Excess PV Power Diverted At Power Point To Heat Water Efficiently


Solar Addict
I've been diverting excess PV power to heat water at my camp for years. It is the only source of hot water as I got rid of my propane heater. A simple method is used to determine when there is excess power. At the power point voltage, the maximum power is obtained from the panels. Use less power and the voltage will rise up to the open circuit voltage depending on the current. Light intensity only changes the panels current, not the power point voltage. This circuit diverts enough current to drop the panels voltage back to the power point. It can be used in parallel with PWM and MPPT controllers. No battery is needed and it can operate in stand alone mode.

The principal of operation is energy is stored in a capacitor bank and is PWMed to a heating element. Here is a simple example. Think of a 10A heating element which is ideally matched to the solar in best conditions. In less than ideal sun that panel can produce 5A. For 50% of the time the panels 5A charges the capacitor bank when the heating element is turned off. In the next 50% the heating element turns on and receives a full 10A, 5A from the solar panel and 5A from the capacitor bank. The panel always sees the same ideal voltage so it operates at maximum power. The duty cycle is changed by sensing thevoltage on the capacitor bank and is kept within a fraction of a volt. Updates are 120 times a second.

Here is a video of a commercial product AC7931 that does this. I have not personally used it. It does miss design targets because it requires an external isolated power supply, has to use an internal fan, and capacitor bank may not be large enough for long life.

There are many circuits that can accomplish this from discrete to micro and I've built most of them. This is the heater control I use at home. It is an unconventional use based on a IR2153 half bridge driver chip which provides many of the functions needed at very low current. It is highly stable with built in voltage regulator, FET driver, oscillator and under voltage protection to prevent FET damage. Pictured in case is a low cost board from China which was modified. I can no longer recommend these because which board you get now totally random regardless of what the listing says. Some have the correct 8 pin chip and some the 16 pin SG3525 chip. Both can be made into controllers. However, they now come with 55V FET which makes them only suitable for 36V arrays max. Poor trace layout also makes them unsuitable for more than 60V. The very small number of components make it suitable for a perf board construction and I recommend using four 110N15 which are 110A 150V FET.

Reasons to divert power from panels instead of battery

1. Lower current and lower resistance losses with higher voltage
2. Smaller charge controller
3. Smaller battery bank
4. No double or tripple conversion loss from charge controller, battery, inverter
5. Battery left as fully charged as possible
6. Does not interfere with charge cycles
7. Captures energy when charge controller is not transfering maximum
power to battery during charge stages
8. Inability to actually know charge state of battery
9. Can capture even small amounts of extra power
10.Often uses standard heater elements instead of battery types

An interesting feature of this circuit is mechanical AC thermostats can be used in switching primary DC supply without arcing when the array is less than 100V. A capacitor bank of about 6,000uF is required made of multiple capacitors to handle the high surge currents. The front panel consists of a power meter and a pot to set the voltage. A switch also allows direct connect to the heater elements for performance comparison. Here are some typical values. Due to low tilt, maximum panel power is about 395W into 7.5 ohms. 44W may not seem like much, but that is what is needed to replace lost heat. The blue wire out of the power meter is a modification to provide external power. Low power levels of direct connect often do not provide enough voltage to self power the meter. Also shown is a breadboard build.

44W 5W
65W 9W
117W 47W
118W 52W
205W 103W
265W 195W
302W 226W


  • 2153front.jpg
    51.5 KB · Views: 43
  • 2153inside.jpg
    60.2 KB · Views: 55
  • 2153schematic.JPG
    63.8 KB · Views: 68
  • IMG_2153.JPG
    56.5 KB · Views: 59


Photon Sorcerer
Thanks for posting your info. Out of curiosity where did you source a 7.5 ohm element from? Are you running two elements in parallel, a bottom and mid/top tank heater?

From what I can see of common types in 110VAC land they top out at 1500 watts at 9.6 ohms. After than they go to 220V with a 4500 watt element being around 10.7 ohms. In 230VAC land (aka Australia) we have 4800 watt elements which are around 11 ohms.

