Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 03:29pm 07 Mar 2019
Copy link to clipboard
Print this post
For several years I've been heating water at my camp by this method. Two years ago I totally disconnected the propane heater from the system as the PV was working so well. I was heating a 9 and 20 gallon tanks in series till almost the end of last season when it started leaking. I replaced the 0 gallon with a 1400W 120V ECOsmart point of use tank as the primary. This is very responsive and takes only 900WH to raise the temp 60F. Tempered water takes even less time. I also run my dishwasher heating element off this which switches in automatically. As a bonus I can still plug it into my generator when I've packed up the solar panels for the season. There is no reason not to have ready hot water around a camp when these cheap 6 gallon tanks are available.It's free hot water from power everyone of you waste and this is almost 100% efficient.It is amazing to see it diverting as little as 5w into the tank.
I'm using a modified 300W inverter board with 60V grid tie panels at my camp to heat water. Power comes from excess solar not used to charge my battery. This operates in parallel with a MPPT charge controller. The modified inverter board that gives a really clean look to the build. Only a half dozen parts need to be added. Made from a $6 ebay 300W inverter board (not the limit of the system). These used to come with 75N75 FET, but recent boards come with IRF3205 which are limited to 55V. I've replaced them with 110N15 which is a 150V FET. Be sure the board pictures an 8 pin chip as there are some new variations. A capacitor bank provides storage of energy when the FET are in the off state. Do not use with PWM controllers unless an isolation diode is used between panels and capacitor bank.
This control keeps the panels at a fixed power point voltage. I haven't seen much need to track as the wattage is consistent over several volts. Seasonal adjustments to the voltage seem acceptable. One neat feature is that existing mechanical thermostat switches work well in this circuit. A 270 ohm 5W or higher resistor is placed across the switch contacts. This does two things. The capacitors are kept charges at the set voltage reducing current surge when the contacts close again. The differential voltage across the contacts can be kept under 30V. When the contacts open, the driver immediately shuts off once the voltage on the capacitors drops less than a volt. This leaves the differential across the contacts as the open circuit voltage minus the set voltage.
The TL431 controls the operating voltage. The voltage divider into the reference pin should be designed for 2.5 volts. This is when it conducts heaviest. The cathode can pull down to a little below 2V. The diode, LED, transistor combination create an offset of .6V + 1.7V + .6V = 2.9 volts. When the cathode of the 431 goes above that (indicating low voltage) the transistor shorts out the oscillator capacitor C3 bringing both FET gates to zero and no conduction thru the heater element.
The yellow scope trace is the cathode of the TL431. The bump up is the sensing of a lower voltage sending the board into shutdown. The low portions of the green trace indicate when the FET is conducting and the upper the capacitor bank voltage. Each FET has a maximum50% duty cycle. The output of each is connected in parallel. This is operating in a test where a very noisy boost module is used. The spikes are not real and just inductive pickup of the leads.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 03:36pm 07 Mar 2019
Copy link to clipboard
Print this post
First modification to the board is to remove the black relay. This serves no purpose other than to protect the board from reverse voltage. That is a real possibility normally since the board terminals are not marked. This provides a space to mount the new components.
Next to the relay space is diode D2. That is removed along with the resistor (10K) above the LED. The diode from D2 is now placed in that resistor location with the band towards the LED. The diode and LED act as a zener ( .65 + 1.7V). The LED also provides some visual indication of inhibit. Although, the board is in some degree of inhibit almost all the time.
On the back side of the board, solder some extra solid wire from the FET source pins to the common connector. The outside lead of the LED needs to be isolated. Cut the trace to common and bridge the 10K resistor from the LED pin to the common foil.
A small signal NPN transistor is bridged across this resistor on the backside of the board. Emitter to common and the base to the LED. The collector connects to pin three of the IR2153 chip which is also the timing capacitor. When the LED is on, this transistor shorts out the capacitor stopping the oscillator. For many transistors, the flat side faces the board. Check your pinout.
This design uses the internal 15V zener of the IR2153 for regulated power. The center terminal of the three lug board connector is supplied with 60+ volts. That power goes to capacitor C1 which must be removed. The diode lead D1 closest to C1 must also be lifted. A 5600 ohm 1W resistor goes from the positive capacitor hole to the lifted lead of D1. Two 3K 1/2W resistors in series can also be used. In operation it uses about 6ma. For higher voltages figure about 10ma in calculating the power resistor. From the center terminal, foil also goes to the front of the board. This must also be cut.
