DC thermostat for water heater

[vtpeaknik]

I've built a backup power system with PV panels.  While the grid power is on, it just sits there with full batteries, so I am always looking for good ways to use the power.  One half-baked idea I've tried out is to switch the raw solar power from the panels directly into the heating element in a small electric hot water tank.

I have an on-demand propane-fueled water heater and the output water temperature is not quite steady enough for a pleasant shower, so I added an Ariston 4-gallon tank on its output side on the way to the faucets.  The tank has a heating element which is not plugged in to power, since the tank is mostly there just to even out the water temperature.  It's rated approximately 115VAC 13A, thus about 10 ohms.  With my PV panels (two 180W panels in series) rated 50V 7A at max power, a 10-ohm load would take most of that power, and thus could help keep the water in the tank hot.

The fly in the ointment turned out to be the thermostat switch inside the tank.  To protect against overheating, I fed the PV power via the wiring of the tank intended for AC, so it was still regulated by the thermostat.  But disconnecting 300 watts of PV at 50V makes quite a spark (as I witnessed recently when I accidentally disconnected the panels under load).  This presuambly destroyed the contacts in the thermostat in short order (weeks?), leading to no connectivity to the heating element.  I've recently taken apart the thermostat and cleaned the contacts and it works again, but I know it won't work for long if I reconnect it to the PV panels.

Any ideas on how to make this sort of thing work?  I can see at least three approaches:

(1) replace the thermostat switch with a more robust one.  This is probably most easily done by leaving the thermostat in place, but using its switch to control a low-energy circuit that will in turn control the real heating power, via a relay.  But can I get a relay that will handle, 5A at 50V (or even 60V)?  How about a solid state relay for this purpose?

(2) alternatively, convert the DC power from the PV panels into AC?  Obviously one can use a regular inverter, but that would use more power than the panels can provide, requiring some sort of timer setup to limit the total daily usage.  But the simple resistive heating element can use something other than 115V 50Hz.  Only need to kill that big spark upon disconnect.  A simple circuit that would reverse the polarity of the DC many times a second would do. Of course, a solid state circuit that could do that could also serve as a relay as in (1) above?

(3) use a smaller portion of the PV power, for a smaller spark in the thermostat contacts.  I've been recently running the water heater off the auxillary load output of a PV charge controller.  I.e., it only gets 12VDC, not 50V.  And the controller automatically turns it on only when the battery is being charged (reaches 13V).  That only yields about 15 watts of heat, so it is of marginal utility.  Perhaps good enough to help keep the water in the small insulated tank hot while no water is being used, but not useful for heating cold water.

[dbcollen]

I would use a DC solid state relay, with a voltage dividing resistor array to drop the PV voltage to 5-10v and limit current for signal voltage and interrupt that with the ac thermo switch. It should interrupt a few dozen miliamps @ 5-10v reliably.

[vtpeaknik]

What happens if you use an AC SSR in a DC circuit?  Wouldn't it work anyway, since AC looks to it like the alternating DC that it is?  Or am I missing something?  There are plenty of $5 25A AC-output SSRs available, but DC ones are much harder to find.  I suppose the AC ones may have more of a voltage drop while conducting, so not great for switching 12V circuits, but I'm talking 50-60 volts DC.

[dbcollen]

The AC SSRs switch at a zero crossing, there is no zero crossing in DC so once you turn it on it will never switch off. Also size your SSR to at least double your load and put it on a good heatsink

[richhagen]

I like option number one.  I have a solid state relay designed to switch 40 Amps of DC at up to 80V, so I know they are made, but I do not know the cost of them.  You might be able to make something similar with a bunch of paralleled FET's too. 

[vtpeaknik]

Thanks for the tips.  It seems that to find DC SSR's on feebay one needs to include the search term "DC-DC", otherwise they get buried between the AC ones  There are some around $14 if shipped directly from China.  Some of them I am not sure what they really are, the picture does not agree with the text.  But some clearly show "DC" on the output labeling.  One more well-known source can be found with the model # "SSRDC100V40A" but it costs more.  Can anybody recommend other sources?

