Author Topic: Cheap Efficient PV Water Heater Controller  (Read 536 times)

0 Members and 1 Guest are viewing this topic.

OperaHouse

  • Hero Member
  • *****
  • Posts: 1089
  • Country: us
Cheap Efficient PV Water Heater Controller
« on: February 02, 2018, 12:24:56 PM »
MPPC Water Heating

The performance of MPPC (constant voltage) is very close to the performance of MPPT (tracking)
for all practical purposes. Panel voltages only change with temperature and a seasonal adjustment
is sufficient for temperature when it comes to all the other places you loose power. Big advantages
are simplicity and the ability to prioritize loads. Those loads with only a minor lower set point
get priority. This method can easily divvy out power to several loads seamlessly. I will be using
this construction as a quick dump load for an off grid system using  a grid tie inverter to power
a washing machine. This load can quickly go from 5W to 300W. Having a consistent load prevents the
GTI from taking time to ramp up. It also prevents over voltages from feeding into the pure sine
wave inverter.  This basic circuit can also be used to heat water or be a dump load for wind.
By placing the set point slightly above the expected power point, this system can divert excess
PV power on panels connected to either a MPPT or PWM system without affecting charge performance.

Losses for a direct connect system without power point control are substantial even if the heater
resistance is well matched to the panels at peak production.  The overall daily performance will
be about 40% of a power point control system. As shipped these 3525 boards operate at about 25KHZ.
I slow this down to less than 1KHZ to lower FET transition heating losses. These FET operate in
parallel with only one operating at a time. Mechanical temp switches do not like to switch DC
at high voltages (>30V DC). An alternate method must be used for temperature control, switching
the power electronically. Mechanical manual reset switches must still be used in series with
the heating element as this will be a "one time event". Switches can be used if there is sufficient
dead time when voltage goes to zero, extinguishing the arc. A time of 1.2ms is considered enough.
In that case, Use each FET separately and power separate heater lines.  Each heater will see half
power so more heater elements would be used. This would be appropriate for zone space heating and
each heater could use its own mechanical thermostat. C4 capacitance should be close to 1uF and pot
RT1 set to maximum resistance. This will give an approximate off time of 4ms. Have a sufficiently
large capacitor bank for best efficiency.

I use a microprocessor to divert excess PV energy to heat water. Many have a computer phobia and
this board is an alternate method to heat water efficiently. It can also work as a dump load for
wind. It uses a common inverter control board from China that costs less than $4. With the addition
of a couple FET and some capacitors for storage it becomes a low cost method for heating water at
the PV panels power point as a stand alone system. With the  of an isolating diode It can be added
to any charge controller for diversion heating.Excess power from 5W to hundreds of watts can be
diverted insuring almost 100% of potential PV power is utilized. This article does not discuss any
applicable electrical or building codes as these can vary greatly. Any water heater should have an
excessive temperature cut out to prevent excess pressure due to boiling in addition to the normal
temperature control. A mechanical or electronic thermostat must be included in the system. Voltages
may be dangerous. This is not for beginners in electronics. Heating water above 120F requires the
use of a tempering valve.     

There is no need to purchase expensive heating elements. I use 120V 2,000W heater elements
cost less than $9 and work well on a system as small as a 36V PV array (about 50V power point).
Any time the voltage applied voltage drops by half, the power drops to one quarter. A 240V element
used at 60V will have one sixteenth the power. 9 gallon point of use water heaters provide quick
recovery and are suitable for camp use. Unfortunately they can cost about as much as a full size
tank and that has two heating elements. A few hundred watts diversion may not be much but it should
be remembered that this is heating all day. Adding additional thermal insulation is a must for these
systems. I use two tanks in series providing warmer make up water. A second controller can be used
for that tank with the diversion set point voltage set slightly higher. Less than half a volt increase
will do it. A single control pot may be used with the two controllers in parallel. The secondary
controller will have a slightly higher resistor for the voltage dropping sense resistor. This is not
a beginner project and you should have a good working knowledge of electronics in order to maintain
it.

Thru Hole 3525 Board Mod

As manufactured these boards operate inverse of how you want a dump load controller to operate.
Inverter controllers turn off when the voltage exceeds a set point.  In a dump controller turns
off when the voltage drops below the set point. The following modifications must be performed to
make this happen. There may be variants of this board that uses different component identifiers.
Pin 2 voltage of 3525 must be greater than pin 1 in order to operate. Search 3525 + 358 to find
a board that looks like this. A good part of this boards components are non functional after
modification. There are cheaper SMD boards with a slightly different circuit. Thru hole boards
are much easier to work with.

Pin #1 now becomes the reference voltage of about 1.92V. Remove R8 (2K) resistor and transistor Q1.
The 2K resistor is now connected C to E (two outside pins of Q1) to make up the voltage divider
consisting of R7, 2K(Q1) and R1 feeding pin 1 of the 3525.