I mounted the TL431 in three holes of the removed relay. Cut traces as appropriate and mount. I drilled a hole near positive C4 and run a 6.8K to 10K resistor from positive to the TL431.15K resistor can be mounted on bottom. A .01uF capacitor from the reference pin to common provides filtering of the signal. The reference pin requires 2.5V in the voltage divider calculation. Pot in series is preferred because any pot failure will turn off controller.Use at least 270K as dropping resistor. The reference input begins to conduct like a zener a little over 2.5 volts and input current should not be allowed to go over a few ma.
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 12:24am 08 Mar 2019
Copy link to clipboard
Print this post
In this particular case, the "MPPT charge controller" is a slighly modified 600W boost convertor, yes?
Do you have a link to the capacitors? I would be interested in purchasing them.
These capacitors cannot be any run-of-the-mill electrolytic capacitors as we are demanding a huge current from them, so low ESR is important.
Also, the electrolytes in these caps will start to boil away regardless due to ESR * I losses.
What are your thoughts?
Is that because otherwise the capacitor bank will try and discharge into the panels?
1. What is the need to connect the modified inverter board with these FETs to the output of the 600W boost convertor? 2. What is the issue with connecting the output of the 600W boost convertor directly to the water heater element(s)? 3. Currently does the modified inverter board output drive the water heater element(s) through a thermostat?
hotwater, thank you again for documenting and sharing this. I will be replicating this very build but with the 1200W TL494 based boost convertor davef and I were talking about in the other thread and need all the help you can give us
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 02:21pm 08 Mar 2019
Copy link to clipboard
Print this post
No, there is a 40A MPPT charge controller off to the side which is charging a battery or at lest maintaining it. At home it has been working with two different controllers at times to look for any adverse effects. It will draw about 15W from the panels. At camp there is another 40A MPPT controller that provides power to the camp and this works with this same water heater board. It is an early software version of MakeBlueSky 40A MPPT controller. This controller has some tracking issues with or without the water heater board connected. This is the point that no one seems to understand. You can harvest what does not go to the battery quite easily.
Typical consumer caps are only rated for about 1A. Anything will work for a while. Would like to talk with someone that has had a techluk more than 3 years. The caps I use are rated at 3.6A ripple each and are 1200uF. I happen to have 800 I bought for a project that went nowhere. My first system used about a dozen 220-470uF out of switching supplies. The secret is to have enough of them to provide the low ESR and prevent heating.
A PWM controller is a direct FET connection from the panel to the battery. Panels are a current source and you can't get more current than the panel can supply. From a capacitor bank that can be hundreds of amps. I've has some wires go the wrong way and short out the water heater element. That current surge easily blew out several 110A FET in parallel with ease. The same thing would happen to a PWM controller. The diode keeps the capacitor bank from feeding back current to the PWM controller. It also allows you to harvest all the energy from the panels when the PWM is in an off period.
This is an inverter board. Normally each FET goes to one side of a center tapped transformer. Each FET is turned on50% of the time. The two FET drain terminals are connected together for up to about 98% duty cycle thru the heater. That deadtime is stored in the capacitor bank and not wasted.
The boost converter should be ignored. It is only there because my heating element is small 3500W and 240V. Frankly it sucks and only gives a marginal improvement boosting voltage to 80V with with this element. Get a life and buy the right panels and heating element. It was only an example of what could be done with a marginal panel system. I would never recommend using a boost converter.
I do not have the thermostat connected in this system. It could never provide enough energy. It would really only be needed for really small tanks or extremely large PV arrays . It takes a lot of energy to raise water to 180F.
That said, there is a 270 ohm resistor from the panel to the capacitor bank with an alligator jumper to simulate a tank thermostat. This pre charges the capacitor bank. When heating water and the thermostat opens, the heater immediately shuts off and capacitor bank never goes lower than the set voltage. This means there is never more than 30V across the switch contacts. There is almost no sign of arcing.
I have seen four versions of this inverter board and now there is a new one with a 14 pin chip. Who knows what that chip is. I get the feeling that it could be a IRS453 which is a full ridge driver that is only half used. I don't see the logic of this as all the cost is in the plastic package and smaller is cheaper. The IRS453 is almost the holy grain of H bridge drivers. While limited in drive it is quire a handy chip. I'd buy a dozen boards with these in a heartbeat.
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 06:25pm 08 Mar 2019
Copy link to clipboard
Print this post
Do you have the datasheet for the IRS453? I cant seem to locate it
If all you had were a bunch of 300Wmpp, 40V Voc, 9A Isc grid panels, what water heating elements would you use and what system would connect the PV to the heating elements?
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 05:54pm 09 Mar 2019
Copy link to clipboard
Print this post
The 2 sometimes doesn't work. You can't seem to edit after posting. IRS2453, like the 2153. Think the data sheet has a better explanation of turning the outputs low than the 2153 does. Look at the capacitor waveform. If just the two LOW outputs are used it looks just like a 2153. Shutdown actually latches it off.