Would it be safe to use one rated for 60V when my panels can reach (or even slightly exceed) that voltage when NOT under load?  Or should I hold out for a 100V model?  I assume the voltage limit is due to the breakdown voltage of the switching transistors?  What kind of transistors do they use in DC SSRs (as distinct from AC ones) and how much voltage drop would there be within the SSR?  The latter affects the heat dissipation needed.    I am hoping that my 5A load would not require giving the SSR much of a heat sink.  I saw one model, CLION brand, rated 25A 250V, described with these notes:
1.When load currency is above 10A, a radiator is necessary; when above 80A, a fan is necessary.
2. When controlling inductive load, a varistor is necessary and its voltage could be 1.6-1.9 times of load voltage.

[madlabs]

There are AC SSR's that are what they call "arbitrary", meaning they are not zero crossing. But I don't know if that works with AC. One rarely sees them on the surplus market but you can order them.

Jonathan

[SparWeb]

I've been doing option #2 for a year, and I find it more convenient than #1.

But first:  Do you have an inverter?  What kind is it?  And what kind of PV charge controller do you have?  Will it regulate a diversion load, or is it just a PWM solar controller?

If you don't have these, or aren't going to invest in them, then my suggestion won't help you.

[kitestrings]

vtpeaknik,

I don't know that you didn't mention what you were using for a PV/battery charge conrtoller.  My bias is not to eat up capicity of your inverter, and secondly not to have to rely on an inverter for anything that we don't need to.  We've been diverting, as in your first option, using a stock tank and thermostats via SS DC relays to control DC elements for a few years now.  Similarly we have a tankless dhw heater that works from the pre-heated inlet.  It has workeded really well for us.  I posted on some time ago here for what it is worth:

http://www.fieldlines.com/index.php/topic,130053.msg845366.html#msg845366

good luck,

~kitestrings

[OperaHouse]

This seems like an easy job for a simple FET.  Generally the gate must be prevented from going over 20V A 12 or 15 volt zener would do that.  Then just tie the gate to the high voltage through a 100K resistor.  Be aware that switching almost non existant gate currents with the thermostat can be as big a problem as switching high current DC.  Over time the contacts oxidize and form a non conductive layer.  Normal current and voltage burn these away.  In your case these probably wont show op for several years and you have contact cleaning experience now anyway.

This is something I will be doing at camp this year.  Inverters are actually a practical solution if you know how to modify them.  I have quite a stock of defective HF 2000/4000W inverters.  I was buying these from TAD on ebay and almost all of them have two shorted FET in the output H bridge.  I generally do no more than wiggle the shorted FET till the leads break.  Then they are a very conservitive 1000W inverter.  Some I removed all the FET in the output H bridge and output only 140V DC.   With these inverters I run CFL lamps and TV DVD and other home electronics that don't voltage double on their power supply.   

These HF inverters ust a TL494 switchmode chip.  It is easy to find the voltage sense pin and force it high to almost turn  off the inverter.  Using that pin to monitor the 12V, the inverter can be turned into a PWM dump controller.

I'm just working on a store animation project that went from simple to a lot of add on items.  Looking for a cheap controller I came upon the ARDUINO UNO.  Less than $20 shipped to you and a free development system that programs it with only a USB cable.  How can you go wrong with that.  Six analog inputs and 12 digital I/O.  This is the way to desigh a smart dump controller that can monitor voltage, temperature, anticipate demand and timing features.

[vtpeaknik]

Cool hacking of semi-dead inverters, Operahouse!  I should try that some time.  Thus far I've been shy about futzing with HV switching circuits, but if they come cheap enough I can sacrifice some magic smoke, uh?

Kitestrings: thanks for the link to that other page, looks like you've done something like I'm trying to do, to use surplus PV power when available.  (And I too am in Vermont, in the town of Essex.)