People need to be careful switching DC with the standard AC thermostat. Check the specs for the particular device. If no DC rating is provided do not assume it can handle it. It's not just the thermostat contacts that need to be considered. There is also a thermal cutout to think about and often those are built into the thermostat housing. Water heater tanks are not something you want to even think about defeating just one of the safety features.

I thought I recognised those pictures!



Solar Addict
You are in the land of 240 which makes 2,000W 120V 7.5 ohm heating elements not only scarce but with different mounting. This tank only has one heater and 60V is a good match for that.

All unrated AC mechanical contacts should be considered 30V DC max. This system when properly set up does not violate that. The manual reset thermal over temp switches have two separate switch contacts for 240V AC. On DC you can quickly exceed the current rating. In two heater systems, use one section for each heating element. That will reduce current to acceptable levels.

Heating water is simple. It just gets complicated doing it right.


Photon Sorcerer
You mention 60v but also then say 30V DC max across the thermostat, how are you achieving that?

*edit to add below*
My own solar array has a Vmp around 60v so a single 4500 watt element would do about 330 watts peak for me. I'd have to either arrange a charge pump or hook another element in parallel with it. My tank is capable of having two hook style elements at the bottom so I could potentially have two 4500 watt elements I could push around 650 watts at it which would be sufficient to ensure that watts into the tank exceeds watts out of the tank on a bad day. Obviously since the tank current max at 60v would be well under the capacity of the array the array voltage would be higher but I don't know where it would really be so I'll assume 60v for the sake of the argument :)

I can buy decent quality elements around $50 AUD.
Last edited:


Solar Addict
Arcing is an issue when switching over 30V DC. This circuit, in combination with a controller that keeps the capacitor bank at
a constant voltage, allows using standard water heater thermostat contacts to switch DC. It only requires that the difference between
the open circuit voltage of the solar array and the differential voltage across the switch will be less than 30V DC. This can only be done with a heater controller that keeps the capacitor bank at a constant voltage. When the switch opens, the controller acts as a shunt regulator preventing a drop in voltage on the capacitor bank below the set voltage. The 300 ohm resistor keeps the capacitor bank charged with a small amount of
current and the controller almost completely shuts down. Heater draws about 2W in off state.



    67.2 KB · Views: 21


Photon Sorcerer
OK, let me make sure I am getting this right. You are using the 2153schematic to charge the capacitors and placing PV across the thermostat which itself is bridged by a 300 ohm resistor? Would you be able to draw all encompassing circuit diagram including the switching FETs for the capacitor charging circuit so that anyone undertaking your system would have a full diagram to follow?

Buzzard Bait

Solar Enthusiast
I am a novice, and I only have one water well running on solar at present. I am working on another array with 1320 watts of panels, and I believe there will be times that my batteries will be charged to capacity. I will be using a MPPT controller that I already own, and have been thinking about heating some water as a diversion load when my batteries are in float. The method that has been described above is beyond my level at present, but it is nice to know I am not alone in my thinking about a diversion load. Thank you for sharing. I must first get this new array going then I will look harder at heating water.


Solar Addict
IR2153 Two Element Priority PV Solar Water Heating At Power Point Feature

Every design has at least one bug. Out in industry we call these bugs selling features. Normally the IR2153 is used as a half bridge FET driver with a 50% duty cycle on each output. In this high/low side mode, the low output (pin #5) MUST turn on first in order to charge the high side
bootstrap capacitor for the high side driver. Without enough voltage, the high side driver would put the FET into a linear region. With excess heating possible, destruction could result. IR2153 is designed to always turn the low side driver on first to charge this capacitor. Shorting the timing cap to control the load not only turns off the oscillator, but resets it. The low side drive has to achieve about 50% duty cycle before the high side driver will start to turn on.

This Power Point control turns the oscillator on and off to maintain a constant voltage on the capacitor bank. Usually the two DRAIN tabs of the output FET are also joined together allowing up to a 100% duty cycle on a single heater element.

In order to quench an arc on a mechanical thermostat contact, the maximum on time must be less than 11ms with an off time of 1ms or greater.