2,000W 120V elements are common in the big box stores and about 7.5 ohms.
This is an interesting thing I thought about trying to add an extra heater to a tank. It connects to the lower drain and gravity feeds hot water to the top.
The company I think is Thermo Dynamics of Canada. This is their solar box. At first glance I thought it was a H bridge using two boards. Since all the wire colors connect, in principle it is the same as mine, just double for their higher current version.
.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 06:03pm 09 Mar 2019
Copy link to clipboard
Print this post
This is the water heater I use now with a 1400W 120V element fed by another 10 gallon tank.
Years ago this was the first water heater control using old PC power supply caps. Bubble wrap provided a protective cover and this stayed on the wall for two seasons.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 01:17pm 13 Mar 2019
Copy link to clipboard
Print this post
This is an example of the regulation. Each output drives two FET in parallel. Notice the skipping of pulses in one channel to lighten load.
Construction is quite simple even without the inverter board and involves very few components. A schematic will follow soon.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 01:11pm 15 Mar 2019
Copy link to clipboard
Print this post
This is the circuit that is fairly easy to build from about a dozen discrete parts. I built one just from an unused circuit board. Anyone better at cad than design is welcome to sell this. Good luck though.
davef Guru Joined: 14/05/2006 Location: New ZealandPosts: 499
Posted: 07:02pm 15 Mar 2019
Copy link to clipboard
Print this post
Is the schematic correct around the opto-isolator?
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 03:07am 16 Mar 2019
Copy link to clipboard
Print this post
The IRS2153/IRS2453 suggestion is brilliant.
It's not only a HBridge controller but also a MOSFET driver, so it really needs a handful of components to build a full transformeless inverter!
HOWEVER, since it's a MOSFET driver, I am sure we want to mount it right next to the MOSFETs themselves to ensure painfree operation?
In your board, it looks like you are connecting this board to the MOSFETs over ribbon cable?
1. How long is the ribbon cable? 2. Are you re using the MOSFETs mounted in your 300W inverter board (bypassing/cutting trace from the controller on board there) and connecting them to this board or you have a separate board with MOSFETs and some passives on them ? I would love to see pictures of the complete system
Dave, Are you confused about the transistor potentially shorting the oscillator cap? If so, that's the neat hack that shuts the inverter down if the PV Panels go into under voltage (UVLO)
davef Guru Joined: 14/05/2006 Location: New ZealandPosts: 499
Posted: 08:11am 16 Mar 2019
Copy link to clipboard
Print this post
No. Is the opto being used as an isolator? If so, I don't see that it is isolating anything. If it is being used as a neat hack then I bow-out.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 05:42pm 16 Mar 2019
Copy link to clipboard
Print this post
Parts is parts and pieces is pieces. This is an alternate design that incorporates the 1V LED side for increased "zener" voltage. It also has nice packing density on a board, fits in a 14 pin socket with the 2153, and eliminates a resistor. They are only 8 cents and potentially allows the driver to be placed long distances from the control. My first water heater was 50 feet from the microprocessor controlling it with no ground loops or interference. Prior design had a transistor and resistor for those in a parts desert.
The 2153 automatically shuts down cleanly at about 9V. Internal zener is about 15.5V and oscillator recovers nicely without glitches. The diode with the 6K resistor is optional and is included for those with a penchant to connect things backwards.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 03:55pm 19 Mar 2019
Copy link to clipboard
Print this post
One disadvantage to the 2153 is the deadtime is very short and can not be changed to create arc suppression. In some cases the thermostat contacts could be switching DC to the water heater element. I switch the main DC using the water heaters contacts at 60V using a special method. I will discuss in another post.
This post discusses a way to switch to multiple heater elements using common relays with just an AC contact rating. The inverter board has two FET outputs that alternate. The two FET output does not reduce the effective on resistance. It does cut the heating of each FET in half. At full power both FET will alternate on and when paralleled will essentially be DC. A DPDT relay can switch the two outputs without a problem of sustained arcing because each output is at 50%. Once passed thru the relay the outputs can then be connected to a single heater element. The following is how two or three tanks could connected together. This would also work for upper and lower heating elements of a tank. Relays are a simple technology easy to understand. This may be preferred by many. If I had multiple heating elements, I would likely go to sets of FET for each heater instead of the relay A microprocessor is preferred with multiple elements.
I have a dishwasher which is connected in this way using a single relay at camp. When the dishwasher demands heat, the wires that went to the heating element go to a 120V AC relay. This provides a set of dry contacts to the water heater control system to operate the water heater/dishwasher relay. Dishwasher has priority and relay switches to the dishwasher heating element.