A 100K resistor is soldered to the backside of the board from pin #2 of the 3525 to the unused
connector pin #5.  This is the voltage sense line for the capacitor bank. As configured the set
point is about 2V with this 100K to 10K (R3) divider with an 18V input. Figure about 100K for
each additional 18V and use a 50K pot to fine tune the set point.  There are online voltage divider
calculators to help you with the math.  I suggest placing the resistor right at the capacitor
bank so that any short to common is limited in current.

Remove diode D4 at the op amp U4. When installed this causes the op amp to latch on and cause a
shutdown of the 3525 when an over current condition exists. If latching is allowed to occur, the
power must be removed to reset.  This shutdown pin will now be used to turn the 3525 on and off
with an external temperature controller.

Short connector pins 8 & 3 together by soldering a wire on the back of the board. This is easy
to do because the two traces are next to each other on one side of the board. Do this on the
board and not on the connector side. This will then give you two common pins and insure the
board is well grounded. Pin 8 is the current sense which has to be disabled.

Short connector pins 6 & 7 together by soldering. This will be one of the external temperature
controllers inputs. These pins are pulled up to 15V on the board by resistor R13. The 3525 is
enabled when pin 7 is shorted to common.  Use a temperature controller with a normally open relay
contact from pin 7 to common.

The oscillator normally operates at 25KHZ. This must be slowed down to reduce transmitted noise
and reduce FET heating. Bridge a capacitor from .33uF to 1 uf of capacitor C4 on the back of the
board. R4 (10K) may be changed as an option up to 100K to use a smaller value capacitor if desired.
Pot RT1 (5K) can adjust the frequency slightly. The oscillator should be above 150Hz and below 400Hz.

Connector pin #1 is the power pin. Voltages above 12V will work.  The board has a regulator to
limit chip voltage to 15V and there is a 5V reference for critical circuits. I do not recommend
supplying pin #1 with more than 20V. Placing a zener at the input is suggested to prevent the
regulator from seeing more than 20V. The board normally draws around 30ma. Use this value to
calculate a suitable voltage dropping resistor.

Two FET are used for the output and their gate is connected to pins 2 and 4 of the connector
through a 4.7 to 10 ohm resistor. Only one FET operates at a time and the source and drain are
paralleled. The 10K gate to common resistor is optional. It insures the FET turns off if the gate
wire becomes disconnected from control board.

Connect  A 15 ohm 1W resistor in series with a .22uF capacitor from drain to source to absorb
spikes. These exact values are not critical. Most heating element are almost pure resistance and
rather low inductance. If the heating element is high inductance, place a diode in parallel with
the heating element. Choose a FET with appropriate voltage and current. For a 10A load choose a
FET rated of at least 50A.  There is a lot of fine print associated with that 10A rating. To reduce
heating, multiple FET may be placed in parallel to obtain current. This will lower the overall on
resistance. The big issue is thermal transfer from the case to the heat sink. TO-247 cases work
much better than TO-220. Any insulating pad greatly reduces heat transfer. If you can make the heat
sink electrically live it will insure much better transfer. Any insulator pad adds a lot of thermal
resistance. A case is required to make it touch safe.

For stand alone operation I suggest using a wall wart to supply power the board.  Most will work
with over 50V DC applied to the AC terminals.  This insures you will never have an over voltage
failure. I buy these 12V (can be adjusted up to 13V) 1.25A on ebay that have a metal case and cost
less than $1.50 shipped.

MPPC water heater applications require a capacitor bank for charge storage during off periods.
Multiple capacitors are better because they have a lower ESR. As a rule of thumb, I don't like to
see more than 1A ripple current for each for consumer grade capacitors. I've used more than a dozen
220-470uF capacitors in a small system successfully. Preferably The capacitance should be over
10,000uF in total.

Every system will have a different voltage. Use the panels maximum power point as a start.
Approximately 1.92 V must be created on the 10K resistor at pin 2. Ignoring that 2V, dividing
the desired operating voltage by .192 gives the total resistance needed in K ohms. Rx will be
that value minus the 100K already on the board and half the pot value. Higher voltage systems
should use a 100K pot to give more range. It is probably a good idea to shoot for a lower
resistance and have some 10K and 22K resistors to add in series for on site selection.

If this system is used with a charge controller, a suitable high current diode should isolate the
capacitor bank from panels connected to a controller. Solar panels are naturally current limiting. 
A capacitor bank could cause controller charge currents that exceed the controllers FET current
rating by a factor of 10. However, this method will allow harvesting a lot of wasted energy when
charge currents are lower and is well worth doing if you are careful. Do not attempt to use this
system unless you have a good understanding of the principals.


OperaHouse

  • Hero Member
  • *****
  • Posts: 1089
  • Country: us
Re: Cheap Efficient PV Water Heater Controller
« Reply #1 on: February 12, 2018, 01:49:53 PM »
The first board I bought was this surface mount board because it was cheaper.  A little harder to modify and requires external circuitry of a TL431. Still a good demonstrator of how a TL431 can be used as an inverter and precision reference. This very similar circuit can turn any buck converter into a power point regulator.