This priority powering of the low side driver can be a very useful feature as the low side will always be on more often giving that load priority to any extra energy.Instead of connecting the two FET drain together, each FET can drive its own heating element with the thermostat contacts in series. This insures more than enough time for the arc to quench when the contacts open. The LOW side drive (pin #5) could be used to power the upper heating element and have priority heating first. This would insure fast recovery in the morning when available power is lower. As noon approached and there was more energy to harvest, the lower heater would also turn on resulting in two heaters operating for maximum load. In the afternoon as excess power lowers the upper heater would be powered until the upper thermostat opened. Then all power would go to just the lower thermostat and element.

This picture was taken when the panels were producing 255W into the heating element. Note that the FET for the high side driver (pin #7) has not turned on the FET.


  • IR2153half_bug.jpg
    62.9 KB · Views: 21


Solar Addict
This is the closest thing I have on file showing how energy is gathered from the off times of a PWM charge controller. Note this for a single heater element with thermostat in the DC supply. If two heater elements are used, only one FET output per element, thermostat can be placed in series with the heater element for a simple two wire connection On time is sufficiently short and off time is long enough to quench arc.


  • Camp_Water.JPG
    93.7 KB · Views: 32


Photon Sorcerer
Surely if you have done this for ages, and presumably are using the setup yourself, you can do up a full circuit diagram of what you actually have in place and implement?


Solar Addict
It is not in my nature to do that. There is no one system that works for everyone. Configuration depends on the hardware you have. This system should not be done by someone who wants to paint by numbers. I will explain in circuit sections and each one should be understood. If they don't understand the functions, it can never be fixed when it breaks. This is fairly simple and something that can be built by a 14 year old. They are curious, believe they can do anything. They haven't learned fear and failure. One does not learn without a little struggle. Those not interested in learning something should just pass this by. This isn't really a DIY board and have little expectation someone will build this. Happy to answer any questions, not in doing engineering work for anyone.


Photon Sorcerer
For anyone that wants to see a better diagram of what is going on don't bother asking here, clearly, just view this youtube clip directly. It's a pretty fundamental arrangement so why all the secret squirrel, I'm not sure.

Principle point is that this is simply dropping your HWS element across the PV/dirty great capacitor bank with the bank being charged by the same PV. A simple voltage sense circuit connects / disconnects the element around your manually set takeoff or somewhere over maximum power point voltage to permit your SCC to pull the voltage down and obtain maximum power for itself.

Your PV and heater element need to be matched in terms of voltage vs desired watts. If you target 60v as the manually set power point you will need an element (or two in parallel) with a total rating over 1500w (110VAC) and achieve approx 450 watts maximum (not accounting for time to charge the capacitors at each cycle). If you have a 220/230VAC element (this is approx 10.5 ohms) you will need to target about 68v for the same power level.

The further you allow your array to go off Vmp the less wattage is produced, the longer the capacitor charge time, the less on time of the element so you do need to be fairly close to Vmp if you don't want to throw away a fair bit of heating capacity.

Water heating needs kWh of energy which is easily derived with l x 4 x c / 3412 = kWh, where l = volume of water in litres and c = temperature differential in degrees c. Suppose you like to shower at 40c, and take a 5 mins shower with typical water flow... about 100 litres of water... and your cold water supply is 20c. 100 x 4 x 20 / 3415 = 2.4kWh for one shower. If you heat your tank to 75c, you have the hot water set to low flow and will be mixing in more cold water to bring the temperature to where you are comfortable, so assuming you don't broil yourself or are using firehose flow rates to shower... the actual watts for your 5 mins shower are the same. 450 watts may not be adequate.

*edits to fix grammar etc*
Last edited:


Solar Addict
That is the schematic of the partial hot water system at the camp, at least at one time. It powers a 6 gallon ECOsmart tank off of a 60V array (60 cell grid tie panels). At 125V AC it is rated at 1440W, 10.8 ohms. As a simple formula, 6 gallons raised 60 degrees F takes 1,000WH. This tank
is in series with a 9 gallon tank used as a pre heater. On average the pre heat tank obtains about 2500WH excess power from the 36V array (50V) a day. The 6 gallon tank typically uses about 350WH and that is primarily recovery from tank heat loss overnight. It is usually supplied with full temp preheat water that requires no heating by 10am in the summer.