In my case, the main water heater is an on-demand one that burns propane (Baxi Combi).  It heats the house too (except when I burn wood).  Alas it does not modulate the flame fully, only partway, and between that and my modest DHW usage I havn't bothered with thermal solar hot water.  But if I can use surplus PV why not?  Currently I have a small (4 gallon) insulated tank on the output side of the on-demand heater, to smooth out the water temperature which does fluctuate some.  (My theory is that small fluctuations get amplified by my waste-water-heat-recovery device since it introduces positive feedback into the system.)  Perhaps later (when I find a good deal on a Marathon tank!) I'll add a pre-heat tank on the input side of the main heater, like you have.  Eventually the Baxi boiler will die, and the newer models modulate the flame further.

Question: if the pre-heat tank sits around at moderate temperatures, are you worried about Legionaire's disease?

Regarding the other controls in my system, I now have two MPPT controllers, with a pair of 200W PV panels on each (800W total).  One of the controllers has an "auxillary load" output (12V derived from the battery) that it can switch on only when the battery is above 13V, i.e., being charged.  They call it "refrigerator mode".  But I think if I put in the SSR system for a water heating tank I'd rather get the control signal from ghurd's dump load controller circuit, which I currently use to turn on a vent fan for the battery box at about 14V.  That very same decision point could serve as the dump load signal.  I just hope the sudden withdrawing of 5 amps from the PV to the water heater (translating to at least 15 amps on the battery side of the MPPT controller) won't bring the battery voltage back down under the ghurd controller's hysteresis point, turning the dump load back off, rinse and repeat relaxation oscillator!

I also have the inclination to add two controlled AC outlets off the inverter output.  Each would be switched on/off by an AC SSR.  One controlled by the 13V aux load output of the PV charge controller (on most of the daylight time), and one controlled by the 14V signal from the ghurd controller (usually on in midday-early afternoon period).  That would give me more flexible ways to run other loads.  E.g., in sunny weather, I could put the freezer on the 14V-controlled outlet, running it only when the batteries are full in the afternoons, using otherwise-lost power, and not running it at night (storing up the demand for the next afternoon).  If that works well enough, I can start running the refrigerator on PV power too, in sunny weather.

My system is intended as a backup power system for when the grid goes down, thus I always have the tension between trying to keep the batteries topped off (in case of an outage) vs. trying to make some use of the PV power other than floating the batteries.  That's why I'm concentrating first on using the surplus power on sunny afternoons.

[ghurd]

I would go with a fet or 2 or 3 too.
Paralled fets are less simple than many websites lead a person to believe, but in this slow switching speed area its not hard or expensive.

Many cheap China/eBay DC-DC SSRs appear (to me) to be large transistors in a Darlington configuration, and have a substantial Vf loss which need a big heat sink and a fan.
Plus the large Vf loss means less heat in the water.

Can you tell if the themostat has sweeping contacts?
If it does, may not have a contact problem before the tank rusts through.
G-

[vtpeaknik]

If the SSR drops 1V out of 50V (raw PV output) that's not too bad.  For switching 12V it's another story.  I'm also thinking of using "surplus" PV power to run a Peltier-based dehumidifier (Eva-Dry EDV-2200), those actuall run on 12V, 6A.  I got one with a dead AC adapter (common) for cheap.  In that case, I actually want some voltage drop, because it would be running off about 14V and I'd rather only give it 12V or so, to not "overclock" it.  I've wired it up with a 20-foot-long 18AWG cable for that reason.  Running that through an SSR with a 1V drop is therefore also OK.  There's another thread going about Peltier cooling, but I'll mention here that I've read that those dehumidifiers are about 4 times LESS efficient (coolth/watt) than good compressor-based ones.  Reason to use one anyway is that it's "free" power, plus low cost and quiet operatoin.

The thermostat I took apart had contacts that did not look like "sweeping" to me.  And they definitely looked pitted and were not conducting until I rubbed them with fine sandpaper.  This water tank has never been plugged into AC supply, and was only connected to my PV panels for a few weeks.  And it's the second one I've had that stopped conducting under that treatment.

Corrosion of the tank is definitely an issue, I'm on my second such tank (Ariston GL4), the first one only lasted about 4 years.  I'd love to find a non-metallic tank.  The only ones I know of are the Marathon brand, they're all bigger than I need for mixing the output of the heater, and start at 3x the price.

