on and off and the relay switches instantly.
Here is the new circuit... This I must admit is working better than the LM741

 

Nice to know you've got it working well.

 

All this is very interesting but it seems rather complicated. I have a solar hot water system. The two panels are on the roof with a cold water inlet and a hot water outlet. The hot water outlet feeds into a 150 or 200 litre insulated storage tank under the roof. Inside the hot water tank is a thermostaticaly controlled immersion heater adjustable, but preset to 70C, the optimum temperature recommended by the electricity company. Normally, the solar panels supply enough hot water. On cold or rainy days I switch on the immersion switch which has a built-in time clock.

 

All this is very interesting but it seems rather complicated. I have a solar hot water system. The two panels are on the roof with a cold water inlet and a hot water outlet. The hot water outlet feeds into a 150 or 200 litre insulated storage tank under the roof. Inside the hot water tank is a thermostaticaly controlled immersion heater adjustable, but preset to 70C, the optimum temperature recommended by the electricity company. Normally, the solar panels supply enough hot water. On cold or rainy days I switch on the immersion switch which has a built-in time clock.

Is your cold water inlet on the solar panel fed from an external source (city, municipality, well, etc)? If so, and assuming you didn't use much hot water, what keeps the water in the storage tank hot (besides the immersion heater)? Insulation?

I use minimal hot water, so I like the concept of being able to circulate the water in the storage tank back into the solar heater, so the sun, rather than an electric heater, provides the heat.

 

And when using an LM358, the relay does not switch off at all, despite the LED being off
I also cannot understand why this is happening!!

The posted fixes for this work, but I don't see where the problem was explained. While the 358 output stage can swing close to its negative rail (very close for its day), it cannot get that close to its positive rail. When the output is saturated high, it still is about 1.5 V below Vcc, more than enough to sink a little current through the BC327 base resistor. I like the fix in post #31 the best.

ak

 

LM35 comes in a TO220 package and has a nice chunky heatsink tab (connected to ground don't forget).
Ideal for flat surfaces. With a little plumbing expertise, you can make up sensors for any situation.
I use them on the flow and return pipes os the gas boiler - in UK, the pipework is 22-mm OD. Cut a "slip joint" in half to fit snugly on the boiler pipe. Cut a 3/4" brass compression nut with flats in half - file out the threads to get a smooth surface. Solder the nut to the slip joint so you have a flat surface for LM35. Drill and tap the flat M4 and screw the LM35 tab to the flat and cable tie the whole thing to the pipe. Arduino or XBee does the rest.
For water proofing, make up a copper tube pocket, again plumbing stuff, a use a little ingenuity to get an IP68 cable gland fitted, or encapsulate with resin. Hot melt might work, but not as good as plug of resin.
I haven't read all the thread, but storage tanks tend to stratify into hot and cold layers. I use a heating circulation pump to keep things mixed up.
If you get a lot of stratification, you essentially set up an unpumped "gravity" heating system

 

The hot water from the solar panels are fed into the bottom of the very well insulated storage tank so there is a continuous circulation of water without a pump. The inlet water supply is from the municipality. In winter when the weather is overcast the immersion heater is switched on automatically with a time clock.

 

My hot water tank is a conventional "wet" system, boiler heated.
When the boiler heats the tank, the heating coils at the bottom mix it up OK, but for some unknown reason, this tank has the immersion heater at the top, so you only ever get half a tank of hot water.
In summer, the off-peak electricity is the cheapest way to do the hot water, so a circulating pump (old C/H pump) wired across the immersion heater and a couple of Essex flanges mixes it a treat.
Before anyone jumps up and down about the wiring, the pump is rated at about 20-W, which is no significant extra load on the 3000-W immersion heater circuit

 

The posted fixes for this work, but I don't see where the problem was explained. While the 358 output stage can swing close to its negative rail (very close for its day), it cannot get that close to its positive rail. When the output is saturated high, it still is about 1.5 V below Vcc, more than enough to sink a little current through the BC327 base resistor. I like the fix in post #31 the best.

ak

Try a Schmitt trigger on the comparator output to get a clean 0-V to Vcc output. 74HC14 (hex inverter) or 40106. Work much better than transistors for gittery signals