The 6 gallon tank controller does double duty as it powers the dishwasher heater when used. Switch over is automatic. Dishwasher can add another 400W to the daily total to increase water temperature and for heated dry. A lot of excess power is still not harvested from the 60V array. Next year I am adding another 30 gallon tank just for the clothes washer to harvest any extra.

15 gallons total may not seem to be a lot, but is enough for two showers and dishes each day. In a camp, just 6 gallons of hot water can add a lot of luxury. The tank uses a 4 wire plug and an adaptor cable allows us to connect it to a gas generator on very bad days. 15 minutes of generator time are needed to raise the temperature 60F.Box says the temperature control can go up to 150F. Probably from potential lawsuits it actually shuts off at 125F.
Higher temperatures and a mixing valve would effectively make it a larger tank. Our propane heater was 9 gallons which was a pain to wait for and needed refill tank quite often.

You have no moral high ground. I try my best to disseminate this information. It is no secret. This is more than enough information to build this for someone who is competent in the field. I am not going to draw pictures of IC's and resistors because people won't look up a data sheet or color code. I don't tell people what they can't do. If this is beyond anyone's understanding, it may be a good indicator that you shouldn't build this. A water heater can be dangerous and this system could be used in a manner not intended. This is experimental and what I have done. It may not pass any applicable codes. You are welcome to find anyone else in the entire world with a working system telling you how to build this. Good luck. Finding anyone interested is just as hard, there has been no interest. When the 'cool kids' finally get around to having this, everyone will just have to have it. I had a farmer in the middle of nowhere build one of my 3W multi element microprocessor systems who had never done electronics before. Never even knew he was building it till he couldn't get the software to load. So unfamiliar with computers he had retyped the entire program instead of using copy. A reload and it worked from the start. I sent built working systems to two individuals who claimed to have some understanding. Neither of those have even been connected after two years. Go figure.


Photon Sorcerer
6 US liq gallons is about 23 litres
60f is about 15c
23 * 4 * 15 / 3412 is 0.4kWh add in a bit of thermal loss from the heating the tank itself and radiation from the tank.

Notional 50v into a 10.8 ohm element gives approx 231 watts. This assumes continous supply, which isn't present with the simple circuit you are using. You have a capacitor charge time which is a function of the amps available from the panels, and a capacitor discharge time which is a function of the voltage in the capacitors vs element resistence, ie you a duty cycle to consider. But let's assume that it is a constant supply. 2.5kWh from 231 watts, well that's over 10 hours of strong sunlight.
Last edited:

Deleted member 783



Solar Addict
Random placement of numbers is not a fair evaluation. It is 60V into 10.8 ohms and 50V into 7.2 ohms. Yes, 2.5KWH is the typical total power sent to heat water a day due to very high loss 9 gallon tank. The 6 gallon takes about 330WH to recover from overnight heat loss. That doesn't take long to make up. It is all about the numbers and anyone can calculate if they enter the right numbers. Heat loss can easily be 2KWH a day with some tanks. Some households use less water than they imagine. I don't make anything up.

The voltage on the capacitor varies only a very small fraction of a volt. Those interested in this technology should watch adiabatic capacitor charging which is similar. This professor has many excellent videos on electronics for the beginner.

Last edited:


Photon Sorcerer
Yeah man. I love it. Thanks for a lead on the tech and your system details.
With oversize arrays being limited by the charge control much of the time in many systems including mine, the diversion is a welcome effect. It would greatly improve our dhw capacity.


Photon Sorcerer
Random figures in one's forum posts doesn't help. So far we've seen 120v, 60v, 75v, 50v, +/- 60 v , 10.x ohms, 9.x ohms and now 7.x ohms. Toss in completely inaccurate calculations of how much energy it takes to heat water along with various incomplete fragments of circuits and even mentions of Arduino (another thread) and it's not hard for a person to become rather tired of it. I wonder, seriously wonder, what the next set of figures will be.

I really don't know why someone would bother posting such a random grab bag like that.