I use two relays for this because I have to include an off delay. In dry mode the dishwasher element power must be normally reduced because water is not cooling it. It is turned on and off in minute intervals to reduce power when drying to not burn out the element. At lower panel voltages it must stay on all the time for effective drying. A capacitor on the gate of the FET forms a RC time constant to keep the FET and relay on for three minutes. Each heating cycle resets that time and the element is on continuously.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 06:13pm 20 Mar 2019
Copy link to clipboard
Print this post
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 MPPT voltage be less than 30V DC. This can only be done with a heater controller that keeps the capacitor bank at a constant 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 this state.
DO NOT TRY THIS WITH ANY OTHER CIRCUIT THAN THE CONTROLLER SHOWN
Scope image shows a rise in voltage for 60V to 72V when the switch opens up, a differential of only 12V across the switch. Yellow trace is the panel voltage. Green is the capacitor bank. Notice that the green line does not sag at all. Scope voltage needs to be multiplied by 2 due to voltage divider necessary to prevent scope overload. This is well under the 30V DC rating of any switch or relay. The transition takes about 60uS, the majority within 10uS. No wild spikes are shown and the voltage on the capacitor bank never drops below 60V DC.
LadyN Guru Joined: 26/01/2019 Location: United StatesPosts: 408
Posted: 09:08pm 20 Mar 2019
Copy link to clipboard
Print this post
What do you think of an alternate way of handling the arc supression:
Feed +-Vpv at say 120Hz through a full bridge.
100% duty cycle with pretty frequent zero crossings.
No need to reduce the duty cycle. (In reality some dead time will be necessary as MOSFETs are not perfect)
No changes needed at the water heater terminals.
Introducing the full bridge increases the BOM cost by $5 - $8.
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 06:19pm 21 Mar 2019
Copy link to clipboard
Print this post
So, is all this added electronics worth it? People will calculate what they think is the ideal resistance for a heater element. Ideal resistance is far from ideal in direct connect. Ideal power happens less than 2 hours a day. Few days have ideal sun unless you live in cloudless AZ. Panels are current sources and below the power point voltage the current is totally dependent on sun intensity. The voltage across the element is the current times the resistance. Voltage is key because power is a function of the square of the voltage. If the voltage drops to half, the power is only 1/4 of what it was. Here is some math using simplified numbers, two grid tie panels for 600W 60V and 10A. Ideal resistance is 6 ohms at 60V for 600W in the absolute best conditions.
At 75% power 7.5A then 6 ohms X 7.5A is 45V and 337W At 75% power 7.5A for power point 8 ohms X 7.5A is 60V and 450W
At 50% power 5.0A then 6 ohms X 5.0 A is 30V and 150W At 50% power 5.0A for power point 12 ohms X 5.0A is 60V and 300W
At 25% power 2.5A then 6 ohms X 2.5A is 15V and 37W At 25% power 2.5A for power point 12 ohms X 2.5A is 60V and 150W
It is a shock to see the effect a fixed resistance has on a solar panel. Do the math for the entire days total power. Solar calculators are available. The eye is log and the solar panel is linear. A small wisp of a cloud can drop the power quite a bit and you won't notice it. Electronics is the only way to harness all the power. A higher resistance in direct connect may overall work better in your conditions if other electronics is not used. Edited by hotwater 2019-03-23
kanchana Regular Member Joined: 08/05/2018 Location: Sri LankaPosts: 56
Posted: 03:37pm 07 Apr 2019
Copy link to clipboard
Print this post
I need little clarification .
Intention of the system is to divert the excess solar power to heat the water when the MPPT charge controller /PWM charger has stopped charging the batteries?
If so is there ant diversion happening when batteries are charging ?
Regards kanchana
hotwater Senior Member Joined: 29/08/2017 Location: United StatesPosts: 120
Posted: 07:46pm 07 Apr 2019
Copy link to clipboard
Print this post
Yes there is diversion any time the panel voltage exceeds the expected power point voltage. That can be when a PWM charge controller is in the off state. Panel voltage charges the capacitor bank thru a diode used it to isolate the bank from the PWM controller. In a MPPT charge controller, panel voltage rises above expected power point voltage. The diversion controller allows as much current as it can to drop the voltage down to the set voltage. I've seen diversions as low as 2W. This also allows the use of a heater much lower in resistance than would be ideal if it was set up for a direct connect to a heater element. It acts like an electronic transmission.
kanchana Regular Member Joined: 08/05/2018 Location: Sri LankaPosts: 56
Posted: 05:40am 08 Apr 2019
Copy link to clipboard
Print this post
In MPPPT setup ,So when ever the sun is bright solar PV voltage goes up and MPPT take some time to adjust, in that time capacitor bank is charged ?
Instead of heater can we some how divert the power stored in the capacitor bank to the batteries ?Regards kanchana
Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me.
Print this post