[electrondady1]

i wonder what the effect wold be of placeing  your p.v. heated water tank on the supply side of your on demand system, rather than the output side as you described.
if the water was already warm/hot it would enable your propane system to more easily reach the proper operating temperature.

[vtpeaknik]

Yes, ideally I'd have two tanks, a bigger one (20 gallons or so) on the intake side, for solar preheat, and a small one (4-6 gallons) on the output for mixing.  As I mentioned above, the specific boiler I have can only modulate the flame down to a certain level, not quite enough for full use of solar preheat potential.  I suppose if the preheated intake water gets too hot, the flame would go out altogether.  Also in that case I can turn the boiler off and enjoy a 100% solar shower :-)

[thirteen]

Is the problem with the tanks or is it in the water. Maybe try a kitchen or dairy supply place for a plastic tank. Is your system a gravity flow or pump pressure system?
 I am putting in a pump booster for my on demand water heater. When some one turns on a faucet and some one is taking a show the flame will go out because of lower pearsure. I have an old water tank that is in the corner and this lets the water get to close to room temperature. and the pump will be just after that.  I am thinking I am going to plumb it over next to the cook stove. I got a stainless steel tank is 6 ft long and 14 inches wide and about 5.5 inches deep. My plans are to mount it behind my stove pipe and heat water. Expansion of the water might be a problem. Got it for nothing and trying to figure out a good use. Hot water from cooking and hot water to wash dishes with. A float valve that can stand up to hot water might work OK.

[vtpeaknik]

In my basement which varies from cool (60F) in the summer to cold (<40F) in the winter, I think I should have a tank with insulation if preheating the water is to do much good.  The Marathon tanks are well insulated.  There might be a way to insulated another kind of tank myself.  But I'd still need a heating element (if I want to use surplus PV power to heat water), and for safety would need a thermostat and an over-pressure safety valve, so in other words I do need a complete electric hot water tank.  The Ariston tanks have a glass liner inside the steel tank but the first one I had corroded anyway, at the edge of the opening where the (unused) heating element inserts.  Presumably the glass liner was chipped there.  Also, the anode was corroded away, despite almost no use of the heating element, perhaps had I replaced it in time it would have prevented the corrosion?  But I don't dare open up the newer tank while it's still good, since if I cause any chipping of the liner then I might cause it to corrode.  Catch 22!  A non-metallic tank must be the way to go.

[ghurd]

was only connected to my PV panels for a few weeks.  And it's the second one I've had that stopped conducting under that treatment.

Direct connection to 50V DC is NOT a great idea, which is why you posted.

With mosfets, the thermostat will be switching maybe a few microamps at 15V.
IMHO, most contacts in something that size will have enough "sweep" to keep the contacts working for a long long time.
A tiny DIY spark/arc arrestor will help too. Cost like a quarter?

Also, almost all contacts are coated with something fancy.
Sandpaper removes the oxidation, but it also removes the material on the contacts that helps keep the oxidation away.
Back in my college days (when the world used relays), we were taught to use a crisp clean new USA dollar bill, because it is less abrasive but enough to take off the crud.
G-

[kitestrings]

My Dad, who repaired small engines and golf-carts for years, had always taught me to use a dollar as the last step on the points of engine.  I can't say that I knew why at the time, but I did it anyway.

Lot's of ways to slice an apple..., but a couple things -

We pre-heat on the supply-side of the dhw before it enters the tankless heater.  Our tankless unit allows you to set the output temperature and modulates to that temperature.  The batteries are happy before the diversion begins, and because the source is not only free, but otherwise untapped, we heat to the max temp that the tank allows.  This is 170 deg F in our case.  With the mixing valve you can adjust the outlet (or inlet to the tankless heater) so that it is safely below maximum inlet (low-end of the burner stage/modulation).  And, this allows a few more Btus to be squirreled away on a good day.  I only wish the tank were a bit larger.

We have three boys under the age of 7, so I don't worry much about unused hot water.  We're in Irasburg (NEK)

~kitestrings