SMD 3525 Board Mod
As manufactured, these boards operate inverse of how you want a dump load controller to operate.
Inverter controllers turn off when the voltage exceeds a set point.  In a dump controller turns
off when the voltage drops below the set point. The following modifications must be performed to
make this happen. There may be variants of this board that uses different componant identifiers.
Pin 2 voltage of 3525 must be greater than pin 1 in order to operate. Search 3525 + 358 to find
a board that looks like this. A good part of the components on the board are non functional after
modification.

The signal to pin #1 must now be inverted by the external TL431. The reference voltage of the 431
is 2.50V. A 10K pull up resistor is added to the backside of the board between connector pins
1 & 8. This connects to the cathode of the 431 and the anode connects to the common pin 3.
The reference pin attaches to a voltage divider from the panel capacitor bank to common.  These
values must be determined from the expected power point voltage. The pot on the board serves no
function and should not be adjusted.

There are voltage divider calculators on the internet. Put simply, 2.5V must appear from the
reference pin to common. If that resistor is 10K, the upper resistor must be 10K for every 2.5V
seen at the capacitor bank. 60V would be 24 X 10K or 240K. Half the resistance of a 50K pot is 25K.
240K - 25K = 215K, a close resistor would be 220K or 200K. If you are in a hot region, go with the
lower value.  When determining the power point, choose to set the pot at maximum power when the
panels are at no more than half power.

Pin #5 will normally be connected to 12v through a 2.2K resistor. This may be located on the back
of the 3525 board. Connector pin #6 is tied to common.  Again this may be done on the back of the
board. As configured, This is the shutdown control for external temperature controller and
requires pins 5 & 6 to be shorted by a relay in order for the driver to work.

Remove resistor R11 at the op amp U4 between pins 1 & 3. When installed this causes the op amp to
latch on and cause a latching shutdown of the 3525 when an over current condition exists. If latching
is allowed to occur, the power must be removed to reset.  This shutdown pin will now be used to turn
the 3525 on and off with an external temperature controller.

The oscillator normally operates at about 25KHZ. This must be slowed down to reduce transmitted noise
and reduce FET heating. Bridge a capacitor from .33uF to 1 uf of capacitor C4 on the back of the
board. R5 (10K) may be changed as an option up to 100K to use a smaller value capacitor if desired.
As this resistor is SMD changing it is not likely an option. Bridging a capacitor on the back of the
board C3 chip side to adjust the frequency lower. The oscillator should be above 150Hz and below 400Hz.
Scope picture is with .66uf in parallel with C4

Connector pin #1 is the power pin. Voltages above 12V will work.  There is no voltage limiter on
the board. Limit chip voltage to 15V to protect the board and FET gates from excessive voltage.
There is a 5V reference for critical circuits so the raw power does not have to be regulated.
I do not recommend supplying pin #1 with more than 18V. Placing a zener at the input is suggested
to prevent the FET from seeing more than 18V. The board normally draws around 30ma. Use this value
to calculate a suitable voltage dropping resistor.

Two FET are used for the output and their gate is connected to pins 2 and 4 of the connector
through a 4.7 to 10 ohm resistor. Only one FET operates at a time and the source and drain are
paralleled. The 10K gate to common resistor is optional. It insures the FET turns off if the gate
wire becomes disconnected from control board.
« Last Edit: February 12, 2018, 01:54:30 PM by OperaHouse »

SparWeb

  • Global Moderator
  • SuperHero Member
  • *****
  • Posts: 3449
  • Country: ca
    • Wind Turbine Project Field Notes
Re: Cheap Efficient PV Water Heater Controller
« Reply #2 on: February 13, 2018, 12:25:41 AM »
OH,
I'm just chiming in to let you know I enjoy reading all this - just not smart enough to contribute anything.
No one believes the theory except the one who developed it.  Everyone believes the experiment except the one who ran it.

System spec: 135w BP multicrystalline panels, regulated by Xantrex C40, DIY 8ft diameter wind turbine, regulated by Tri-Star TS60, 800AH x 24V AGM Battery, Xantrex SW4024

OperaHouse

  • Hero Member
  • *****
  • Posts: 1089
  • Country: us
Re: Cheap Efficient PV Water Heater Controller
« Reply #3 on: February 13, 2018, 03:21:40 AM »
This one is being sent off to a guy in CA who has like 4KW of panels on a water heater and is looking for more performance. I thought it was always sunny there. He is installing another 10KW and wants to do floor heating with water.  Should be interesting.

ontfarmer

  • Full Member
  • ***
  • Posts: 132
  • Country: ca
Re: Cheap Efficient PV Water Heater Controller
« Reply #4 on: February 13, 2018, 06:21:12 AM »
I enjoy reading and following your posts.  This will be interesting to follow it's progress.